Welcome to another edition of Why Join.

Every company reports the same three numbers, and only one of them is hard to bend. Revenue is a story about demand. Earnings are a story about accounting. Cash is what’s left after both stories are told.

That’s why the cash flow statement is the most honest document a company produces. It shows you what management actually believes, not what they said on the call. A company that talks about a decade-long bet and spends nothing is bluffing. A company that talks about discipline and burns everything is telling you the truth by accident.

The uncomfortable part is that conviction and panic look identical on paper. Both show up as a number going down. You can’t tell them apart in the quarter. You can only tell them apart in five years, which is exactly when nobody is checking anymore.

So the market prices the story instead. It has to. Nobody has the other data yet.

Spending is a claim about the future that costs something to make. Everything else is a press release.

On to who raised, who's hiring, and what we're reading.

In partnership with: THNKS

You didn’t automate your relationships. So why did you automate every message in them?

Every email, sequence, and follow up now asks your customer for something. The one touch that asks for nothing is the one they remember. Send a coffee, lunch, or ride from the tools you already use. No form, no reply, no meeting. Small gestures. Big results.

Send a Thnks

Partner with us →

Who Recently Raised 💰

🇮🇱 Enigma

💵 Raised: $71M seed, led by Index Ventures and Ribbit Capital. Sarah Guo’s Conviction also in, plus leaders from OpenAI, Anthropic, DeepMind, xAI, Cognition, and Wiz.

One-liner: AI models and interfaces to make controlling any robot feel effortless, starting by putting 100 robots online for anyone to use.

Why it’s a fave ツ: Most robotics companies are racing on capability. Can the robot fold the shirt, can it do the dexterous thing it wasn’t trained on. Enigma is chasing a different problem: even with the most capable models today, telling a robot what you want is miserable. If you had to spend 15 minutes explaining to a robot where every dish goes, you’d give up and do it yourself. Everyone’s at that point right now. Capability isn’t the only wall. Interface is.

Think of a car volume knob. You turn it, instantly hear whether it’s too loud, and adjust. Now imagine having to set the volume as an exact percentage without hearing the result. That’s roughly what controlling a robot feels like today. Enigma wants to find the robotics equivalent of that knob, the interaction so intuitive nobody has to think about it.

To find it, they’ve been running an unusual experiment. Anyone in the world can visit robots.online and control one of more than 100 real AI robots in real time. The robots are housed in hangars in Israel and California, where they can draw with a paintbrush, sword-fight each other, and run simple chemistry experiments with flasks. Every interaction teaches Enigma how people instinctively try to communicate with a machine (text, voice, showing a video, tap and drag) and that data feeds back into both the interface and the underlying model. They built the robotic arms and the models from scratch, and the models are hardware-agnostic, meaning they can adapt to a new machine with a fraction of the training data usually required.

🇺🇸 Fish Audio

💵 Raised: $52M seed, led by Coreline Ventures and Capital Today. 359 Capital, HF0, and others also in.

One-liner: Voice AI with word-level emotion control.

Why it’s a fave ツ: Most synthetic voices are flat. Fine for a menu prompt, useless when you need a game character to sound scared or an avatar to sound genuinely warm. The technical gap Fish Audio went after is control. Their S2 model lets you drop instructions inline at the word level using free-form tags like [whisper] or [professional broadcast tone]. Not a fixed menu, open-ended description. They count 15,000+ of these controls. The model is 4 billion parameters, trained on 10 million+ hours of audio across roughly 80 languages, and can clone a voice from a 5-second clip in about 15 seconds. On independent-style benchmarks like EmergentTTS-Eval and Seed-TTS Eval, S2 scores at or above closed models from Google and OpenAI, which is a real achievement for a small team.

The consent problem is real. Fish Audio built part of its voice library by letting users upload voices and earn money when those voices were used. But a UK voice actor found her voice had been cloned and downloaded more than 900 times on Fish Audio without permission. The company now says verified takedowns happen within three minutes, but only after the artist discovers the upload and files a claim. That is the core tension: Fish Audio’s open model drove growth, but it only works long-term if consent, attribution, and revenue-sharing are built in from the start.

🇺🇸 Andrenam

💵 Raised: $18M Series A, led by Upfront Ventures. Valor Equity Partners, Also Capital, and First Round Capital also in. $30M total. The $10M seed last year closed in 36 hours.

One-liner: Distributed network of underwater acoustic sensors, plus AI software to make sense of what they hear.

Why it’s a fave ツ: The ocean is enormous and almost impossible to watch. We track planes across the sky in real time, but underwater is a near-total blind spot, especially the sneaky stuff happening below the surface. The reason is physics. Radar and cameras don’t work through water. Sonar is the only thing that reaches long range, and building sonar coverage across an ocean is a genuinely hard problem nobody has solved at scale.

Andrenam builds two things. Pearl is a networked buoy with a suspended passive sonar array that listens for the bumps, whirs, and rumbles of vessels moving underwater. They’ve built 35 of them, logging over 4,500 in-water hours off the California and Washington coasts. All that sensor data feeds in real time to Obsidian, their software, which detects, tracks, classifies, and estimates the probability of what type of vessel it’s hearing, then plugs into whatever command systems the end user already runs.

Why this matters now: China is building submarines at roughly six times the annual rate of the US, and criminal organizations are increasingly moving product underwater. There’s a real intelligence gap in wide-area underwater sensing, and there are a lot of unmanned systems in the water now that need something to actually listen for them.

Hiring: Head of Finance, Senior Software Engineer, Federal Account Executive, Supply Chain Manager - Torrance

🇺🇸 Paper

💵 Raised: $34M Series A, co-led by Accel and ICONIQ. Designer Fund, Michael Grinich (WorkOS), Anton Osika (Lovable), and engineers from Anthropic and OpenAI also in.

One-liner: Design tool where everything on the canvas is real HTML and CSS, built for a world where agents write the code.

Why it’s a fave ツ: Every design tool ever made has the same gap. You design in one format (Figma’s proprietary canvas), then someone translates it into code. That handoff is where things break, timelines slip, and the final product drifts from the design. It was an annoying but tolerable problem when humans did all the coding. Now that coding agents like Claude Code and Codex write more and more of the actual software, the translation step becomes a real bottleneck, because the agent works in code and the designer works in a format the agent can’t natively touch.

Paper removes the translation step entirely. Every element you place on the canvas is actual HTML and CSS. What you design is what ships. No export, no conversion, no handoff gap. And their MCP server exposes 24 bidirectional tools, which is the part that matters for agentic workflows. Figma’s MCP server gives an agent a window into your design (read-only, it can look but not touch). Paper’s gives the agent a steering wheel. An agent can read the design and modify the canvas directly, sync tokens, pull in live API data, convert a design to React and commit it to GitHub. A third-party benchmark found Paper used roughly 6K tokens and 3 minutes per AI design task versus Figma’s 20K tokens and 10 minutes.

The timing is hard to ignore. Figma’s stock is down 40% in 2026, partly on fears that agentic design tools could weaken the proprietary-canvas model it spent a decade building. Paper is a working example of that threat. But Figma still has roughly 90% market share, deep enterprise adoption, strong network effects, and high switching costs. Paper barely appears in industry usage data, and design tools depend heavily on collaboration and ecosystem, where Figma has a decade-long lead. Paper is growing, but from a small base. The real question is whether a code-native canvas can shift the market faster than Figma can adapt.

Open Tabs (stuff we’re reading) 📖

• OpenAI giving 100K academics free ChatGPT access through 2027: Researchers in biology, chemistry, CS, engineering, math, physics get access to OpenAI’s most advanced models including GPT-5.6 Sol Pro. Roughly equivalent to the $200/mo Pro tier. Each researcher can invite four collaborators. Data won’t be used for training. First 10K getting access this summer. Part of $250M+ commitment to external scientific research through 2027. The catch: program covers math and sciences but not AI research itself. And researchers still don’t get model weights or training data, which is what AI researchers actually want for independent safety evaluations. Tools yes, transparency no.

• Fed holds rates at 3.5-3.75%: 9-3 vote. Fifth straight hold. Chairman Warsh: "There is no soft inflation target. Not on this committee's watch." Fed's preferred inflation gauge at 3.4% through May. The fractured vote suggests holding is getting harder. Uncertainty from tariffs, the Iran conflict, and AI-driven demand all adding to inflation fears.

• Meta misses, free cash flow nearly gone: Q2 revenue $60.8B (beat slightly), but EPS missed: $6.18 vs $7.22 expected. Q3 revenue guidance $61-64B, below the $63.15B Street estimate. The real number: free cash flow collapsed to $784M from $8.55B a year ago. AI spending is the reason. Capex guidance narrowed to $130-145B for the year. Total costs up 55% YoY, includes $2.4B in legal charges and $1.18B in severance from the May layoffs. Net income fell to $15.85B from $18.34B. Reality Labs lost $4.6B on $431M revenue.

• Microsoft beats, Azure crosses $100B: Azure grew 43% in fiscal Q4, fastest in four years. Annual Azure revenue topped $100B for the first time. Total quarterly revenue $90B (up 18%), net income $35.8B (up 31%). Full year revenue $331.8B. Intelligent Cloud segment: $39.3B (up 32%). Copilot hit 30M paid users, up from 20M last quarter. $3.2B gain from Anthropic investment baked in. More Personal Computing segment declined 4%, Xbox content and services revenue down 10%, Windows OEM down 7%. Still free-cash-flow positive.

• Microsoft launches first cybersecurity AI model: MAI-Cyber-1-Flash, built to find vulnerabilities in complex codebases. Scores 96% on CyberGym, beating Anthropic’s Mythos 5, GPT-5.5 Cyber, and GPT-5.6 Sol. Combined with GPT-5.4 inside the MDASH harness, handles 90% of tasks cheaply and routes the hard 10% to the bigger model. 50% cost savings. Also launching Perception, an agentic security platform with AI red/blue/green teams that can discover, prioritize, and fix vulnerabilities in minutes instead of hours. Directly competing with Anthropic’s Mythos/Glasswing and OpenAI’s Daybreak program.

  

• Trump admin bans new Chinese humanoid robots and power inverters: FCC barred imports of new Chinese humanoid robots, quadruped robots, and connected power inverters. China dominates humanoids with ~85% global market share. Of the ~15,000 humanoid robots shipped globally in 2025, Chinese companies Unitree and AGIBOT each shipped 5,000+. Tesla and Figure AI each shipped a few hundred or less. Unitree (~20% market share) was recently added to the Pentagon’s Chinese military-backed companies list despite having a Nvidia partnership using Blackwell chips. Inverter ban hits Sungrow and Huawei. The fear: robots could spy and send data to Beijing, inverters could be compromised like the Volt Typhoon router hack.

• Moonshot open-sources Kimi K3, but with a catch: The Chinese AI startup released its model’s technical details publicly, but large-scale users will need commercial licenses. First Chinese AI company to make that requirement explicit. K3 shook Silicon Valley when it dropped earlier this month, matching top OpenAI and Anthropic models on some tasks. The model is too large to run on a basic computer, so most users will either subscribe through Moonshot or go through a licensed third party. Two days after announcing K3, Moonshot had to stop accepting new users because it couldn’t get enough chips to serve them.

  

• Apple launches device leasing program with Klarna: Called “Apple Upgrade.” Monthly installments starting at $17.99 for iPhone, $11.99 for Apple Watch, $24.99 for Mac, $11.99 for iPad. 1-2 year leases for iPhone/Watch, 2-3 year for Mac/iPad. At the end you can upgrade, buy outright, or return. Replaces Apple’s existing in-house financing programs (iPhone Upgrade Program, iPhone Payments) in the US. The context: Apple recently hiked Mac and iPad prices due to “RAMageddon,” the industry-wide memory chip shortage driving up hardware costs. This is Apple’s way of softening the sticker shock. 3% Daily Cash back if you pay with Apple Card.

• PayPal leaves door open to Stripe deal, but not at $60.50: Q2 earnings beat expectations: adjusted EPS $1.38 (vs $1.28 expected), revenue $8.68B (up 5% YoY, above $8.47B estimates), adjusted FCF $1.8B. CEO Lores didn’t shut down the Stripe/Advent bid but didn’t accept either, saying PayPal would consider any path that creates “superior value.” Cantor values PayPal closer to $70/share. Stock trading at ~$58. Turnaround continues: reorg into three segments, targeting $1.5B in cost savings over 2-3 years, migrating from data centers to cloud. Lores still framing PayPal as an AI-driven tech company, not a legacy payments business.

• DoorDash building its own drone delivery business: Called DoorDash Air. Just got FAA Part 135 air carrier certification. Building custom aircraft in-house through DoorDash Labs, the same team behind Dot, its autonomous sidewalk delivery bot (already operating in Phoenix suburbs and Fremont, CA). Software called Autonomous Delivery Platform decides in real-time whether to dispatch a human driver, drone, or sidewalk bot. Keeping existing partnerships with Wing and Flytrex. No timeline on when drones actually start delivering. Still needs FAA Beyond Visual Line of Sight approval to fly autonomously, which Amazon, Wing, and Zipline already have.

• X Money app launches in the US: Paid X subscribers get a Visa debit card (instant Apple Pay, physical card coming), fee-free P2P transfers, no foreign transaction fees, free ATM withdrawals worldwide. Premium+ users ($40/mo) get 6% APY. Premium users ($8/mo) can get 6% too if they link direct deposit, plus early access to paychecks and up to 3% cash back. This is Musk’s 25-year-old dream coming full circle. He founded X.com as a financial services startup in 1999, it merged into PayPal, and now he’s essentially rebuilding it inside the platform he renamed X.

Who’s Hiring 💼

• Techstars: Operations Associate - Los Angeles

• Unusual Ventures: Operating Partner - San Francisco

• Prime Intellect: Head of Marketing, Head of Growth, Full-stack Engineer - San Francisco; AI Research Resident - Open Source AGI - Remote

• Venice AI: Software Engineer - Remote

• Super.com: Senior Data Analyst, Full-stack Engineer, Payments Engineering Intern (12-month), Software Engineering Intern - Product (Full-stack), Legal Intern, Travel Supply Analyst Intern (8-month) - Remote

• Osmo: Head of IT - New Jersey

• Rivet: Account Executive, Tax Manager, SDR - Remote

• Kaizen: Federal Growth Lead, Federal Proposal Writer & Capture Strategist, Head of Brand & Storytelling, Digital Designer - New York, Remote

• Firefly: Visual Designer - Brand, Enterprise BDR, Founding Growth Marketer, Software Engineer - New York

• Paces: Product Manager, Demand Gen Lead, Product Marketer, Software Engineer (Junior) - New York

• Koah Labs: Content Marketer - New York; Founding Head of Ad Sales - San Francisco/New York; Product Designer - San Francisco/London/New York

• Fresco: Sales Development Representative, Founding Software Engineer - San Francisco

• Willow: Founding Growth & Content Lead, ML Engineer - San Francisco

• Alta: GTM, Sales, AI Engineer - Tel Aviv/US, Remote

• Monumental: BizOps/Special Projects, General Manager, Technical Recruiter, GTM, Head of People & Performance - Amsterdam/US, Remote

• Spade: Finance Lead, Account Executive - New York

• Monk: Full-stack, Strategy & Operations Lead, BDR - New York

• Prism Data: Accounting and Finance Manager, Partnerships Manager, Sales Associate, Software Engineer - NYC/San Diego, Remote

• Stuut: People Operations Manager, Director of Product Marketing, Applied AI, Full-stack, Business Development Representative (BDR) - New York

• Metal: Account Executive, Forward Deployed Investor - New York

• Thira: Founding AI Engineer - Seattle

• Sable: Growth, Founding Account Executive, Founding Designer, Research Engineer - San Francisco

• Closure: Founding Deployment Strategist, Founding AI Engineer, Business Development/Account Executive - New York/San Francisco

• HealthKey: Account Executive (Founding GTM) - Austin

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

Share Why Join

See you Saturday, Ryan

Sponsorships: We are now accepting sponsors for Q3/Q4 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome to another edition of Why Join.

When two companies are racing to own the same category, watch what they spend money on. Not the product. The narrative.

The product differences between competitors are almost always smaller than either side wants to admit. Same underlying infrastructure. Same basic capabilities. Same vendor powering the core engine. You could swap the logos and most customers wouldn’t flinch.

So the fight moves upstream. Whoever defines the category wins the category. This has been true for fifty years. The best companies don’t just launch products. They launch words. They name the problem. They coin the metric. They create the axis on which their product sits at the top and everyone else is playing catch-up.

By the time the competitor responds, they’re already playing defense. Not because the product is worse. Because the story belongs to someone else. And in technology markets, the story is the product.

The most underrated skill in tech is not building. It’s framing.

On to who raised.

Why Join Exclusive: You Could Be an Investor in High-Potential Startups Like These

Why Join and Alumni Ventures are teaming up to give readers early access to high-growth startup opportunities, including some of today’s most exciting AI, Deep Tech, Quantum Computing, Cybersecurity, and Space companies co-invested alongside top VC firms like Andreessen Horowitz (a16z), Bessemer, & Y Combinator.

You get:

• Curated deal flow of high-potential startups

• AV is already investing alongside elite lead venture firms in these deals

• No cost to see deals

• No obligation to invest

Don’t miss your chance before access closes.

Join Alumni Ventures Syndicate Today →

Who Just Raised 💰

🇺🇸 Manifest OS

💵 Raised: $60M Series A at $750M valuation, led by Menlo Ventures. Kleiner Perkins, First Round Capital, and Quiet Capital also in. Largest Series A in legal tech history.

One-liner: AI-native law firm model that kills the billable hour and replaces it with fixed-fee pricing.

Why it’s a fave ツ: 80% of American businesses and consumers can’t afford to hire a lawyer when they need one. The reason is the billable hour. It ties firm economics to time spent, not outcomes delivered. Billing rates hit $3,000/hour last year. Rates went up 7.4% in 2025 alone. And 78% of lawyers work at small or solo practices where they spend up to two-thirds of their time on admin, not actual legal work. The $6B invested in legal tech last year mostly went to AI tools that make traditional firms more profitable... without actually making legal services cheaper for clients.

Manifest OS took a completely different approach. Instead of selling software to existing law firms (who are incentivized to bill more hours, not fewer), they built AI-native law firms from the ground up. They provide everything: the brand (all firms operate under “Manifest Law”), the software (AI platform for client comms, legal research, document drafting, billing), and a centralized back office (paralegals, admins, intake, collections, quality assurance). Lawyers just do legal work. Everything else is handled.

The key structural difference: traditional law firms distribute year-end profits as partner bonuses. Manifest reinvests them into R&D. So the cost of delivering legal services keeps dropping over time instead of going up.

Started in business immigration under Arizona’s ABS program (which lets non-lawyers have ownership stakes in law firms). 18 months in: 3,000+ client engagements, 100+ immigration attorneys, visa approval rate 15% above the national average, 3x faster client response times than traditional firms. Out of 5,000+ attorney applicants, fewer than 1% accepted. Fixed-fee on everything.1

Hiring: Sales Development Representative, Strategy & Operations, GTM, Junior Design Engineer, Product Manager, Full Stack Developer - New York, NY/ Events Manager - Remote

🇺🇸 Chord

💵 Raised: $7M, led by Equal Ventures. M13 returning, Chingona Ventures and CEAS Investments new.

One-liner: AI platform that gives commerce teams a single brain for their entire operation.

Why it’s a fave ツ: The problem they're solving is deceptively simple. E-commerce brands run on a dozen disconnected tools. Analytics here, inventory there, marketing somewhere else. You can't bolt AI onto that mess and expect it to make good decisions because it has no context. Chord built what they call a "context graph" for commerce. Think of it as a living operational memory that captures how a business actually runs... metrics, rules, historical decisions, constraints, trade-offs. It compounds over time. So when you ask the AI a question or tell it to do something, it's working with real-world context, not just raw data from a dashboard. Teams can ask questions, generate insights, and take action in the same place.

Hiring: Account Executive - New York, NY; Remote (US)

🇺🇸 Windmill

💵 Raised: $12M seed, led by Inspired Capital. Primary Venture Partners, Founder Collective, and Oceans Ventures also in.

One-liner: AI that builds a living map of your workforce so companies can actually manage their people.

Why it’s a fave ツ: Every company says people are their most important asset. Then you ask how they actually track who’s performing, who’s growing, who’s underutilized, who elevates the people around them... and the answer is: it lives in the heads of a handful of managers. Maybe some spreadsheets. Maybe an annual survey nobody reads. You literally have a system of record for every asset in the company except the people.

This matters more now, not less. AI is flattening orgs, merging roles, giving individuals the leverage that used to require whole teams. The value of each person goes up. But companies have no infrastructure to understand what’s actually happening with their workforce in real time. You can’t ask an AI agent to identify your best teams or surface bottlenecks or help staff a critical project because the context doesn’t exist anywhere.

Windmill started with the most broken process in the workplace: performance reviews. Their platform plugs into Slack, GitHub, Google Workspace, and 30+ other tools where work actually happens, so reviews never start from a blank page. Completed 90% faster with 93% employee satisfaction. But the reviews are really just the first proof point. Underneath is what they call the context graph... a continuously updated understanding of who people are, what they’ve done (traced back to specific Slack threads, pull requests, documentation), what’s expected of them, and what others think of them. Everything cited, nothing hallucinated from memory.2

Hiring: Design Engineer, ML Engineer, Software Engineer - New York, NY

Open Tabs (stuff we’re reading) 📖

• GameStop makes unsolicited $55.5B bid for eBay: $125/share, half cash half stock, 20% premium. The math is wild: GameStop’s market cap is ~$12B, eBay’s is $46B. So how does Cohen pay for it? ~$9.4B cash on hand, $20B commitment letter from TD Bank, plans to issue stock, and potentially Middle Eastern sovereign wealth funds. He’s built a ~5% eBay stake (mostly derivatives) starting Feb 4. The pitch: cut $2B in annual costs, nearly double EPS from $4.26 to $7.79 in year one, use GameStop’s 1,600 stores as physical infra for authentication, fulfillment, live commerce. Cohen would run the combined company with no salary, comp tied entirely to performance. ($35B+ payout if it hits $100B valuation.) eBay board says it’s reviewing. No conversations with management yet. Proxy fight on the table if they don’t engage.

• Meta acquires robotics AI startup for humanoid push: Bought Assured Robot Intelligence, a startup that builds AI models helping robots understand and adapt to human behavior in real-time. Team joining Meta Superintelligence Labs to work on humanoid technology. Co-founders came from Nvidia and Fauna Robotics (Amazon bought Fauna in March for its own humanoid effort). Meta’s bigger play here: they want to be the Android of humanoid robots. Build the foundation layer, sensors, software, AI, and let the rest of the industry build on top. Tesla, Google, Amazon all chasing the same space.

• Ask.com shuts down after 30 years: The search engine formerly known as Ask Jeeves is done. Launched in 1996, it let users ask questions in natural language, arguably making it a precursor to today's AI chatbots. Never came close to competing with Google though. IAC bought it in 2005, dropped the Jeeves branding, scaled back search by 2010. Officially closed May 1, 2026. The farewell message: “Jeeves’ spirit endures.”

• Musk’s companies more than doubled spending on each other: Tesla’s amended 10-K shows related-party transactions hit ~$600M in 2025, up from ~$240M in 2024. xAI paid Tesla $430M (mostly for Megapacks), up from $198M. SpaceX spent $143M on Tesla vehicles for the first time. Tesla sold about 1 in 5 Cybertrucks registered in the US to Musk’s other companies. Tesla also paid $4.8M to Musk’s personal security company. All before Tesla’s separate $2B investment in xAI, which post-merger now gives Tesla a stake in SpaceX.

  

• xAI launches Grok 4.3 at rock-bottom pricing: Big performance jump over Grok 4.2 on third-party benchmarks but still below OpenAI and Anthropic's best. The real play is price: $1.25/$2.50 per 1M input/output tokens, roughly 40% cheaper than its predecessor and way below GPT-5.5 ($5/$30) and Claude Opus 4.7 ($5/$25). 1M token context window, reasoning baked in permanently. Tops legal and finance benchmarks (CaseLaw, CorpFin) but users report it "falls asleep" on agentic tasks and struggles with hard math. Also launched a voice cloning suite: 120 seconds of audio creates an eerily accurate clone of your voice, $3/hour for voice agent interactions. US only (excluded Illinois over biometric privacy laws). All 11 original xAI co-founders are now gone but the product machine is still shipping.

  

• Anthropic forming $1.5B joint venture with Blackstone, Goldman, others: JV will sell AI tools to private equity-backed companies. Anthropic, Blackstone, and Hellman & Friedman each putting in ~$300M. Goldman adding ~$150M. General Atlantic and others also in. The entity acts as a consulting arm for Anthropic, helping PE portfolio companies integrate AI across operations. OpenAI is in talks for a rival JV with PE firms. Both see PE-backed companies as prime targets since they’re already focused on efficiency and cost-cutting.

• Cerebras IPO targeting $4B raise at $40B valuation: Up from the earlier $2B target. Banks already have $10B+ in indications of interest. Valuation trajectory: $8.1B in Sept, $23B in Feb, now $40B. Revenue hit $510M in 2025 (up 76% YoY) and the company swung to profit after a $485M loss in 2024. The OpenAI relationship is the anchor: $20B+ compute deal through 2028, $1B loan from OpenAI at 6%, plus warrants for 33.4M shares. Ticker: CBRS on Nasdaq.

• Legal AI startup Legora hits $5.6B valuation: Swedish-born Y Combinator alum just crossed $100M ARR, 18 months after launching. Nvidia’s NVentures made its first legal AI investment here. $50M Series D extension on top of a $550M Series D a month ago. 1,000+ law firms and in-house teams across 50 markets. Going head-to-head with Harvey ($11B valuation, backed by Sequoia, a16z). The marketing war is fun: Harvey signed the actor from “Suits,” Legora countered with Jude Law under the slogan “Law just got more attractive.” Both built on top of LLMs from the big AI labs, which could become their competitors. (When Anthropic launched a legal plug-in for Claude, several legal software stocks dropped.)

Who’s Hiring 💼

• Fyxer: Senior Creative Strategist, Member of Product Team - London/ Senior Product Support Specialist - Austin, TX

• Granola: Community Manager - San Francisco/ Product Engineer (Backend), Product Engineer (Full Stack), AI Engineer, Web & Brand Designer, Content Lead, SDR, People Operations Lead - London

• Wispr Flow: Technical Support Engineer - Bangalore; San Francisco; New York/ SDR (PLG / Install-Base Expansion), Product Lead, B2B Marketing Lead, Paid Search Performance Marketing Manager - San Francisco; New York/ QA Engineer (Windows / iOS) - Bangalore

• Littlebird: Product Analyst, Applied AI Engineer - Agent Intelligence & Retrieval, Brand (Visual) Designer, Growth Analyst - X, Senior Content & Community Marketing Manager, Social Media & GTM Associate - Remote

• Coefficient: Support Specialist - Philippines/ QA Automation Engineer - Bangalore/ Early Product Engineer - San Mateo, CA

• Redpine: Business Development - Internship, Design Lead, Partnerships Manager, GTM - Stockholm, Sweden/ GTM - US

• Marloo: Founding Engineer, Design Lead, Growth Lead, GTM Associate, SDR - London

• Trata: Founding Engineer - New York, NY

• Magical: Junior Software Engineer, AI (New Grad OK), San Francisco, CA; Toronto, Canada/ Senior Software Engineer, Agent Platform - Toronto; San Francisco/ SDR, AI Deployment Strategist - Remote (Canada/US)

• Orb: Software Engineer, Core Product (New Grad OK) - San Francisco/ Solutions Architect, Software Engineer, Infrastructure - Remote (US)

• Durable: AI Marketer, AI Engineer - Vancouver; Remote - Canada

• Ahrefs: Backend Engineering Intern - OCaml, Social Media Manager, Product Designer - Palo Alto, CA; Singapore; Acton, MA/ Japanese Speaking Customer Success & Support Specialist - Remote (Japan)

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

Share Why Join

See you Wednesday, Ryan

Sponsorships: We are now accepting sponsors for Q2/Q3 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome to another edition of Why Join.

There’s a famous story about the Panama Canal. The French started it in 1881. They had the most celebrated builder on the planet, Ferdinand de Lesseps, the man who built the Suez Canal. Same playbook, bigger ambition. And for eight years they dug. Tens of thousands of workers, $287 million in 1880s dollars, and they could absolutely move earth. That was never the question. The machines worked. The dynamite worked. They could excavate with the best of them.

They just couldn’t control where the water went afterward.

Landslides refilled what they’d dug. The Chagres River flooded the cuts every rainy season. Mosquitoes killed workers by the thousands because standing water was everywhere and nobody understood drainage or disease control. The project went bankrupt in 1889. De Lesseps was convicted of fraud. He died a year later, broken.

When the Americans took over in 1904, they didn’t just show up with better shovels. They showed up with a different theory of the problem. The lock system, the Gatun Dam that tamed the Chagres instead of fighting it, a massive mosquito eradication campaign that cut the death rate by over 90%. The canal opened in 1914. The breakthrough was never the ability to move earth. It was the ability to decide what happened after the earth moved.

Every major technology goes through this exact sequence. The capability arrives first and gets all the attention. Then someone has to build the infrastructure that makes the capability actually obey you. The first part is a spectacle. The second part is a company.

On to who raised.

In partnership with

Stop drowning in research papers.

Anara is the AI workspace built for researchers, scientists, and students who need answers they can actually trust. Upload any paper, textbook, or study and get instant, cited responses pulled directly from your sources, not hallucinated from memory. Search PubMed, arXiv, and JSTOR from one place. Auto-generate citations as you write. Over 2 million researchers at Stanford, MIT, Mayo Clinic, and GSK already use it daily. Free to start, no credit card required. Backed by Y Combinator.

Try Anara free →

If you’re interested in sponsoring this newsletter, please reach out here.

🇺🇸 ComfyUI

⛀⛁ Raised: $30M at a $500M valuation, led by Craft Ventures. Pace Capital, Chemistry, and TruArrow also in.

One-liner: Node-based tool that gives creators precise control over AI-generated images, video, and audio.

Why it’s a fave ツ: If you’ve used Midjourney or DALL-E, you know the problem. You prompt something, it gets 60-80% there. Then you try to fix the remaining 20%, and the model rewrites everything, including the parts that were already good. Like a slot machine. You keep pulling the lever hoping to land on what you want, but every pull reshuffles the whole thing.

ComfyUI solves this with a node-based interface (think visual building blocks you connect together) where creators can control every individual step of the generation process. Want to change the lighting without touching the composition? Swap a texture without regenerating the whole image? You can do that. You can’t do that in a prompt box.

Started as an open-source project in 2023, right after diffusion models (the tech behind AI image generation) dropped. Back then, the models could barely draw hands correctly. The tool got so popular with creative professionals that it turned into a real company. Now has 4 million+ users. That open ecosystem is the structural advantage no closed tool can replicate.

Hiring: Partnership & Events Marketing Manager, Lifecycle Growth Marketer, BizOps Strategist, Product Manager, ComfyUI, Software Engineer, ComfyUI Desktop, Graphic Designer, Design Engineer, Creative Artist, Senior Product Designer - San Francisco; Software Engineer, Frontend - Remote/San Francisco

🇵🇱 GRAI

⛀⛁ Raised: $9M seed, co-led by Khosla Ventures and Inovo VC. a16z Scout Fund, Tensor Ventures, Tiny.VC, Flyer One Ventures also in.

One-liner: AI music platform that lets people play with tracks instead of generate them from scratch.

Why it’s a fave ツ: Every AI music startup right now (Suno, Udio) is focused on generation. Type a prompt, get a song. But most people don't actually want to create music from scratch. They want to mess around with songs they already like. Remix a track, change the style, send it to a friend. Music is one of the last major consumer categories that hasn't gone social and creator-first. Discovery is broken, listening is passive, and there’s basically zero social context around any of it.

GRAI is building apps where people can play with real tracks. Remixing, restyling, sharing. Not generating AI slop and dumping it on Spotify. And here’s where they diverge from basically every other AI music company: they’re talking to labels first. Not building and asking forgiveness later. The whole system is designed so artists opt in and opt out. If remixing takes off it creates a new royalty stream for artists and labels rather than undercutting them. Whether labels actually go for that... we’ll see. But at least they’re asking.

They built their own infra for this. Taste and participation graph, real-time audio systems that preserve the identity of original tracks while allowing transformation, a derivatives pipeline for modified versions. Two apps live now, one iOS (Music with friends), one Android. Going after Gen Z and Gen Alpha who discover music through friends and fandoms and short-form content, not playlist algorithms (over 80% of US recorded music revenue now comes from streaming, and on those platforms the algorithm decides what you hear).

Hiring: Android Developer, Python Engineer (MLOps/Data Infrastructure), QA Engineer, Research Engineer, Head of Finance & Operations - Warsaw

🇺🇸 Era

⛀⛁ Raised: $11M total. $9M seed led by Abstract Ventures and BoxGroup, with Collaborative Fund and Mozilla Ventures. $2M pre-seed from Topology Ventures and Betaworks.

One-liner: Software platform that lets anyone build AI-powered gadgets without building the AI stack themselves.

Why it’s a fave ツ: AI hardware has been... rough. Humane got sold to HP. Rabbit went quiet. Nobody’s really figured it out yet. But the founding team at Era thinks the problem isn’t that people don’t want AI gadgets. It’s that every hardware maker has to build their own intelligence layer from scratch, which is expensive and hard and most of them aren’t AI companies.

Era doesn't make devices. They're the software underneath. Their platform gives hardware makers access to 130+ LLMs from 14+ providers, handles multimodal inputs (voice, vision, text), manages real-world constraints like connectivity, and routes dynamically across models. So whether you’re building AI glasses, a smart ring, a home speaker, or a souvenir that tells you jokes about France (someone actually made that), you plug into Era instead of building the AI orchestration yourself.1

Hiring: Researcher in Residence, Supply Chain & Manufacturing, Storytelling, Product Engineer, GTM, Partnerships & Business Development, Interaction Designer/Product Designer, Software Engineer, iOS & Android, Hardware Engineer, Embedded Systems & Electronics - NYC, SF, or Remote

Open Tabs (stuff we’re reading) 📖

• China blocks Meta’s $2B acquisition of AI startup Manus: Beijing’s National Development and Reform Commission (NDRC) ordered the deal canceled. The acquisition of the agentic AI startup was announced in December and had been largely completed. It triggered a probe into illegal foreign investment and tech exports. Initially seen as a template for Chinese AI startups with global ambitions, critics called it a loss of valuable AI tech to a geopolitical rival. China has since tightened scrutiny of key AI firms.

• Google investing up to $40B in Anthropic: $10B now at a $350B valuation, another $30B tied to performance targets. Google Cloud providing 5 GW of fresh compute capacity over five years, on top of the 3.5 GW Broadcom/Google TPU deal announced earlier this month. So in one week Anthropic locked in: $25B from Amazon, $40B from Google, plus the CoreWeave data center deal. Google is simultaneously a competitor and key infrastructure supplier (Anthropic relies heavily on Google’s TPUs).

• XChat launches on iOS: X’s standalone encrypted messaging app launched on iOS, aiming to compete with WhatsApp and Signal. Claims E2E encryption, no ads, no tracking. But: encryption keys live on X’s servers (protected by a four-digit PIN), images retain GPS coordinates and camera metadata, and the app shares account/usage/device data with third parties including potentially advertisers. X itself has acknowledged that its architecture “could allow a malicious insider or X itself” to access conversations. Security researchers say it’s fine for casual chat, but for anything sensitive, use Signal. Android release date TBD.

• X shutting down Communities: Used by less than 0.4% of users but generated 80% of spam reports, financial scams, and malware on the platform. Head of product Nikita Bier called it a “Temu version of subreddits.” Most successful Communities were just user-acquisition channels for Kick or paid clipper networks. Shutting down May 6, members can migrate to XChat group chats (supports up to 500 members, aiming for 1,000 soon).

• Cohere acquiring German AI company Aleph Alpha: Canadian AI lab buying the startup once considered Germany's "national AI champion." Schwarz Group (Lidl, Kaufland, owned by Germany's richest person Dieter Schwarz, $45.6B net worth) is investing $600M into Cohere's upcoming Series E as part of the deal. FT says combined entity valued at ~$20B. Purchase price not disclosed. Cohere raised $1.6B to date from Nvidia, AMD, others, valued at $7B last October. Both companies sell AI to governments and regulated industries. Aleph Alpha tried to compete with OpenAI/Anthropic directly, couldn't keep up, pivoted to customized services. Brings German government contracts including the digital ministry. Schwarz Group is also building its own cloud and AI data centers.

• DeepSeek drops V4: Long-awaited flagship model, biggest release since R1 in Jan 2025. Two versions: V4-Pro (coding and complex agent tasks) and V4-Flash (faster, cheaper). Both offer 1M token context window. Performance matches Claude Opus 4.6, GPT-5.4, and Gemini 3.1 on major benchmarks, beats all other open-source models on coding, math, STEM. Pricing is wild: V4-Pro at $1.74/$3.48 per 1M input/output tokens, V4-Flash at $0.14/$0.28. Key innovation: new attention mechanism that compresses older context, cutting compute by 73-90% and memory by 90-93% vs V3.2 for long-context tasks. First DeepSeek model optimized for Chinese chips (Huawei Ascend), though training still appears to have relied partly on Nvidia. Prices could drop further when Huawei’s Ascend 950 supernodes ship at scale in H2.

  

• OpenAI publishes five principles for its AGI push: Altman released a new operating framework covering democratization, empowerment, universal prosperity, resilience, and adaptability. Replaces the 2018 OpenAI Charter. Key shift: less emphasis on AGI specifically, drops the old language about helping rival labs if they got to safe AGI first, and gives OpenAI more room to restrict capabilities when “resilience matters more than user freedom.” The fifth principle (adaptability) is the most telling: the closer AI gets to AGI, the more access may become conditional. OpenAI is now talking like a cloud company, energy customer, and national security asset all at once. The paradox: wants broad access to AGI, but getting there requires massive centralized investment in compute, data centers, and energy that only a handful of companies can afford.

• Meta signs deal for space-based solar power: Partnering with startup Overview Energy, which is developing a system to collect solar energy in orbit and beam it to ground facilities. Meta gets early access to up to 1 GW of capacity. Orbital demo expected 2028, commercial power delivery 2030. Meta is building gigawatt-scale data centers across the US and has already partnered with Vistra, Oklo, and TerraPower for nuclear power. AI’s energy appetite is pushing Big Tech into increasingly exotic power sources.

Who’s Hiring 💼

• Alumni Ventures: Treasury Analyst - Manchester, NH; Senior Associate - Menlo Park, CA

• a16z: GTM Strategy & Ops, Speedrun - Remote (US)

• Planet A: Investment Associate - Berlin

• 500 Global: Analyst, Global Investment Team - Palo Alto, CA

• Lightfield: Community Lead, AI Product Engineer, New Grad, AI Product Engineer, Machine Learning Engineer, Full Stack Engineer, Product & Systems, Founding Account Executive - San Francisco

• Gradient: AI Researcher, Growth & BD, Head of BD, Gradient Ascent Program (Intern) - Remote (Anywhere)

• Natter: Senior Frontend Engineer, Product Manager, Machine Learning Engineer, Senior Backend Engineer - Remote - GMT */-3hrs; Chief of Staff - New York; Founder Associate, Marketing Associate, EA, People & Culture Manager - London

• Atlas: iOS Engineer, Software Engineer, Web Engineer - San Francisco

• Pickle: Founding Frontend Engineer, Founding Platform Engineer - Hillsborough, CA

• OffDeal: Full Stack Engineer, Investment Banker II - New York

• Hamilton AI: Senior Product Builder, Staff Full Stack Engineer - San Francisco

• Stainless: Business Operations, Recruiting Coordinator, Product Designer, Product Manager; Support Engineer - New York/Remote - Pacific Time; Software Engineer, Generalist, Product Marketing Manager - New York/San Francisco

• Bland AI: Content Marketing Lead, Growth Coordinator, Customer Engineer, BDR, Head of Partnerships, Machine Learning Researcher, Audio, Machine Learning Researcher, Multimodal LLMs - San Francisco

• Phonely: Founding Marketing Lead, Channel Partner Manager, Full Stack Engineer, Customer Success Representative, Enterprise Account Manager, DevOps Engineer - San Francisco

• Sett: Senior Software Engineer, Partnership Growth Manager, Partnerships Operations, AI Creative Expert - Tel Aviv-Yafo

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

Share Why Join

See you Wednesday, Ryan

Sponsorships: We are now accepting sponsors for Q2 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome to another edition of Why Join.

Most founders start with the technology and go looking for a problem. The ones who win usually do it backwards. They start with a problem so obvious that everyone assumes someone already solved it, go look, and find out nobody did.

That’s a strange thing about markets. The size of a problem has almost no correlation with whether anyone is working on it. Trillion-dollar inefficiencies survive for decades not because they’re hard to see but because they’re hard to believe. Bills of lading, the documents that prove who owns cargo on a container ship, haven’t meaningfully changed since the 1600s. Truckers still show up to ports with paper originals to get goods released. In 2026. With everything we’ve built. The paper still has to physically cross an ocean before the container moves.

Or take something closer to home. Your dentist’s office probably still sends referrals by fax. Not because fax is good. Because every system it connects to was built around fax, and ripping that out means rewiring the entire chain. So it stays. And a trillion dollars a year in healthcare admin piles up around it like sediment.

The gap between “that can’t be real” and “wait, it is” is where most of the best companies get started. Not the ones building new categories. The ones walking into old ones and doing the work everyone else assumed was already done.

On to who raised this week.

In partnership with

The IPO Pipeline Is Heating Up - Here’s How to Potentially Profit

SpaceX is reportedly preparing for what could be one of the largest IPOs in history.

Canva and several other AI companies may also go public this year as well.

The problem: By the time these companies list, most investors will be buying at the opening price - or higher.

Augment Collective funds give accredited investors access to pre-IPO shares in companies like SpaceX, OpenAI, Anthropic, and Databricks.

• Minimums start at $10K and there’s no management fees and no carried interest.1

Augment is a FINRA-registered broker-dealer that has facilitated over a billion in private market assets, and handles the full process - from accreditation to settlement - in one place.

See available pre-IPO companies before you miss out on Augment.

If you’re interested in sponsoring this newsletter, please reach out here.

Who Just Raised 💰

🇺🇸 Crewline AI

💵 Raised: $7.1M seed, led by Initialized Capital and Nebular. Ford Street Ventures, Entrepreneurs First, and others also in.

One-liner: Aftermarket kit that turns existing construction rollers into self-driving machines.

Why it’s a fave: US construction productivity has fallen over 30% since 1970. Every other industry got more efficient. Construction went backwards. The workforce is aging out... median age is 42, nearly half are over 45, younger workers aren’t replacing them. Operators ghost shifts. Machines sit idle.

Everyone says autonomy will fix this, but Musk has spent a decade and billions on self-driving for open roads and it’s still not fully solved. So why would construction be different? Because a construction site isn’t an open road. A drum roller goes 3 mph in straight lines inside a fenced zone. If it gets confused it just stops. Nobody rear-ends a steamroller. Crewline’s CEO likes to paraphrase Karpathy: the first 90% of autonomy takes as long as the next 9%. Construction lives entirely in that easy 90%.

The product is an aftermarket kit that installs on an existing roller in about an hour without cutting a single wire. Foreman draws a geofence on an iPad, hits go. Five-layer safety system with stereo depth cameras, on-edge object detection, vision-language models that can recognize construction-specific hazards from a text description, independent safety controller, and emergency stops. First real deployment was 30 days on a 30-acre airport job in Austin. Downtime went from six hours a day to under one. Productive hours nearly doubled. Zero accidents.

Hiring: Founding Perception Engineer - San Francisco

🇺🇸 Monk

💵 Raised: $25M Series A, co-led by Footwork and Acrew Capital. BTV returning from seed. Total funding $29M.

One-liner: AI that automates accounts receivable so companies actually get paid.

Why it’s a fave: Software automated basically every other part of running a business. Operations, customer management, paying your own bills. But getting paid by other companies? Still the same. Company sends an invoice. Someone on the finance team starts chasing. Emails, follow-ups, uploading things to payment portals. Customer pays a different amount than what was billed because something doesn’t match. Or they dispute a charge. Or the money shows up in the bank and nobody can figure out which invoice it’s for.

This is trillions of dollars in the US. Most of it still runs on email, spreadsheets, and bank portals. The tools that exist handle the simple part... sending reminders. Monk built for everything else. The mismatches, the disputes, the weird one-off situations that break normal workflows, the back-and-forth that eats 25+ hours a month per finance team. Their AI handles the full cycle from invoice to money in the bank, but every step is locked down with hard rules and tested against thousands of scenarios. When real money is moving between real accounts you really can’t have the system guessing wrong.

Hiring: Founding Growth Marketer, Account Executive, Business Development Representative, Founding Full-Stack Engineer - New York

🇺🇸 Coral

💵 Raised: $12.5M seed, led by Lightspeed and Z47.

One-liner: Automates healthcare admin by working with the fax machine, not replacing it.

Why it’s a fave: The thing nobody outside healthcare realizes is that fax machines still run the entire system. Referrals, prior authorizations (when your doctor has to prove to your insurer that a treatment is necessary), patient intake... all faxes. McKinsey estimates over $1T is spent annually on healthcare admin alone. That’s a quarter of the entire healthcare market just on paperwork.

For a decade, legacy RPA (robotic process automation) vendors like UiPath and Automation Anywhere tried to fix this by mimicking keystrokes and mouse clicks. Combined $4-5B in revenue and they still fail over 30% of the time. Because healthcare is messy. Unstructured data, constantly changing interfaces, multiple stakeholders, fragmented legacy systems. Brittle scripts don’t survive that.

Coral took a different approach. Don’t replace the fax, automate around it. Plug into existing EHR systems, fax lines, payer portals. Handle intake, eligibility verification, prior authorizations, patient communications. Their AI reads handwritten fax forms, scanned insurance cards, prior auth templates at 99.7% accuracy. Traditional RPA in healthcare hits maybe 50%. Patient intakes that took 30+ minutes now run in under five.2

Hiring: ML Engineer, Account Executive, Business Development Representative - New York; Backend Engineer - Platform, Finance Lead - Remote

Open Tabs (stuff we’re reading) 📖

• SpaceX partners with Cursor, has option to buy for $60B: Deal combines Cursor’s coding product with SpaceX’s Colossus supercomputer (~1M H100 equivalent). SpaceX will either pay Cursor $10B for its work or acquire the company for $60B at some point this year. Cursor was valued at $2.5B in Jan 2025, $9B by May, $29.3B by November. Two senior Cursor engineers already left for xAI last month, both reporting to Musk. Neither Cursor nor xAI has models that match Anthropic or OpenAI, and Cursor still sells access to Claude and GPT even as both compete directly with it.

• SpaceX files for $1.75T IPO, biggest ever: Targeting a $75B raise. IPO filing reveals the full financials for the first time. Revenue hit $18.67B in 2025 but the company swung to a $4.94B consolidated loss (was profitable the year before) after absorbing xAI. Starlink is the bright spot: $11.4B revenue, $4.42B operating profit, 63% EBITDA margin, 10M users. The AI segment (xAI) is the drag: capex there surged to $12.7B from $5.6B the prior year. Total capex hit $20.7B. SpaceX ended 2025 with $24.8B cash, $92B assets, $50.8B liabilities. Musk gets dual-class super-voting shares (10 votes each vs 1 for public investors), stays on as CEO, CTO, and board chair. His salary was $54,080 last year but he stands to get 60M shares if market cap hits $6.6T. He also bought $1.4B of SpaceX stock from employees last year through his trust.

• OpenAI launches ChatGPT Images 2.0: Major upgrade to image generation. Two modes: Instant (fast) and Thinking (slower, reasons through the task, maintains character consistency across frames). Supports up to 2K resolution, flexible aspect ratios, up to 8 outputs per run. Biggest shift: image gen is now conversational. You can refine, zoom, adjust without restarting. Better handling of non-Latin scripts (Japanese, Korean, Chinese, Hindi, Bengali).

• Cerebras files to go public: The AI chipmaker filed on Nasdaq under ticker “CBRS.” Revenue hit $510M in 2025 (up 76% YoY), swung to $87.9M net income from a $485M loss in 2024. $24.6B in remaining performance obligations. Huge OpenAI dependency: $20B+ deal to provide up to 750 MW of compute through 2028, OpenAI loaned Cerebras $1B at 6% interest, and got warrants to buy 33.4M shares. Revenue still concentrated in the UAE (G42 at 24%, MBZUAI at 62%). Raised $1B at $23B valuation in February, up from $8.1B in September. Fun fact: Musk tried to buy Cerebras back in 2018. Sam Altman is a personal investor.

• Unauthorized users accessed Anthropic’s Mythos: A few people in a private Discord channel that hunts for unreleased AI models gained access to Mythos via a third-party Anthropic contractor. They guessed the model’s URL based on Anthropic’s naming patterns. The group has been using Mythos regularly since then, though not for cybersecurity purposes. They also claim access to other unreleased Anthropic models. Mythos was only supposed to go to select vendors (Apple, etc.) through Project Glasswing. Anthropic says no evidence its systems were impacted. Still alarming given what Mythos can do: the UK’s AI Security Institute warned last week that it’s a “step up” in cyber-threat capability, able to carry out multi-step attacks and find IT vulnerabilities without human help.

• DeepSeek raising its first outside capital at $20B+ valuation: Tencent and Alibaba in talks to invest. First ever capital raise for the company, which is owned by Chinese hedge fund Zhejiang High-Flyer Asset Management. Was initially looking at $300M at $10B+, now reportedly above $20B. DeepSeek shook the AI world in Jan 2025 with a model matching US rivals despite limited chip access. The investment is likely a hedge for Tencent and Alibaba since open-source models (plus OpenClaw) have made it hard for bigger Chinese players to monetize their own AI. DeepSeek also expanding into agents to compete with OpenClaw, posting 12+ agent-related job listings in March.

• Amazon investing up to $25B more in Anthropic: On top of the $8B already in. Total potential commitment: $33B. $5B going in now at Anthropic’s $380B valuation, remaining $20B tied to commercial milestones. In return, Anthropic committed to $100B+ in AWS spending over the next decade, including current and future Trainium chips. Bringing ~1 GW of Trainium2/3 capacity online by year-end, with up to 5 GW secured total. This lands just two months after Amazon agreed to invest up to $50B in OpenAI.

• New York sues Coinbase and Gemini over prediction markets: AG Letitia James filed suits against both crypto exchanges, accusing them of operating prediction markets that violate state gambling laws. Notable: Kalshi and Polymarket were left out of the suits. The two platforms have dominated the “event contract” space where users bet yes/no on things like the economy or elections. Federal and state regulators still sparring over who actually has oversight of these platforms.

• Tim Cook stepping down as Apple CEO, John Ternus taking over: Cook becomes executive chairman Sept 1 after 14 years as CEO. Ternus, 50, has been SVP of hardware engineering and has been at Apple for 25 years. Apple’s market cap went from ~$200B to $4T under Cook. Revenue nearly 4x’d to $400B+. Ternus’s biggest challenge: AI. Apple has lagged its megacap peers, delayed a Siri upgrade last year, and replaced its AI chief with a Google veteran in December. Updated Siri coming this year, built on Google’s Gemini. Johny Srouji becomes chief hardware officer.

• 44% of songs uploaded to Deezer daily are now AI-generated: That’s ~75K AI tracks per day, 2M+ per month. Up from 10K/day in Jan 2025 when Deezer first launched its detection tool. Actual consumption is tiny (1-3% of streams), and 85% of those streams are flagged as fraudulent and demonetized. AI tracks get removed from algorithmic recs and editorial playlists. An AI-generated track topped the iTunes charts last week in the US, UK, France, Canada, and NZ. Deezer survey found 97% of people couldn’t tell AI music from human-made. Spotify and Apple Music still don’t tag AI tracks at the platform level.

Who’s Hiring 💼

• Roadrunner Venture Studios: Venture Architect - Albuquerque, New Mexico

• ICONIQ: Associate, Investment Specialist - New York

• Balerion AI: Software Engineer, Software Engineer, Applied AI, Founding Account Executive, Strategy & Operations - New York

• Solidroad: Account Executive, Founding Product Designer, Senior Software Engineer - San Francisco/ Business Development Representative, EMEA, Senior Product Engineer - Dublin, Ireland

• Amperos Health: Product Manager, Product Designer, AI Deployment Strategist, Operations (RCM Leader) - NYC/ AI Research Engineer, Software Engineer, Full Stack - NYC; SF/ Billing Associate (AI Quality), Billing Associate (AR & Denials) - India (Remote)

• Mosaic: GTM Associate - Banking & Credit, Founding Designer, GTM Associate - Private Equity, Head of Marketing - NYC

• Realm: Customers & GTM, Business Development Representative, AI Engineer, Product Engineer - Helsinki, Finland

• BuildForever: Backend Engineer, ML / AI Engineer, Product Engineer, Social Growth Lead, San Francisco/New York

• TextQL: Demand Generation / Growth Marketing Manager, Events Marketing Manager, Partner Manager - NYC/Remote

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

Share Why Join

See you Friday, Ryan

Sponsorships: We are now accepting sponsors for Q2 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome to another edition of Why Join.

The market is a grading machine. It grades you on what you are, not what you were. Allbirds made a comfortable shoe. The market gave it a $4B grade in 2021. Then it gave it a $21M grade in 2026. Revenue dropped from $298M to $152M in three years. Same shoe. Fewer people wanted it.

So they sold the brand for $39M, shut every US store, typed “AI” into the company name, and the stock went up 800% in a day.

Like. Come on.

This wasn’t a pivot. Pivots require a product. A fire sale followed by a press release followed by retail traders doing what retail traders do. There’s no AI. There’s no product. There’s a ticker symbol that learned the password.

Which, okay, actually tells you something real. There’s this concept in economics called a signaling game. Markets run on imperfect information, so sometimes the signal becomes worth more than the thing it’s supposed to represent. A Rolex signals wealth whether or not the person wearing it has any. An “AI company” press release signals growth whether or not any growth exists. Signals work because they’re hard to fake long term. Short term though? Embarrassingly cheap. Allbirds up 800% is what the short term looks like.

It happens every cycle. Blockchain in 2017. Metaverse in 2021. A dying company finds the magic word, says it loudly, collects the pop, and we all move on when reality catches up. The word right now is AI. Some companies are borrowing it. A few are actually earning it. Worth knowing the difference.

On to who raised this week.

In partnership with

The AI Insights Every Decision Maker Needs

You control budgets, manage pipelines, and make decisions, but you still have trouble keeping up with everything going on in AI. If that sounds like you, don’t worry, you’re not alone - and The Deep View is here to help.

This free, 5-minute-long daily newsletter covers everything you need to know about AI. The biggest developments, the most pressing issues, and how companies from Google and Meta to the hottest startups are using it to reshape their businesses… it’s all broken down for you each and every morning into easy-to-digest snippets.

If you want to up your AI knowledge and stay on the forefront of the industry, you can subscribe to The Deep View right here (it’s free!).

If you’re interested in sponsoring this newsletter, please reach out here.

Who Just Raised 💰

🇺🇸 Auctor

💵 Raised: $20M (including Series A), led by Sequoia. Microsoft’s M12, HubSpot Ventures, Workday Ventures, OneStream, YC, Dig Ventures, and Tercera also in.

One-liner: AI platform that fixes enterprise software implementations before they go sideways.

Why it’s a fave: Here’s a stat that should make every CTO uncomfortable: more than two-thirds of large-scale technology programs at large companies aren’t delivered on time, on budget, or within scope. The cost of these failures often runs into tens of millions per year per company. And it’s getting worse as enterprises take on more ambitious deployments, especially AI.

The dirty secret is that most failures happen early. Not in the coding or the rollout, but in the discovery phase, when the enterprise is trying to figure out what it actually needs, how the new software integrates with existing systems, and how delivery should work. Get that wrong and everything downstream is expensive garbage. Consultants and integrators charge enormous fees to do this work manually, and it still goes wrong constantly.

Auctor automates that early-stage context building. It helps enterprises and integrators understand needs, map existing systems, and build a clear plan that links decision-making to execution.

Hiring: Software Engineer, Software Engineer, Applied AI, Founding Account Executive, Strategy & Operations - New York

🇺🇸 Pillar

💵 Raised: $20M seed, led by a16z. Crucible Capital, Gallery Ventures, and Uber CEO Dara Khosrowshahi also in. $23M raised to date.

One-liner: Automated hedging platform for commodity-driven businesses that can’t afford a Goldman Sachs trading desk.

Why it’s a fave: If you’re a metals recycler or a food importer, commodity price swings can wreck your margins overnight. Geopolitics has made this worse, not better. The problem is that sophisticated hedging tools have always been a big bank luxury. Goldman has the infrastructure, the talent, the derivative trading desks. A mid-sized metal recycler in Ohio does not. Risk management has been treated as a luxury despite being essential for the businesses that actually produce and move physical goods around the world.

Pillar uses AI to ingest data from client contracts, cash flows, inventories, ERP systems, spreadsheets, even WhatsApp messages (because that’s how global trade actually works), and continuously analyzes exposure across commodities, FX, and freight. It builds and manages a hedge portfolio, adjusts positions automatically based on market conditions, volatility, and risk tolerance, and executes trades. Turns hedging from a periodic manual decision into a continuous autonomous system.1

Hiring: Staff Software Engineer - Full stack, Quantitative Software Engineer Lead, GTM Associate, Business Development Representative, Customer Success Manager - New York; General Manager, Singapore - SG

🇺🇸 Sora Fuel

💵 Raised: $14.6M, co-led by Spero Ventures and Inspired Capital. Engine Ventures and Wireframe Ventures also in.

One-liner: Makes jet fuel from air, water, and clean energy.

Why it’s a fave: Aviation is roughly 3% of global carbon emissions and growing. Everyone agrees sustainable aviation fuel (SAF) is the answer. The problem is nobody can make it cheaply enough. The bottleneck is carbon feedstocks, green hydrogen, and clean power. Traditional direct air capture alone costs $500+ per ton of CO₂. That math doesn’t work for fuel that needs to compete with $3 kerosene.

Sora built technology that captures CO₂ from ambient air and converts it into syngas in a single integrated step, co-producing hydrogen while skipping the sorbent regeneration step that accounts for over 90% of the cost in conventional DAC. They claim carbon capture at under $50 per ton, roughly one-tenth the cost of existing approaches. If that holds at scale, it could enable SAF production below $5 per gallon, which starts to approach unsubsidized cost parity with fossil fuel. That’s the number that matters. Subsidized cost parity is interesting. Unsubsidized cost parity changes the industry.2

🇩🇰 Spektr

💵 Raised: $20M Series A, led by NEA. Northzone, Seedcamp, and PSV Tech also in. ~$26M total raised.

One-liner: AI agents that actually do compliance work, not just manage it.

Why it’s a fave: Financial compliance is one of those areas where the labor is rule-based, repetitive, and absurdly expensive. Analysts spend their days cross-referencing documents, mapping ownership structures, researching registries, manually assessing risk. It’s exactly the kind of work modern AI should be eating, but most “compliance tech” still just helps you organize the workflow rather than doing the actual work inside it. Moody’s, Fenergo, Pegasystems... they manage cases and data. Someone still has to make the determinations.

Spektr’s distinction is small but meaningful: their AI agents perform the analysis end-to-end. Document reviews, ownership mapping, risk analysis, sanctions monitoring. The human stays in the loop for the final decision but doesn’t have to do the underlying grunt work to get there. They sit one layer below the workflow tools and execute.3

Hiring: SDR, Growth associate, Product Manager, Compliance Specialist, Partnership Manager - London; AI Engineer, DevOps Engineer, Implementation Specialist, Customer Growth Manager, Product Designer - Copenhagen

Open Tabs (stuff we’re reading) 📖

• OpenAI acqui-hires personal finance startup Hiro: Hiro offered AI-powered financial planning for consumers. Founded by Ethan Bloch, who previously sold neobank Digit for ~$230M. Backed by Ribbit, General Catalyst, Restive. App is shutting down and data gets deleted, so this is purely a talent grab. Not OpenAI's first fintech acquisition either. Interesting fact: Bloch also built his own autotrading OpenClaw agent called RoboBuffett.

• Snap cuts 16% of staff, blames AI: CEO Evan Spiegel announced ~1,000 layoffs, citing AI improvements that let employees “move more quickly.” Also closing 300+ open roles. Cuts will reduce annualized costs by $500M+ by H2. Q1 revenue ~$1.53B (up 12%), adjusted EBITDA ~$233M. Stock down 31% YTD. AI is the convenient scapegoat of 2026. The reality: just last month activist investor Irenic Capital took a stake and demanded layoffs, telling Spiegel “Like many of your peers, you over-hired. Unlike your peers, you haven’t course corrected.” Snap also cut 20% in 2022 and 10% in 2024. Not exactly out of character.

• Google launches Gemini app on Mac: Option + Space pulls up a floating chat bubble. Can share your active window so Gemini can see what you’re looking at and answer questions about it. Also handles file uploads, image/video/music generation. Looks a lot like Apple’s upgraded Spotlight. But Anthropic’s Claude and ChatGPT for Mac already go further with Computer Use features that let the AI take actions on your behalf. Google playing catch-up on the desktop AI assistant race.

• Anthropic closing in on OpenAI with US businesses: Per Ramp data ($100B in annual spend, 50K+ customers), nearly 1 in 3 US businesses paid for Anthropic tools in March, up 6+ points MoM. OpenAI flat at 35%. Claude downloads tripled to 21M in March while ChatGPT downloads grew just 5%. ChatGPT weekly active users fell MoM in the US for the first time since early 2024. Anthropic says it hit $30B annualized revenue, up from $9B at end of 2025. Claude is now more widely adopted than OpenAI in US info, financial, and professional services.

  

  

• Anthropic shrugs off $800B+ valuation offers: VCs pitching preemptive rounds that would value Anthropic at $800B+, potentially matching or passing OpenAI ($852B post-money). Anthropic isn’t biting, for now. Revenue is at $30B annualized (up from $9B end of 2025), so investors are saying “worth it.” The company has massive capex commitments though: $50B for its own data centers, $30B on Microsoft’s cloud, plus billions/year on AWS. Secondary market demand for Anthropic shares is reportedly “nearly insatiable.”

• SpaceX eyeing $1.75T IPO: Starlink hit $11.4B revenue in 2025 (up 50% YoY) at a 63% EBITDA margin, the only profitable SpaceX segment. Users doubled to 10M by Feb 2026. Aviation revenue up nearly 10x, maritime projected at $1.9B in 2026. The other businesses are a different story: xAI burned $9.5B through Q3 2025 on just $210M in revenue, and rocket launch growth has slowed. SpaceX's $1.75T target works out to a 109x price-to-sales multiple. Quilty Space projects Starlink hitting $20B revenue in 2026. (Amazon's Kuiper is racing to meet its FCC deadline of 1,618 satellites by July to keep its spectrum.)

• AWS launches Amazon Bio Discovery: New AI tool that helps scientists design and test new drug molecules without writing code. Comes with pre-built AI models that generate drug candidates and an AI agent that walks researchers through the process. Early users: Bayer, Broad Institute, Voyager Therapeutics. (19 of top 20 global pharma cos already use AWS.) In a Memorial Sloan Kettering pilot, it generated ~300K possible antibody molecules and narrowed them to 100K worth lab-testing. Months of work done in weeks.

• Amazon buys Globalstar for $10.8B to take on Starlink: Stock and cash deal. Amazon will use Globalstar’s satellites to offer voice, data, and messaging services starting 2028. Also announced an Apple partnership so Amazon’s satellites connect to iPhones and Apple Watches for emergency texting (Globalstar was already doing this with Apple). Comes on top of Amazon’s own Project Kuiper, recently renamed Leo, which launched its first 27 satellites last April. Starlink is still way ahead with thousands of satellites and millions of customers, plus a $1T+ IPO filing this month.

• The Wayback Machine is in trouble: Major news outlets are blocking the Internet Archive from preserving their content. Per Originality AI, 23 major news sites are blocking ia_archiverbot, including the New York Times. USA Today (which owns 200+ outlets) is also blocking. Reddit too. The Guardian doesn’t block but excludes content from the Internet Archive API. Publishers’ stated reason: fear that AI companies will use the archive to train models. The irony is thick. USA Today just used the Wayback Machine in a major ICE accountability story, even while blocking the tool from archiving its own work. Over 100 journalists (Rachel Maddow, Taylor Lorenz, etc.) signed an EFF/Fight for the Future letter supporting the Archive. The Internet Archive has preserved over a trillion web pages in 30 years. No comparable public tool exists.

Who’s Hiring 💼

• Outlander VC: Junior Partner or Partner - San Francisco, Los Angeles, or New York City

• [untitled]: Software Engineer (iOS/MacOS), Software Engineer (Full stack), Software Engineer (Android), Product Designer - New York

• Roamr: Head of Marketing - Remote Europe or US

• Protege: Business Development Representative, GM, New Vertical, Product Manager, Platform, Product Manager - Privacy, Rights & Trust, Solutions Engineer (Media), Partnerships Operations Lead (Healthcare) - Remote

• maintouch: Engineer (Full stack + AI), Founding Growth Lead - New York

• Plume: Software Engineer, Founding Engineer, Founding Deployment Strategist - Paris, France

• AfterQuery: Software Engineering Intern, Software Engineer - Platform/Applied AI (Full stack), Marketing Lead, Strategic Projects Intern - Coding, Strategic Projects Associate - Coding, Growth Associate, Growth Intern - San Francisco

• Fly.io: Support Engineer - Remote (UTC-7 to UTC-4, UTC+7 to UTC+13)

• Peec AI: Brand Designer, Community Manager, Customer Experience Associate, Financial Analyst - FP&A, Founding SDR, Senior Full stack Engineer, Data Scientist, Internship (Product, GTM, Talent, HR, or Operations) - Berlin, Germany

• Stacks: Growth Marketer, Founding Product Marketing Manager, Senior Full stack Engineer, Accounting Engineer, Account Executive - Mid Market, BDR - Mid Market & Enterprise - Amsterdam; London

• Wordware: Senior Frontend Engineer, Ex-YC Founder, Product Designer - San Francisco

• Retell AI: Research Scientist - LLM, GTM Recruiting Lead, Founding GTM Engineer - San Francisco

• Superset: Former Founder - San Francisco

• Superblocks: Software Engineer - AI Platform, Infrastructure Engineer / SRE, Sales Engineer, Product Manager - AI Agent - New York

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

Share Why Join

See you Saturday, Ryan

Sponsorships: We are now accepting sponsors for Q2 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome to another edition of Why Join.

Software ate the world. Everyone knows that line. What people missed is the footnote: it ate the parts of the world that were already legible. The parts that could be modeled, digitized, put in a database.

Everything else stayed analog. Not because it was unimportant. Because it was too messy, too contextual, too dependent on judgment calls made by humans who’d seen a thousand similar situations. You couldn’t write rules for it so you just hired people to do it.

That’s the actual unlock with AI. Not that it’s smarter. That it can operate in mess. In ambiguity. In the parts of the economy that software couldn’t touch.

That is an absolutely enormous amount of the economy.

Let’s get into it.

Why Join Exclusive: You Could Be an Investor in High-Potential Startups Like These

Why Join and Alumni Ventures are teaming up this week only to give readers early access to high-growth startup opportunities, including some of today’s most exciting AI, Deep Tech, Quantum Computing, Cybersecurity, and Space companies co-invested alongside top VC firms like Andreessen Horowitz (a16z), Bessemer, & Y Combinator.

You get:

• Curated deal flow of high-potential startups

• AV is already investing alongside elite lead venture firms in these deals

• No cost to see deals

• No obligation to invest

Don’t miss your chance before access closes.

Join Alumni Ventures Syndicate Today →

Who Just Raised 💰

🇺🇸 Moda

💵 Raised: $7.5M seed, led by General Catalyst. Pear VC, Jeffrey Katzenberg’s WndrCo, and execs from Dropbox, Stripe, Segment, Google, and Scale AI also in.

One-liner: AI design agent that actually learns your brand and produces fully editable work.

Why it’s a fave: AI image generators made it easy to create visuals, but the outputs are generic, can’t be edited, and have zero brand awareness. As AI-generated content floods every channel, looking on-brand is becoming the clearest way to stand out. And the most important materials in a company (the pitch deck that closes a round, the sales collateral that wins a deal) still require taste.

Moda gives every company an AI agent that learns their visual language. It indexes your website, Google Drive, past decks, existing assets, and builds an understanding of your brand that gets sharper with every design you create. The output is fully editable on a layered canvas, not a static image. Presentations, social posts, reports, ads. It understands layout, typography, color. Imports and exports from Google Slides and PowerPoint so there’s no workflow disruption.

Founding team is stacked. CEO Anvisha Pai co-founded Dover (AI recruiting, $23M raised from YC, Founders Fund, Tiger Global). COO Ravi Parikh co-founded Heap ($960M valuation, acquired by Contentsquare) and Airplane (acquired by Airtable). CTO John Holliman was employee #1 at Dover.

Hiring: Chief of Staff, Founding AI/ML Engineer, Founding Engineer, Software Engineer - New Grad - NYC

🇺🇸 Glimpse

💵 Raised: $35M Series A, led by a16z. 8VC and YC also in. $52M raised to date.

One-liner: AI agents that recover money retailers wrongly deducted from brands.

Why it’s a fave: Here’s how retail works: a brand ships product to a retailer, sends an invoice, retailer pays. Except retailers routinely pay less than what’s billed and attach a reason (damaged goods, short shipment, whatever). Some deductions are legit. A lot aren’t. A brand might ship inventory perfectly and still get charged for a short shipment. If nobody catches it, that’s just... lost revenue. Quietly, over and over.

The process to actually dispute these is brutal. Teams log into multiple retailer portals, pull scattered documents, review line items, reconcile against internal records, manage disputes end to end. It’s fragmented, unstructured data across siloed systems and teams. Most brands either eat the losses or throw bodies at the problem.

Glimpse’s AI agents log into retailer portals, centralize all the documents, classify each deduction, validate it against internal data (supply chain records, promotion calendars), then automatically file disputes, follow through, apply recovered cash, and sync everything back to the brand’s ERP. Humans stay in the loop for quality assurance and following up on resolution. The system gets smarter with every deduction it processes, building a compounding data advantage across the network.

Hiring: Sales Development Representative, Head of Operations, Growth Marketing Manager, Software Engineer, Customer Success Manager - NYC

🇺🇸 Trayd

💵 Raised: $10M Series A, led by White Star Capital. YC, Suffolk Technologies, and RXR Realty also in. Total funding now $17M.

One-liner: Back-office operating system for specialty trade contractors in construction.

Why it’s a fave: Specialty trade contractors (electricians, plumbers, ironworkers, concrete crews) outnumber general contractors, but almost all construction tech has been built for GCs. The trades get stuck with ADP, Paychex, or Excel. And construction payroll is uniquely painful. A single worker might earn four different pay rates in one day depending on the trade task, project scope, and jurisdiction. Union rules, prevailing wage requirements, multistate taxes. Payroll admins are literally receiving paper timesheets, hand-calculating variable rates, then double-entering numbers into a generic payroll system and again into accounting software. What takes 14 hours of manual work, Trayd does in under 30 minutes.

Hiring: Product Operations Associate, Senior Software Engineer, Business Development Representative - Sales, Sales Account Executive, Senior Implementations Analyst, Enterprise - New York

Open Tabs (stuff we’re reading) 📖

• Nvidia-backed Reflection AI raising $2.5B at $25B valuation: The “DeepSeek of the West” is building open-source US AI models to counter Chinese offerings. Nvidia invested ~$800M in a previous round and has been introducing Reflection to potential customers, including foreign governments building sovereign AI infrastructure. JPMorgan in talks to participate through its new $10B Security and Resiliency Initiative. Already cut a deal with South Korea’s Shinsegae Group to build Korean-language models. Part of Nvidia’s new Nemotron Coalition alongside Cursor and Thinking Machines Lab. CEO Misha Laskin, ex-DeepMind: “Open models are Trojan horses for the infrastructure they bring with them.” Investors include Sequoia, Lightspeed, CRV, and 1789 Capital (where Donald Trump Jr. is a partner).

• Physical Intelligence raising $1B at $11B valuation: The two-year-old robotics startup founded by ex-DeepMind researchers is in talks to double its valuation in four months. Founders Fund and Lightspeed set to participate. The company builds AI models for robots, has deployed on robotic arms for tasks like folding clothes, making coffee, assembling boxes. Already raised $1B+ from Bezos, CapitalG, Thrive, Lux. Robotics AI is hot right now: Skild AI at $14B, Figure AI at $39B, and Big Tech (DeepMind, Apple, Meta) all piling in.

• ChatGPT ads hit $100M annualized revenue in 6 weeks: And that’s from less than 20% of eligible users actually seeing ads. 600+ advertisers on the platform now. Self-serve access launching in April. OpenAI hired former Meta ad exec Dave Dugan to run sales. Expanding to Canada, Australia, NZ soon. This is key to the IPO story: execs told investors they expect $17B+ from ChatGPT consumers in 2026. (Early pilot had rough CTR numbers, nearly 7x below Google benchmarks, but scale is clearly there.)

• And then there were none at xAI: Musk’s last remaining cofounder, Ross Nordeen, is out. That’s all 11 original cofounders gone in under three years. Eight left since January alone, most after the SpaceX merger. Musk has cut teams working on Grok Imagine and Macrohard (its AI agent project), gone through multiple reorgs, and admitted publicly that “xAI was not built right first time around.” The company’s valued at ~$250B but still trails OpenAI and Anthropic on scale. They’ve hired a few senior Cursor engineers but the talent bleed is hard to spin.

• Mistral secures $830M in debt financing for Paris data center: On top of the €1.2B Sweden investment from February, Mistral is now building a Paris-area data center powered by 13,800 Nvidia GB300 GPUs (44 MW capacity). Backed by seven banks including BNP Paribas and HSBC. Aiming for 200 MW across Europe by end of 2027. For context: Mistral has raised $2.9B total. OpenAI has raised $180B, Anthropic $59B.

• NYSE owner drops another $600M into Polymarket: ICE (parent of NYSE) added $600M on top of the $1B it invested in October, bringing total commitment to nearly $2B. Polymarket also acquired a licensed exchange and clearinghouse this year and partnered with Palantir to build a trading surveillance system. Meanwhile rival Kalshi raised $1B+ at a $22B valuation and is pulling ~$1.5B in annual revenue. The prediction market space is getting real, fast, even as lawmakers question whether these platforms are vulnerable to manipulation.

• Claude’s consumer growth is skyrocketing: Analysis by Indagari of billions of credit card transactions from ~28M US consumers (conducted for TechCrunch) shows Claude gaining paid subs in record numbers. New subscriber growth spiked between the late January DoD feud coverage and Amodei’s public statement in February. Anthropic says paid subs have more than doubled this year. The Super Bowl ads mocking ChatGPT’s decision to run ads helped too. Claude Code, Cowork, and the new Computer Use feature (lets Claude navigate your computer independently) are all driving upgrades. ChatGPT still way bigger overall, but OpenAI’s uninstalls spiked after the Pentagon deal while Claude hit #1 on the App Store.

  

  

  

• Waymo hits 500K paid rides per week: 10x growth in under two years (50K/week in May 2024). Now in 10 US cities, added seven Sun Belt markets in the past year alone. Fleet is still ~3,000 vehicles, meaning they’re squeezing way more utilization out of each car. Still tiny vs Uber (13.5B trips in 2025) but the lead over every other robotaxi company keeps growing. Tesla launched paid rides in Austin in January but still lacks permits for California. 6th gen system coming on the Zeekr Ojai minivan and Hyundai Ioniq 5.

  

• OpenAI kills Sora, $1B Disney deal dies with it: OpenAI scrapped its video gen app and the three-year Disney partnership collapsed with it. Disney had pledged $1B in equity plus licensing of 200+ characters from Disney, Pixar, Marvel, Star Wars. No money ever changed hands. Sora was eating compute without the revenue to justify it, downloads cratered from 6.1M in November to 1.1M in March, and it was losing to Google, Kling, and others. Apps chief Fidji Simo told staff to drop “side quests” and focus on coding and enterprise. OpenAI also deprioritized the “adult mode” sexting feature. The pivot is clear: all in on competing with Anthropic on coding and enterprise.

Who’s Hiring 💼

• Oximy: GTM Intern (Founder’s Office), Content Intern (Product + Founder Content), Founding Full Stack Engineer - San Francisco; Remote

• Bastion: Account Executive, Senior Software Engineer - NYC

• Truewind: Accounting Agents Solutions, Growth Business Development Representative (BDR) - San Francisco; Remote

• Mirai: Machine Learning Engineer - Model Optimization, Inference Engineer - Remote; San Francisco; Europe

• Clasp: Data Engineer, Full Stack Engineer, B2B Demand Generation Lead, Executive Partner to the COO - Boston, MA/ Recruitment Marketing Manager, Manager, Student Engagement - Remote

• Eragon: Applied AI Intern, AI Product Manager, Applied Research Engineer - San Francisco

• Mutable Tactics: ML Engineer (perception & state estimation), Robotics Engineer (planning & behaviour) - London

• Notch: Product Marketing, Demand Generation & ABM Lead - New York/ Product Designer - Ramat Gan, Israel

• Chexy: Marketing & Events Intern - Summer 2026, Customer Success Manager, Full Stack Engineer, Chief of Staff - Toronto

• Ambrook: Software Engineer, AI, Content Marketing Lead - New York; Denver; San Francisco; Remote/ Founding Operations Associate - New York

• Top Tier Capital Partners: Investment Senior Analyst or Associate - London

• Alumni Ventures: Platform Operations Analyst Internship - New York/ Nurture Marketing Intern - Manchester, NH

• a16z: Partner 6, Executive Assistant, Crypto - New York

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

Share Why Join

See you Wednesday, Ryan

Sponsorships: We are now accepting sponsors for Q2 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome to another edition of Why Join.

One of the more interesting patterns in technology is that progress rarely breaks things all at once. It accumulates quietly… until the systems built for a previous constraint start to fail under new conditions.

We’ve seen this before. Databases had to evolve when data exploded. Cloud computing emerged when fixed infrastructure couldn’t keep up with variable demand. And now, AI is creating a similar pressure, not just on models, but on the workflows built around them.

What’s notable is that the shift isn’t purely about capability. It’s about coordination… how work is generated, tracked, and executed when both humans and machines are contributing at a much higher pace.

This issue looks at a few places where that pressure is already visible, and how new systems are emerging to handle it.

A message from Guideless

Most product and ops teams document the same way. Someone records a Loom, drops it in Notion, and calls it done. Six months later the product has changed, the Loom is wrong, and new hires are still asking the same questions on Slack.

Guideless is a faster path. Install the Chrome extension, click through your workflow like you normally would, and it generates a narrated video guide on the other side automatically. AI writes the script, generates the voiceover, adds captions, and applies your branding. No recording setup, no editing, no script prep.

Teams use it for customer onboarding, sales demos, internal SOPs, and support docs. Guides are shareable via link or embeddable straight into your help center. When something changes in your product, you update the guide without starting over.

If your docs are always two releases behind, this closes that gap fast.

New users get 20% off for 3 months.1

Try it for free →

Who Just Raised 💰

🇺🇸 Namespace

💵 Raised: $23M Series A, led by NEA. Susa Ventures and Burst Capital also in.

One-liner: High-performance compute environments for CI/CD in a world where AI agents are shipping code too.

Why it’s a fave: CI/CD pipelines were designed for a world where humans submitted a manageable number of pull requests. Now you’ve got agents and humans shipping code across multiple repos at much higher volume. Branches, builds, and test runs pile up, queues form, and every extra minute of build time gets multiplied across far more PRs. The pipeline as we know it, a static, sequential checklist built for human-paced development, isn’t designed for this.

Namespace replaces that with on-demand compute environments where both agents and humans can build, test, and deploy at their own speed. It drops into your existing setup (GitHub Actions, self-hosted runners, Kubernetes) without forcing you to rearchitect anything. Turbo Docker builds, remote caching, workflow analytics, built-in observability, and reproducible dev environments that spin up in seconds.

Hiring: Product Engineer, Developer Marketing Lead, Infrastructure Software Engineer - Switzerland; Germany; New York; San Francisco surroundings; United Kingdom

🇺🇸 PointOne

💵 Raised: $16M Series A, led by 8VC. Bessemer, General Catalyst, and YC also in. Total funding now $20M.

One-liner: AI-powered time tracking that turns law firm time data into real business intelligence.

Why it’s a fave: Law firms run on time data. Every pricing decision, staffing call, and profitability analysis flows from it. And the tools to capture that data have been, to quote the founders, “embarrassingly broken” for decades. Lawyers manually reconstruct their days at the end of the week, entries are vague or missing, and the downstream mess means pricing is a guess and profitability is a quarterly surprise.

PointOne automates time entry with AI so every hour is captured and the data is clean from the start. When that foundation works, everything else gets better. Pricing becomes defensible, staffing decisions are actually informed, and profitability becomes something you manage instead of something that happens to you.

They’ve expanded beyond time entry into bill review, outside counsel guidelines compliance, and pricing intelligence. Basically, treating timekeeping as the entry point to a much larger data strategy for firms.

Hiring: Customer Success Lead, Founding Forward Deployed Engineer, New Grad Software Engineer, GTM Engineer, Software Engineer, Growth Fellowship - NYC

🇺🇸 Littlebird

💵 Raised: $11M seed, led by Lotus Studio. Angels include Lenny Rachitsky, Scott Belsky, Gokul Rajaram, and DocSend co-founder Russ Heddleston.

One-liner: Desktop AI that reads your screen all day and builds a deep understanding of your work and life.

Why it’s a fave: Littlebird is built around what it calls “quiet computing.” It sits in the background, reads the text on your screen across apps, and builds a running picture of who matters to you, what you’re working on, and what you care about this week and this year.

You’ve seen this idea before. Rewind (now sold to Meta) and Microsoft Recall both tried to capture everything on your screen. The problem: they stored screenshots, which is data-heavy, kind of creepy, and the search experience wasn’t great. Littlebird takes a different approach. It “reads” your screen but only stores text, no visual data. Way lighter, way less invasive.

It runs quietly in the background, captures context from whatever you’re doing, and only shows up when you ask it something. You can customize which apps it ignores (password managers and sensitive fields are excluded automatically). Connect Gmail, calendar, etc., and it builds a live picture of your work life.

There’s also a Granola-style meeting notetaker built in that uses system audio, plus a “Prep for meeting” feature that pulls context from past meetings, emails, and company history. It also has “Routines,” basically recurring prompts like daily briefings or weekly summaries.

Hiring: Product Analyst, Brand (Visual) Designer, GTM Associate - Growth Marketing, Swift Engineer, Applied AI Engineer - Agent Intelligence & Retrieval - Remote

Open Tabs (stuff we’re reading) 📖

• OpenAI plans to nearly double workforce to 8,000 by end of 2026: Hiring ~12 people/day, just leased 1M+ sq ft in SF. Big reason: Anthropic is eating their lunch with business customers. Ramp data shows first-time AI buyers choosing Anthropic at 3x the rate of OpenAI, a full reversal from last year. (OpenAI called the data "insane.") Altman issued a "code red" back in December, and apps chief Fidji Simo told staff to ditch "side quests" and focus on Codex, enterprise, and turning ChatGPT into a productivity tool. OpenAI expects half its revenue from business customers by year-end, up from 40%.

  

• ChatGPT ads pilot and nobody can prove they work: OpenAI got WPP, Omnicom, Dentsu into an alpha test but early numbers look weak. Per The Information, one brand saw ~0.91% CTR, nearly 7x below Google search benchmarks. Another spent just 3% of a $250K budget after weeks. OpenAI’s own Ad Manager is buggy, advertisers can’t properly see their data. There’s also a structural issue. Ads only reach free and Go users, paying subs never see them. Altman used to call ads a “last resort.” Pilot ends this month and so far no clear ROI. Despite that, OpenAI is doubling down. Hired Meta ads veteran David Dugan to lead global ads, building out its ad stack, and asking brands for ~$200K minimum commitments.

• Tesla, SpaceX announce $25B “Terafab” chip factory: Musk unveiled a joint venture between Tesla, SpaceX, and xAI to build what he calls the biggest semiconductor fab ever in Austin, TX. Targeting 2nm chips, 100K wafer starts/month scaling to 1M (that’s ~70% of TSMC’s entire global output, from a single building, by companies that have never made a chip). 80% of compute would go to orbital AI satellites, 20% ground-based. Tesla’s CFO said the $20-25B cost isn’t even in their 2026 capex plan yet. For context: TSMC spent $165B and decades building six fabs in Arizona. Tesla has zero chip fab experience. As Electrek put it: “This is Battery Day on steroids. And if you’ve been following how that turned out, you should be very skeptical.”

• SoftBank might go past its own leverage limit to push harder on AI: CFO said the 25% LTV cap can be exceeded. That cap was supposed to be their safety line. Now it’s flexible, so expect more borrowing. Reason is simple. They’re spending big on AI. $100B Stargate with OpenAI and Oracle. That kind of spend needs cash. Either sell assets or take on debt. Looks like debt. Arm is the swing factor. It’s a big part of their assets and the stock has been moving a lot. If it drops, leverage goes up even without new borrowing. This matters because once you bend a “hard limit,” people stop trusting it. SoftBank already takes big swings. More debt on top makes things more fragile. Main risk is if Arm falls while debt is high. That’s how things got messy in 2020.

• Bipartisan bill targets prediction markets like Kalshi, Polymarket: US senators want to ban sports betting on these platforms, saying it’s basically gambling under a different label. Would not touch FanDuel or DraftKings since those are state-regulated. Prediction markets sit under federal rules (CFTC), so they’ve been able to operate nationwide. Lawmakers are saying that’s a loophole. Kalshi alone saw ~$1B in Super Bowl trades, up ~2700% YoY. Pushback is predictable. Kalshi says this is casino lobbying trying to kill competition and will just push users offshore. There’s also real concern around addiction. Data shows gambling-related searches jumped ~61% after online betting expanded. Platforms are already under pressure. Kalshi is banned in Nevada, facing issues in Arizona, and recently cracked down on insider-style betting. Polymarket also tightened rules around using non-public info.

• OnlyFans billionaire Leonid Radvinsky dies at 43: The owner who turned OnlyFans into a massive porn platform passed away after cancer. He bought the company in 2018 and scaled it into a cash machine, reportedly making ~$1.9M a day at peak. OnlyFans exploded during Covid. Users went from ~13M in 2019 to ~188M by 2021. Revenue hit ~$1.4B in 2024, with ~$7.2B spent on the platform. Radvinsky’s net worth reached ~$4.7B. His background was controversial. Started young with spammy porn sites that promised illegal content but mainly monetized clicks. Later built MyFreeCams before acquiring OnlyFans. He kept a low profile. Rarely gave interviews and stayed out of the spotlight despite running one of the biggest creator platforms online.

• Anthropic pushes deeper into AI agents with Claude controlling your computer: New feature lets you send a task from your phone and Claude completes it on your PC. It can open apps, browse, edit files, even send emails. This is part of the agent race. After OpenClaw went viral, everyone is trying to build “do things for you” AI, not just chat. Nvidia called OpenClaw the next ChatGPT. OpenAI already hired its creator. Use case is simple. Ask Claude to export a deck, attach it, send it. It just does it. Still early though. Anthropic says it can make mistakes and will ask permission before accessing apps.

• Delve “fake compliance” scandal getting worse and investors are backing off: YC-backed startup accused of fabricating audit evidence for customers. Claims include fake board meetings, fake tests, and reports that never actually happened. This is not small. Delve sells compliance for things like SOC 2, HIPAA, GDPR. If that’s fake, customers could be exposed to real legal risk. Allegations say the whole system was inverted. Delve generates the evidence, then audits get “rubber stamped” instead of independently verified. Fallout is already showing. Insight Partners pulled its investment post. Delve also reportedly paused demos, which is basically growth on hold. Delve denies it. Says it’s just automation + templates, and real audits are done by third parties. Bigger issue here is trust. Compliance is not a nice-to-have product. It’s the product. If that breaks, the whole thing collapses.

Who’s Hiring 💼

• BlueFlag Security: Product Manager, Software Engineer - Hyderabad, India; Data Scientist - SF Bay Area

• Cape: Operations Analyst, Enterprise, Full Stack Software Engineer - New York/ Software Engineer - New York; Washington

• RunSybil: Account Executive, Software Engineer - NYC/ Customer Success Manager, Machine Learning Engineer - NYC; SF

• Gimlet Labs: Founding Marketer, Member of Technical Staff (Intern), Member of Technical Staff - AI Research (Intern) - SF

• Dash0: Head of Design, Senior Marketing Operations Manager, RevOps Manager, Digital Marketing Manager - EMEA - Remote/ Talent Acquisition (Intern) - Germany - Remote/ Commercial SDR - US, SDR Manager - US - New York/ Partnerships Manager - Amsterdam; London

• Parallel: Account Executive, Operations Associate - Paris

• Edra: AI Engineer, Software Engineer - Full Stack - London/ AI Engineer, Product Designer, Software Engineer - Full Stack, Office Manager & Executive Assistant - New York

• Corridor: AI Lead, AI Engineer, Software Engineer, Founding Technical Recruiter, Marketing Manager, Community Manager - SF

• redalpine: Visiting Analyst (early stage investments) - Zug; Berlin; London

• M13: Investment Summer Associate - AI Tooling - SF

• defy.vc: Analyst - Remote

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

Share Why Join

See you Thursday, Ryan

Sponsorships: We are now accepting sponsors for Q2 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome back to Why Join.

So there’s this pattern in tech that keeps repeating. The most exciting new thing and the most dangerous new thing are always the same thing. Every major platform shift starts like this. Useful before it’s secure. Adopted before it’s understood. The people who move first get a massive edge and the people who move first without thinking get wrecked. That’s where we are right now with agents. And if you’ve been building long enough you know the window between “this is a toy” and “this is infrastructure” is when the real companies get built.

Alright let’s get into it.

Sponsored by CoinTracker

File crypto taxes in minutes

Easily import your crypto transactions and file your taxes in minutes with TurboTax, H&R Block, or your own tax professional. Save time and maximize your refund using our crypto tax software.

Get started

💰 Who Just Raised

🇺🇸 Advocacy

💵 Raised: $3.5M seed, led by Relentless. Relativity (via Rel Labs) and Am Law 100 firm Fenwick & West also in.

One-liner: AI litigation workspace that actually remembers your case.

Why it’s a fave: Most AI legal tools right now do the same thing: draft a motion, summarize a deposition, maybe do some research. But litigation isn’t a series of one-off tasks. It’s a months- or years-long process where context is everything, and right now that context lives scattered across e-discovery platforms, research tools, and... litigators’ heads.

Advocacy built what they call a “case memory” layer. It centralizes all matter-critical intelligence (documents, audio, video evidence, timelines) into one workspace that gets smarter as the case develops. Instead of keyword search, it does context orchestration. So when you’re drafting a brief at month 14 of a case, the AI actually knows the full history, the evidence, the arguments, all of it. They can also ingest raw audio and video and map it into the case timeline automatically.

🇩🇪;🇺🇸 Lio

💵 Raised: $30M Series A, led by a16z. SV Angels, Harry Stebbings, and YC also in.

One-liner: Agentic AI workforce that actually does procurement, not just tracks it.

Why it’s a fave: Enterprises spend $180B a year on procurement talent versus $10B on procurement software. That ratio tells you everything. Despite all the eProcurement tools out there, decisions still happen at human speed. Requests move through layers of systems, rules, approvals. Teams scale by adding headcount or outsourcing at up to 20x the cost of software.

Lio’s approach is different from most procurement startups. Instead of building another tool that sits on top of the process, they built AI agents that do the actual work. Triaging requests1, analyzing quotes, comparing suppliers, negotiating, onboarding vendors, executing purchases. End to end across ERPs, inboxes, contracts, the open web. They call it Agent Operating Procedures (AOPs)2, basically teaching AI agents to follow the same playbooks as experienced buyers.

Hiring: Founders Associate - CEO Office, Sales Development Representative, Partnership Manager, Event Marketing Manager, Software Engineer, Account Executive - New York; Munich

🇮🇱;🇺🇸 Fig Security

💵 Raised: $38M (seed + Series A). Investors include Team8, Ten Eleven Ventures, plus former Splunk CEO Doug Merritt and former Palo Alto Networks CMO Rene Bonvanie.

One-liner: Monitors your security stack to make sure your defenses actually work.

Why it’s a fave: Modern enterprise security stacks are absurdly complex. Dozens of tools all wired together, and a small change in one can quietly break detection and response somewhere downstream. The scary part is you might not find out until something bad happens. It’s like an alarm you haven’t tested in months... do you actually trust it still works?

Fig traces data flows through the entire security stack, from source through pipelines and data lakes to your SIEM3 and response platforms. But instead of following data forward to see where it goes, they work backwards from your detections4. That’s the thing that needs to work, so they back-trace what has to happen for each detection to fire correctly, then alert you in real time when something breaks the chain. You can also simulate how a new patch5 or change would affect things before you deploy it.

📖 Open Tabs (Stuff We’re Reading)

• Anthropic sues the Pentagon: The company is taking the Defense Department, Hegseth, and a bunch of other officials to court over the supply-chain risk designation. Anthropic's argument: this was retaliation for a policy disagreement, not an actual security threat. (That designation is usually reserved for adversary-nation entities like Huawei, so yeah.) Trump ordered all federal agencies to stop using Claude with a six-month transition window, which Anthropic points out kind of proves how essential its tools are. The White House called Anthropic a "radical-left, woke company." Microsoft and Google say they'll keep working with Anthropic commercially. The $200M Pentagon contract? In limbo.

• Hayden AI sues ex-CEO: The SF startup that builds spatial analytics tools for cities is claiming co-founder and former CEO Chris Carson forged board signatures, sold $1.2M in stock without authorization to buy a home in Boca Raton and a gold Bentley, then had an employee download his entire 41GB email file onto a USB stick on his way out. He’s since started a rival company called EchoTwin AI. Oh, and Hayden AI says his entire CV is fabricated—claims he wasn’t getting a PhD from a Tokyo university in 2007 but was actually running a paintball equipment shop in a Florida strip mall. Hayden is valued at $464M.

• Google hands Pichai a $692M pay package: Three-year deal, almost entirely stock-based and performance-tied. Includes PSUs linked to Alphabet’s total shareholder return vs. S&P 100 companies (0-200% payout), plus ~$130M in incentives tied to Waymo’s growth and $45M tied to drone delivery unit Wing. His base salary hasn’t changed since 2020: $2M, no bonus.

• Palmer Luckey’s retro gaming startup wants $1B valuation: ModRetro, Luckey’s side project that makes high-end vintage gaming handhelds, is raising at a billion-dollar valuation per the FT. First product was the Chromatic, a Game Boy-style device that reviewers loved. Now working on an N64-inspired device. (As The Verge put it: “If Lockheed Martin made a Game Boy, would you buy one?”) Meanwhile, his day job is going well too: Anduril is reportedly raising at $60B.

• Anduril revenue doubles to ~$4.3B, but losses widen: Defense tech company saw sales double, with its Altius long-range strike drone now the biggest revenue driver. Missiles and fighter jets are expected to take a larger share going forward. The company is spending $900M+ on a factory outside Columbus, Ohio, that should open within months. The catch: $1.2B in operating losses expected this year, up from $800M in 2025, with $1B+ annual losses projected for the next four years. Gross margin dipping to 36% in 2026 before climbing back to ~40% by the end of the decade.

  

• Kalshi and Polymarket both eyeing $20B valuations: Per the WSJ, both prediction market platforms are in early fundraising talks that would roughly double their late-2025 valuations. Kalshi (CFTC-approved, $11B last round) is at ~$1.5B annualized revenue run rate. Polymarket ($9B after Intercontinental Exchange’s $2B investment) is right behind. Both are doing nearly $2B in weekly trading volume. The space is getting crowded though, with Coinbase, Robinhood, and now Nasdaq and Cboe all exploring prediction markets.

• OpenAI hardware exec quits over Pentagon deal: Caitlin Kalinowski, who ran OpenAI’s robotics team, resigned over the company’s new DoD agreement, saying surveillance without judicial oversight and lethal autonomy without human control “deserved more deliberation than they got.” OpenAI jumped on the Pentagon deal after talks with Anthropic fell apart over those exact same safeguards. (Trump then ordered all government departments to cut ties with Anthropic and the Pentagon slapped it with a supply-chain risk designation, the kind of thing usually reserved for companies like Huawei.) Even Altman admitted the whole thing looked “opportunistic and sloppy.” The consumer fallout has been something: ChatGPT uninstalls surged 295% and Claude climbed to #1 on the App Store. Anthropic says it’s taking the designation to court.

• Flipkart moves HQ back to India, IPO coming: Walmart-owned Indian e-commerce giant relocated from Singapore to India as it targets a domestic IPO by March 2027. GMV hit ~$30B in 2025, up from $23B in 2021, with 500M+ customers and 1.6M sellers. Part of a broader wave of Indian startups redomiciling ahead of listings, following Zepto (filed confidentially for an IPO in December) and Groww (went public last year).

💼 Who’s Hiring

• Sage: Sales Development Representative, Client Implementation Specialist - New York; Remote - CST, MST, or PST

• Diligent: Full Stack Engineer - London

• Cosmon: AI/ML Engineer - Palo Alto

• Guided: Founding Product Engineer (full stack, backend), Founding AI Engineer - Paris

• Rillet: Founders Associate, Digital Marketing Manager, Content Marketing Manager, ML/AI Engineer, Software Engineer, Accountant - New York City; San Francisco

• David: Marketing Designer, Creative Strategist - New York

• City Detect: Business Development Representative, AI Engineer, Full Stack Engineer - Remote (US)

• Patlytics: Full Stack Engineer - Remote (APAC); Full Stack Engineer, Business Development Representative, IP Attorney / Strategic Business Development Lead - New York

• 360 Capital: Analyst - Paris

• Equal Ventures: Analyst/Associate - New York City

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

See you Wednesday, Ryan

Sponsorships: We are now accepting sponsors for Q2 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome back to Why Join.

So uh I read this piece from Jason Lemkin on SaaStr a few weeks back and it really stuck with me. For like a decade there was this almost mechanical formula in SaaS right. Hit $20M ARR, show 110%+ NRR, don’t burn too much, grow at least 20%. Do those four things and PE shows up with a term sheet. That was the safety net. You didn’t need to be a rocket ship, you just needed to be like good enough. That whole floor? Gone. PE is buying AI-accelerated growth stories and hot sectors now. Plain vanilla SaaS with okay growth and no AI angle? Crickets. Nobody’s coming to save you. Build accordingly.

Alright let’s get into it.

In partnership with Nibbles

The first pet rewards credit card with free pet insurance, meet Nibbles

Vet bills aren’t getting any cheaper. Costs have jumped more than 60% in the last decade, and pet owners are feeling it. But one new card is actually helping lighten the load.

The Nibbles Pet Rewards Credit Card offers 3x rewards on pet purchases—from vet visits to toys—and even includes free pet insurance for one eligible pet, with no annual fee. It’s a smart way to save while keeping your best friend happy and healthy.1

Reward your pet now

💰 Who Just Raised

🇺🇸 InScope

💵 Raised - $14.5M Series A. Investors include Norwest (lead), Storm Ventures, Better Tomorrow Ventures and Lightspeed Venture Partners.

One-liner: AI platform that automates the grunt work of financial reporting.

Why it’s a fave: Anyone who’s been near a 10-K knows the process is brutal. Financial statements get patched together across spreadsheets, moved into Word docs, emailed back and forth. Even just making sure dollar signs and commas are consistent eats up 20% of an accountant’s time. Legacy tools like Workiva exist but accountants still hate them.

InScope automates the manual stuff, verifying math, formatting, all the tedious checks that slow everything down. They’re not fully auto-generating income statements and balance sheets yet (accountants are risk-averse, shocker) but that’s the end goal. Founded by two former Flexport accountants who kept running into the same problems no matter where they worked. Customer base grew 5x in the last 12 months, with top-15 firms like CohnReznick already on board.

Hiring: Full-stack Engineer, Full-stack AI Engineer, Founding Marketing Manager, Customer Success Lead, Founding Account Executive, Founding Product Designer - San Francisco

🇺🇸 Pepper

💵 Raised: $50M Series C. Investors include Lead Edge Capital (lead), ICONIQ, Index Ventures, Greylock, Harmony Partners, and Interplay.

One-liner: End-to-end tech platform for independent food distributors.

Why it’s a fave: Independent distributors handle over two-thirds of food distribution in North America. That’s a $1.4 trillion market. And most of it still runs on phone calls, faxes, and duct-taped legacy systems. The big guys (Sysco, US Foods) have the tech budgets to modernize. The independents don’t, so they’re stuck.

Pepper gives them a unified platform covering ordering, sales, accounts receivable, payments, the whole stack. They’ve built 70+ ERP integrations so distributors can plug in without ripping out what they already have. 500+ distributors on the platform now, roughly $30B in gross merchandise volume flowing through it. They’re rolling out AI agents for ordering and growth, and building an ad platform where food manufacturers can market directly to distributors.

Hiring: FP&A Manager, Finance Operations Analyst - Remote (US)

🇺🇸 Freeform

💵 Raised - $67M Series B. Investors include Apandion, AE Ventures, Founders Fund, Linse Capital, Nvidia’s NVentures, Threshold Ventures, and Two Sigma Ventures.

One-liner: AI-native 3D metal printing for mass manufacturing.

Why it’s a fave: Industrial metal 3D printing has been around for a while but the machines are expensive, slow (require extensive calibration), and not built for scale. CEO Erik Palitsch figured this out firsthand building rocket engines at SpaceX. Freeform's current system (called GoldenEye) uses 18 lasers to fuse metal powders into precision parts. The next version (Skyfall) will use hundreds of lasers to pump out thousands of kilograms of parts per day.

The wild part is they’re running H200 GPU clusters on-site doing real-time physics simulations of the printing process. They claim to have more data on metal-printing physics than anyone in the world, and they’re using it to improve quality and speed constantly. The category is heating up (Hadrian just hit a $1.6B valuation, VulcanForms and Divergent have raised hundreds of millions) but Freeform’s bet is that software-first, AI-native manufacturing wins over traditional automation.

Hiring: Manufacturing Engineer (New Grad Summer 2026), Software Engineer, C++ (New Grad Summer 2026), Electrical Engineer (New Grad Summer 2026), Mechanical Engineer (New Grad Summer 2026), Process Engineering Internship/Co-Op (Summer/Fall 2026) - Los Angeles

📖 Open Tabs (Stuff We’re Reading)

• New 15% global tariff replaces struck-down emergency duties: SCOTUS ruled Trump's emergency tariffs (up to 50%) were illegal under IEEPA. Hours later, he invoked Section 122 of the Trade Act of 1974, a provision no president has ever used, to slap a 15% blanket tariff on most imports. It caps at 15%, expires in 150 days unless Congress extends it, and exempts goods under the US-Mexico-Canada deal. Sector-specific tariffs on steel (50%), aluminum (50%), autos (25%) stay in place. Deals struck with 42 countries including the EU and Japan are now in legal limbo. China remains at 30% from an earlier truce. Trump says more tariffs are coming via Section 301.

  

• Hegseth summons Anthropic CEO to Pentagon over AI safeguards: Hegseth called Amodei to the Pentagon for a meeting officials say isn't friendly. Claude is the only AI in military classified systems. The Pentagon wants unrestricted access; Anthropic says no to mass surveillance and autonomous weapons. Pentagon is threatening to declare Anthropic a "supply chain risk," which would void contracts and block vendors from using Claude. Feud escalated after Claude was used in the Maduro raid in January.

• OpenAI planning $600B compute spend through 2030, IPO could hit $1T: 2025 revenue came in at $13B, beating its $10B target, while spending $8B (under its $9B budget). Now raising $100B+ at an ~$830B valuation, with Nvidia alone putting in $30B. Expects $280B+ cumulative revenue by 2030, split evenly between consumer and enterprise. The catch: inference costs quadrupled in 2025, dragging adjusted gross margin from 40% to 33%. Altman previously committed to $1.4T to build 30 gigawatts of compute capacity.

  

  

  

• Peak XV raises $1.3B: The former Sequoia India arm raised $1.3B across new India and APAC funds, with most earmarked for India over the next 2-3 years. The firm now manages $10B+ and has 450+ portfolio companies. AI, fintech, and consumer are the primary targets, with 80+ AI investments already made. This drops as India’s AI moment accelerates: General Catalyst just committed $5B to the country over five years, and OpenAI, Anthropic, and Google all showed up at New Delhi’s AI Impact Summit. The firm has returned $7B+ to LPs since inception with 35 portfolio IPOs.

• China’s BCI industry is moving from lab to market: While Neuralink grabs headlines, China has quietly completed 50+ clinical trials for implantable BCIs and just finished the world’s second fully implanted wireless BCI trial. Seven government agencies released a national BCI roadmap targeting a full supply chain by 2030, backed by an 11.6B yuan ($165M) brain science fund. Provinces are already setting medical pricing for BCI, fast-tracking insurance coverage. The market hit ~$530M in 2025 with projections of $16B+ by 2040. Key startups: NeuroXess, BrainCo (quietly filing for HK IPO after raising $287M), StairMed ($48M Series B), and ultrasound BCI startup Gestala launching its first product Q3 this year.

• Anthropic launches Claude Code Security: New Claude Code feature that scans full codebases for vulnerabilities and suggests patches. Goes beyond pattern-matching by reasoning about code like a human researcher. Found 500+ bugs in production open-source repos that went undetected for decades. Available in research preview for Enterprise/Team customers, free for open-source maintainers.

💼 Who’s Hiring

• Clarion: Software Engineer (Full-stack), Marketing Lead, Founding Product Designer, Growth Associate - NYC

• Conveo: Social Media Manager, Enterprise Account Executive, Customer Marketing Lead - New York, Antwerp, London

• Playground: Growth Marketing Intern (Intern-to-Hire Track), BizOps Lead, Sales Development Representative, Senior Software Engineer, Frontend - NYC, Denver

• Hadrius: BDR, Founding Operator - New York

• Fractional AI: AI Engineer, Software Engineer, Marketing and Developer Relations - San Francisco, New York; Founding Engineer - Dubai

• Bessemer Venture Partners: Associate or Senior Associate - AI - San Francisco

• SoftBank: Investment Associate - Menlo Park

• Harmonic: Growth Lead, Solutions Engineer, New Grad (Class of 2026), Growth Analyst, New Grad (Class of 2026), Full Stack Product Engineer (AI), Search - NYC

• Numeric: Content Marketing Manager, Founding Business Operations and Strategy, Software Engineer, Frontend, Software Engineer, Product - New York, San Francisco, Vancouver; Business Development Representative - London

If you haven’t yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

See you Wednesday, Ryan

Sponsorships: We are now accepting sponsors for Q1 ‘26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Welcome to another edition of Why Join. I keep thinking about this thing. The people who win never look like they're trying that hard. And the ones trying the hardest keep tripping over themselves. Weird how that works. Anyway lets jump in.

In partnership with Anything

You know that app idea you’ve been sitting on? Anything lets you build it by literally just... describing it. Tell it what you want, and it spins up a full-stack app (mobile, web, or both) with a real backend, database, payments, the works. No code. Over 1M people are already building with it, and some of them have shipped actual App Store apps without writing a single line.

We got WHY JOIN readers 30,000 free credits to try it out.

👉 Claim your 30k free credits

💰 who just raised

🇺🇸 Meridian

💵 Raised - $17M seed at $100M post-money valuation, led by Andreessen Horowitz and The General Partnership

One-liner: AI workspace for financial modeling - like Cursor but for spreadsheets.

Financial modeling is painfully manual. Analysts spend hours building valuation models, DCFs, all that. And the thing is, unlike software engineering where ten different implementations are fine, in finance ten analysts modeling the same company should get basically the same workbook. That’s a huge tension with AI, because LLMs are inherently non-deterministic. You ask the same question ten times, you get ten different answers. Existing tools (like Shortcut AI) just bolted agents onto Excel, which... kind of limits what you can do. You're still stuck in Excel's world. The gap between "AI flexibility" and "auditability" has kept most financial teams from trusting AI tools.

Meridian built their own IDE-style workspace from scratch. That lets them integrate data sources natively and layer conventional tooling on top of the agentic stuff to keep outputs auditable. You can see exactly where every assumption comes from, how the logic flows, removing what they call the "doubt layer" around LLM outputs.

🇺🇸 Didero

💵 Raised - $30M Series A, co-led by Chemistry and Headline

One-liner: AI autopilot for manufacturing procurement - sourcing suppliers, negotiating, tracking orders, handling payments, all the way through.

If you're a manufacturer buying raw materials from suppliers across, say, 15 countries, that whole process is just... chaos. It's emails, WeChat threads, phone calls, purchase orders, packing lists. People literally spend their entire day chasing suppliers and copy-pasting updates into ERPs. The existing AI procurement tools (Zip, Levelpath, SAP Ariba, etc.) mostly handle corporate purchasing, like buying software licenses or office supplies. Not the same thing. And the smaller players that do touch supply chain only really work for small brands and don't cover the full quote-to-payment flow.

Didero plugs into your existing ERP as an agentic layer and basically reads all that messy communication, the emails, messages, documents, then does the updates and tasks automatically. In Spencer's (CEO and co-founder) words, “the goal is to go from ‘I need a good’ to payment without having to lift a finger.”

Hiring: Senior Customer Success Manager, Growth Marketing Lead, Senior Product Manager - NYC/Remote

🗞️ the rundown/ 📖 open tabs (stuff we can't stop reading)

• Anthropic raises $30B at a $380B valuation. Series G led by GIC and Coatue. Co-leads include D.E. Shaw Ventures, Founders Fund, ICONIQ, MGX. Plus basically every name you’ve heard of: Sequoia, Lightspeed, Fidelity, BlackRock, Blackstone, Goldman, JPMorgan, Jane Street, Temasek, you get the idea. The numbers are wild. $14B run-rate revenue, 10x annual growth for three consecutive years. 500+ customers spending $1M+ annually (that was a dozen two years ago). Eight of the Fortune 10 are Claude customers. Claude Code hit $2.5B run-rate on its own and has doubled since January. Anthropic says 4% of all public GitHub commits are now authored by Claude Code. First dollar of revenue was less than three years ago.

  

• OpenAI drops Codex-Spark, 15x faster coding on Cerebras chips. GPT-5.3-Codex-Spark is a smaller, speed-first version of Codex built for real-time back-and-forth coding rather than the slow batch-style agent workflow. 15x faster code gen, 80% less roundtrip latency, 50% faster time-to-first-token. Runs on Cerebras’ Wafer Scale Engine 3, making it the first real output from their January partnership. The tradeoff: it’s less capable than full GPT-5.3-Codex on benchmarks and OpenAI says it doesn’t hit their cybersecurity capability threshold. So faster, but not as smart. Pro tier only ($200/mo) for now. OpenAI says they’re working toward blending both modes so you don’t have to choose. Not there yet.

  

• Crypto flows to suspected human trafficking up 85% YoY. Chainalysis released part of its 2026 Crypto Crime Report and it’s grim. Hundreds of millions flowing to suspected trafficking ops, mostly tied to Southeast Asian scam compounds and Telegram-based laundering networks. Telegram “escort” services are highly professionalized, nearly half of transactions over $10K. CSAM operations shifted to subscription models under $100/month and are increasingly overlapping with extremist communities targeting minors. One positive: blockchain transparency actually gives law enforcement way better tracking than cash. German authorities shut down “KidFlix” and US arrests of CSAM buyers went up in 2025.

• TikTok rolls out Local Feed in the US. New location-based feed shows nearby restaurants, events, shopping, local news. Opt-in, 18+ only, location sharing off by default. Timing’s not great though. It showed up right after updated terms of service said TikTok would start collecting precise location data under the new US joint venture. TikTok is framing this around small business value (7.5M businesses on the platform, 28M jobs supported). Basically the same argument Meta makes when regulators come knocking. Already live in UK, France, Italy, Germany since December.

• Two more xAI co-founders out the door. Tony Wu and Jimmy Ba resigned, which means Musk’s AI company has lost half its 12 co-founders in under three years. The FT says Ba left after internal friction over model performance pressure as xAI tries to catch up to OpenAI and Anthropic. Comes right after SpaceX said it would buy xAI to create a $1.25T combined company planning to IPO this year. Oh and at an all-hands, Musk floated building a factory on the moon to make satellites for AI. No timeline, no roadmap. Two former SpaceX execs told the NYT the moon was never really a priority there. Lot happening at xAI right now, not all of it reassuring.

• Zhipu AI launches GLM-5. The Chinese AI company (also known as Z.ai) hit #1 on Artificial Analysis’s Intelligence Index for open-weights models, scoring 50. First time any open model has crossed that mark. 744B total parameters, 40B active, trained on 28.5T tokens, uses DeepSeek’s Sparse Attention, MIT licensed. Ranks third overall on agentic tasks behind Claude Opus 4.6 and GPT-5.2. Cut hallucination rate 56% by basically just saying “I don’t know” more often. Can run on Chinese-made chips from Huawei and others, notable since Washington blacklisted Zhipu last year. Stock jumped 29% in Hong Kong. Available via APIs around $1/$3.2 per million tokens. Part of a bigger wave of Chinese model drops ahead of Lunar New Year.

  

• Mistral investing €1.2B in Swedish data centers. CEO Arthur Mensch announced Mistral’s first data centres outside France at the Techarena conference in Sweden. Partnering with EcoDataCenter, chose Sweden for clean energy. Mistral is valued around €11.7B, claims 20x growth over the past year, targeting €1B+ revenue in 2026. Mensch pushed hard on Europe needing to act as one market to compete with the US.

💼 come work here

• Handshake: Software Engineer - Intern - San Francisco

• Hook: Sales Development Representative (Graduate) - London

• Thera: Sales Development Representative (SDR), Founding Engineer - NYC

• The Compression Company: ML Compression Engineer, Deployment Engineer, Forward-Deployed Engineer, Earth Observation - US

• Reflow: Customer Success Manager, GTM Manager, Product Manager - Remote - US; Software Engineer, AI Engineer - Remote - Anywhere in the world

• Seamflow: Founder’s Associate, Founder’s Associate - Intern - London

• Stanhope AI: Chief of Staff, AI Developer - Robotics & Embodied Active Inference Systems - London

• Take2 AI: SDR, Account Executive, Head of Growth, Senior Software Engineer, Fullstack - New York

• Anterior: Member of Technical Staff - New York

• Ando: Fullstack AI Engineer, Chief of Staff - San Francisco

• Ever: Product Designer, Software Engineer, Head of Brand & Marketing, Head of People - San Francisco

If you haven't yet, please subscribe, like, leave a comment, and share it! It helps us continue to bring you the most interesting news + nuance in tech and business every week.

See you Monday, Ryan

Sponsorships We are now accepting sponsors for Q1 '26. If you are interested in reaching my audience of founders, investors, and tech executives, send me an email at chief@whyjoin.xyz.

Ship Videos Without a Video Team

Launch coming up, but no budget for video pros? With Descript, you can create polished, on-brand videos fast. Here’s how founders and marketing teams use it:

✅ Record in Descript or upload a file—transcription is automatic
✅ Edit the transcript to edit the video, just like a doc
✅ Add customizable layouts and transitions with a click
✅ Remove filler words, fix audio, and add captions instantly
✅ Generate custom images and B-roll to keep it interesting

Whether it’s launch videos, product demos, or thought leadership clips—Descript helps you make all the video you need to reach more customers.

Get a video out today

Welcome to another issue of Why Join.

Sponsored by Delve

Don’t Let SOC 2 Kill Your Startup

A startup was on the rise—steady growth, happy customers, investors circling. Then a big prospect asked for SOC 2.

They signed up with one of the big platforms. Suddenly: endless tasks, confusing checklists, engineers pulled off product. Deals stalled. Competitors moved in.

Those competitors? They used Delve.

15 hours later, they were audit-ready and closed the deal.

The first startup never recovered. One by one, their deals slipped away. Keys handed over. Game over.

Don’t be that startup.

Delve automates compliance—SOC 2, ISO 27001, HIPAA, GDPR, PCI-DSS and more—so you’re ready in days, not months.

Today, compliance is done in Delve.

Book a demo, use WHYJOIN1KOFF for 1K off.

P.S. Want to sponsor this newsletter? Simply reply to this email or book one here.

Latest in startups

🇺🇸 Zingage

Funding - Seed, $12.5M; Investors - Bessemer Venture Partners (lead), TQ Ventures, South Park Commons, WndrCo

Zingage is an AI-powered care delivery platform for home-based healthcare providers. It acts as an “AI Secretary” that automates scheduling, staffing, and compliance, helping agencies manage 24/7 care logistics with reliability and scale. The platform is built around two products: Operator, which manages everything from patient intake to billing, and Perform, a caregiver engagement app that improves retention through gamified incentives.

The home care system is under pressure. Aging populations, rising costs, and patient preference are driving a $500 billion shift toward care delivered at home, yet most agencies still run on phone calls and spreadsheets. Nearly 80% of professional caregivers leave within their first 100 days, leaving schedulers scrambling to fill visits with tools built for billing, not logistics.

Zingage replaces that manual coordination with AI trained on real operational data. Operator learns from staffing patterns to predict gaps, optimize assignments, and communicate directly with caregivers and patients through voice AI. Shifts are filled before they fail, documentation is completed automatically for payers, and managers can handle twice the patient volume without burnout.

Hiring: Founding Backend Engineer, Head of Talent Engineering, Product Operations, Software Engineer (Intern or Co-Op), Software Engineer (New Grads) - New York

🇺🇸 Quilter

Funding - Series B, $25M; Investors - Index Ventures (lead)

Quilter is building a physics-driven AI engine for printed circuit board (PCB) layout. After engineers design a circuit schematic, they still spend weeks manually routing thousands of electrical connections, placing components, and validating that every trace meets electrical, thermal, and manufacturing constraints. The process involves constant back-and-forth between design and fabrication tools, slowing down hardware development.

Quilter automates that entire stage. Rather than learning from human layouts, its platform is trained on the first principles of electronics—electromagnetic interactions, signal integrity, thermal dissipation, and manufacturability. Using reinforcement learning, it generates and evaluates millions of potential layouts, optimizing each for performance, reliability, and cost.

The output is a fully routed, DRC-clean board that meets all design rules and is ready for fabrication, often produced in hours instead of weeks. Engineers upload their schematics and constraints—board outline, stack-up, critical nets, keep-outs—and receive fabrication-ready files that can be opened directly in tools like Altium, KiCAD, or Cadence.

In short, Quilter aims to do for circuit boards what a compiler does for software: take high-level intent and automatically produce a manufacturable design.

Hiring: Senior ML Ops Engineer, Senior Fullstack Engineer, Platform Engineer (Senior, Staff, or Principal level), Senior/Staff Software Engineer (C++) - Remote (US)

🇧🇲 Meanwhile

Funding - $82M; Investors - Bain Capital, Haun Ventures, Apollo, Northwestern Mutual Future Ventures, Pantera Capital, Stillmark

Meanwhile is a fully regulated life insurance company that runs entirely on Bitcoin. Licensed by the Bermuda Monetary Authority, it issues policies, collects premiums, and pays death benefits all in BTC—no conversions to fiat. The company operates like a traditional insurer but swaps bonds for Bitcoin: it keeps a portion of reserves in cold storage and lends the rest to large, regulated institutions through secured, overcollateralized BTC loans. Those loans earn predictable returns that fund policy growth and future claims, allowing Meanwhile to function as a solvency-backed insurer with a Bitcoin balance sheet.

Bitcoin holders finally have real financial infrastructure. Selling BTC to fund insurance or inheritance triggers taxes and defeats the store-of-value thesis. Meanwhile lets users save, borrow, and plan estates directly in Bitcoin—keeping exposure while earning yield through a regulated, audited framework.

🇺🇸 Peer AI

Funding - $12.1M; Investors - Flare Capital Partners (co-lead), SignalFire (co-lead), Greycroft, Atria, Alumni Ventures, Gaingels, Mana Ventures

Peer AI is developing an agentic AI platform for life sciences that accelerates regulatory documentation across clinical, pre-clinical, and manufacturing phases. It provides a unified workspace for authoring, managing, and validating regulated documents, with human oversight built directly into the workflow.

Regulatory writing remains one of drug development’s biggest bottlenecks. Peer AI cites that new drugs often require over 200,000 pages across 1,500+ regulated documents. Around one-third of FDA submissions contain quality issues, and roughly three-quarters of applications face rejection or delay, often extending review cycles by over a year.

Peer AI uses document-specific agents to draft, propagate data, edit, and perform quality checks. Its Intelligent Template Builder auto-generates document frameworks aligned with regulatory standards, while human medical writers verify outputs at key control points to ensure compliance and traceability. Reported results show drafting speeds 55-94% faster, with clinical study reports reduced from 40 to 17 days and protocol development from 6-8 weeks to one week.

Hiring: Account Executive - Emerging & Mid-Sized Biotechs, Regulatory Medical Writer (Contract), Senior AI Backend Engineer, Senior AI  Engineer - Agentic Systems  &  LLM  Evaluations, Senior Full Stack Engineer - Generative AI & Product Engineering - Bay Area/Remote

Top news

• Anysphere, maker of Cursor, is in talks over new funding at a $30B valuation: The AI coding startup hit $500M in ARR and raised $900M in June at a $9.9B valuation from Thrive, Accel, and Andreessen Horowitz.

• China tightens rare earth exports (again): Beijing added five more rare earth elements to its export control list, bringing the total to 12. Foreign producers now need licenses to export products containing Chinese-origin minerals or tech. Defense users are banned, while semiconductor use will be reviewed case-by-case.

• Samsung’s AI lab in Montreal trained a 7M-parameter model called Tiny Recursive Model (TRM) that beat Gemini 2.5 Pro, o3-mini-high, and DeepSeek-R1 on the ARC-AGI benchmark. TRM hit 45% accuracy on ARC-AGI 1 and 7.8% on ARC-AGI 2, trained in just two days on four H100 GPUs for under $500. Instead of scaling parameters, TRM refines answers in loops. The study, “Less is More,” argues small recursive models can reason better on focused tasks than large LLMs.

• OpenAI expands ChatGPT Go: The $5 plan is now live in 16 new Asian countries including Malaysia, Thailand, Vietnam, Pakistan, and the Philippines (with local payment options). It offers higher limits, more memory, and faster responses. After launches in India and Indonesia, paid users in India have doubled and SEA usage is up 4x.

• n8n $180M Series C: Berlin-based AI automation startup n8n raised $180M at a $2.5B valuation led by Accel with Sequoia, Visionaries Club, and Nvidia’s NVentures joining. The round follows its $60M Series B at $350M (Mar ’25). n8n automates repetitive tasks, now has 230K users and $40M ARR.

• xAI $20B raise: Elon Musk’s AI startup xAI is expanding its ongoing funding round to $20B, backed by Nvidia, Apollo Global, Valor Capital, and others. The deal combines $7.5B in equity and up to $12.5B in debt through a special purpose vehicle that will buy Nvidia chips and lease them to xAI for its Colossus 2 data center in Memphis. Nvidia is investing about $2B directly, part of a broader push to accelerate AI infrastructure spending. xAI has raised $10B this year and burns about $1B a month.

• AI insurance is pricey: OpenAI and Anthropic are struggling to secure full coverage for AI-related risks as insurers hesitate over the scale of potential losses. OpenAI’s broker, Aon, reportedly arranged up to $300M in coverage (some say less), far short of what’s needed for massive copyright suits or broader “systemic” AI failures. With insurers unable to price existential risk, OpenAI may set up a captive insurance vehicle using investor funds to insure itself.

• Polymarket $2B deal: NYSE parent ICE is investing up to $2B in Polymarket, valuing it at $8B and making 27-year-old founder Shayne Coplan the youngest self-made billionaire. It follows Polymarket’s $112M acquisition of QCX, a CFTC-licensed exchange, paving its US return. The platform still awaits regulatory approval; globally, it holds 31% market share behind Kalshi’s 67%.

• Gemini 2.5 Computer Use: Google launched a new Gemini model that can use a web browser like a person, clicking, typing, and scrolling to fetch data not available through APIs. It can fill forms, test interfaces, and handle basic on-screen tasks. Unlike ChatGPT Agent or Anthropic’s tools, Gemini’s system runs only inside a browser, not across an OS, and supports 13 actions so far. Developers can try it through Google AI Studio, Vertex AI, or via a live demo on Browserbase.

• Tesla unveils cheaper EVs: Tesla launched new Model 3 and Y trims priced at $37K and $40K—about $5K less—as federal tax credits expire. The stripped versions have cloth interiors and fewer features but retain ~320-mile range, aiming to compete with Hyundai, GM, and Nissan’s sub-$35K EVs.

Open roles

• Antler: Associate - San Francisco

• Andreessen Horowitz: Research Intern, Crypto - New York

• Bandana: Content Creator/Producer, Social Media Intern, Animation/Illustration Intern, Business Development Intern, Business Development Representative (New Grad) - New York

• Cogent: Fullstack Engineer, AI/ML Engineer, Backend Engineer - San Francisco/New York

• Paramark: Integrated Marketing Manager, Marketing & Content Manager, Marketing Program Manager - San Francisco

• Resolve AI: Backend Engineer, Frontend Engineer, Growth Marketing Lead, Marketing Operations Lead, Enterprise Account Executive, Social Media Lead, Senior Brand Designer, Sales Development Representative, Recruiting Coordinator, Technical Recruiter - San Francisco

• Harmonic: Software Engineer Intern (Summer ‘26), Product Engineer, Discovery, Software Engineer (Fullstack), Growth Analyst, (New Grad), Account Executive - New York

• First Resonance: VP of Engineering, Customer Success Manager, Senior Technical Product Manager, Senior AI Engineer - Los Angeles

• Fold: Lead Mobile QA Engineer (iOS & Android), Communications Manager - Remote (US)

• Dune Security: Solutions Engineer, Senior Fullstack Engineer - New York

That’s all for today. Add us to your contacts or hit reply so we stay out of spam. Not your thing? Unsubscribe anytime below.

See you Saturday,
Chief

Welcome to another issue of Why Join.

Sponsored by HubSpot

GTM Performance Lab: Live Testing Unified Campaigns October 9, 2025 | 2PM EST

See what really happens when siloed vs. unified GTM campaigns go head-to-head.

Join HubSpot’s VP of Data and experts from Salted Stone for a live controlled experiment that exposes the hidden costs of fragmentation and shows how a unified approach changes the outcome.

In real time, you’ll watch GTM leaders plan, run, and report campaigns using HubSpot’s Smart CRM, Marketing Hub, and Marketing Studio in one connected system of record. Then compare performance side by side, from faster time-to-launch to stronger conversion rates.

This session is built for marketing leaders and GTM operators who are tired of disconnected tools and ready to prove ROI with confidence.

Register now

Latest in startups

🇺🇸 Hipp Health

Funding - Seed, $6.2M; Investors - RTP Global (lead), Swift Ventures, Rackhouse Venture Capital, Difference Partners

Hipp Health is an AI-powered all-in-one platform for ABA (Applied Behavior Analysis) practices that automates clinical documentation, data collection, scheduling, and billing, cutting admin time to a fraction.

About 1 in 31 children (3.2%) in the U.S. has been identified with Autism Spectrum Disorder, according to the CDC. Yet the behavioral health system is stretched thin, with more than half the country living in a Mental Health Professional Shortage Area and many regions lacking any providers. Clinicians spend up to half their work hours on intake, scheduling, progress notes, and claims validation instead of care.

Most practices still rely on legacy systems like CentralReach and Rethink, built more than 15 years ago and unable to keep up with modern needs. The result is burnout. Registered Behavior Technicians (RBTs) and Board-Certified Behavior Analysts (BCBAs) often stay late or work weekends to meet Medicaid and insurance requirements.

Hipp uses AI to write clinical notes in seconds. Clinicians can type, record, or photograph their notes on the go, and Hipp turns them into clean, compliant reports. The platform is built on privacy-safe retrieval-augmented generation (RAG), allowing providers to query patient data securely without exposing information to commercial language models.

Hiring: Billing Specialist, Implementation Engineer, Customer Support Representative - Remote; Software Engineer, Product Designer - San Francisco

🇺🇸 FurtherAI

Funding - Series A, $25M; Investors - Andreessen Horowitz (lead), Nexus Venture Partners, Y Combinator

FurtherAI is building domain-specific AI for commercial insurance. This isn’t a horizontal copilot; it’s trained on insurance language and processes, designed to read policy documents, extract fields, compare terms, run eligibility checks, and push data into internal systems and portals.

Insurance is one of the largest industries in the world, with more than $7 trillion, but its processes are outdated and inefficient. Manual work dominates underwriting, claims, and compliance across disconnected legacy systems. A single mid-sized MGA can generate thousands of quotes each month, each requiring 10-20 minutes of manual processing. Multiply that across carriers and brokers, and the inefficiency compounds.

The labor picture makes the problem worse. More than 3M insurance professionals already spend most of their time on repetitive tasks, and the industry is projected to lose over 400,000 workers by 2026. That shortage means fewer people are available to handle growing volumes of policies, renewals, and claims, while compliance demands continue to rise.

This is the kind of repetitive, rules-based work large language models now handle with precision. Traditional RPA tried and failed—it broke on unstructured documents and missed regulatory nuance. Fine-tuned language models trained on insurance data are finally strong enough to manage these tasks without human babysitting.

Hiring: Enterprise Account Executive, Customer Success Manager - San Francisco/Remote (US); Software Engineer - Backend/Fullstack, Software Engineer - Frontend - India

🇺🇸 Moonlake

Funding - Seed, $28M; Investors - Threshold Ventures (formerly DFJ), AIX Ventures, NVentures (NVIDIA’s venture capital arm), Steve Chen, Jeff Dean, Naval Ravikant, Ian Goodfellow, Guillermo Rauch, Jared Leto

Moonlake AI is building an AI reasoning engine that lets anyone “vibe code” immersive interactive worlds in minutes.

Building games and simulations has always been a big-studio process, requiring teams of artists, engineers, and months of pipeline just to test an idea. A single change can ripple through physics, logic, and assets, breaking the build. For indie developers and the billion-plus individual creators out there, that makes iteration nearly impossible.

Moonlake’s reasoning models translate natural language into fully working interactive environments with physics, agents, and programmable rules. You describe what you want — “a medieval town square with vendors trading” or “a warehouse with three robot arms testing grasp policies” — and the model compiles it into a playable scene.

Edits happen through conversation in real time. You can adjust gravity, modify behavior, or drop in new agents without restarting a build. No 3D modeling skills or programming expertise required.

The goal is to compress the prototyping loop from months to minutes while preserving expert-level control. Beyond games, the same system powers robotics simulations, 3D environments for film, and embodied-AI training.

🇺🇸 Summation

Funding - $35M; Investors - Benchmark, Kleiner Perkins

Summation is building a decision-grade AI platform for executives who can’t wait weeks for answers. It connects to ERP, CRM, EPM, and warehouse systems and runs AI-scale distributed computations automatically. Behind the scenes, AI agents explore and validate thousands of business questions in parallel, surfacing verified, audit-ready insights for high-stakes decisions.

Executives today walk into board meetings with dashboards and spreadsheets, only to face new questions: why did this change, what’s driving it, and what should we do next. Analysts can only cover narrow slices, and traditional warehouses can’t handle the scale or concurrency that AI-driven analysis demands. By the time teams reconcile data across systems, the window for action has passed.

Summation’s engine is designed to break that bottleneck. It performs up to 100 times more drill-downs than legacy analytics stacks while maintaining full data lineage, permissions, and audit trails. Its three core products are:

• AI Calculation Engine, for distributed computation built for agentic workloads.

• Verified Tasks, which automate reconciliations and reporting with audit-grade accuracy (SOC 2 Type II).

• Deep Dive, which orchestrates reasoning agents to test and validate thousands of strategic hypotheses in real time.

Every number ties back to its source, with traceability leaders can trust. The platform integrates seamlessly with existing enterprise systems, requires no heavy migration, and can go live in weeks.

Hiring: Founding Marketing Manager, Sr. Product Designer, Senior Platform Engineer, Senior Product Engineer, Senior Design Engineer - Bellevue, WA

In partnership with

The $2 Trillion Asset Class You’re Missing

While retail investors chase volatile tech stocks, institutions like BlackRock and KKR are quietly allocating billions to private credit—a market that’s grown 19x since 2006 and now approaches $2 trillion globally.

Why? Private credit offers what traditional portfolios increasingly lack: consistent yield potential (Percent’s marketplace delivered 14.9% average net returns in 2024), lower correlation to public markets, and shorter investment horizons averaging just 9 months.

With Percent, accredited investors can now access the same institutional-quality deals that were once exclusive to the ultra-wealthy. Start with as little as $500 and get up to $500 on your first investment. Invest like the smart money.

Access Institutional-Grade Private Credit

Top news

• OpenAI Dev Day 2025: OpenAI is bringing apps directly into ChatGPT, starting with Booking.com, Canva, Spotify, and Zillow. Users can interact with these apps using natural language, while developers can now build their own through the new Apps SDK, built on MCP for full in-chat UI and video playback. The company also launched AgentKit, an “Agent Builder” for creating AI agents in a drag-and-drop Canva-style interface. Codex is officially out of preview, GPT-5 Pro and Sora 2 APIs are in testing, and Altman says ChatGPT now reaches 800M weekly users and handles more than 6B tokens a minute.

  

• OpenAI x AMD deal: OpenAI signed a massive multi-year deal with AMD to secure hundreds of thousands of AI chips—roughly six gigawatts of compute—starting in 2026. The deal includes a warrant letting OpenAI buy up to 160 million AMD shares (around 10% of the company) for 1 cent each, tied to chip shipment and stock price milestones that go up to $600 a share. AMD expects the partnership to generate more than $100B in revenue over four years as it builds a one-gigawatt facility powered by its upcoming MI450 chips.

• Why Cerebras pulled its IPO: Cerebras withdrew its IPO filing days after raising $1.1B at an $8.1B valuation led by Fidelity and Atreides. CEO Andrew Feldman said the original 2024 S-1 had gone stale after a year of record revenue and major changes in the business. The company plans to refile with updated financials and strategy reflecting its shift toward AI cloud services. Feldman called it “a procedural step, not a strategic shift,” adding that Cerebras still intends to go public once the new materials are ready. Translation: the IPO’s still on, they just want to show up with a better deck.

• OpenAI x Jony Ive device: OpenAI and Jony Ive are running into hurdles on their secret AI gadget due next year. It’s a palm-sized, screenless assistant that sees and listens—meant to feel more like a “computer friend” than a voice bot. The team’s stuck on compute limits, privacy, and how to give it a believable personality. OpenAI bought Ive’s firm io for $6.5B, hired ex-Apple and Meta engineers, and is working with Luxshare to bring it to life.

• Supabase $100M Series E: Supabase raised $100M at a $5B valuation led by Accel and Peak XV, with Figma Ventures in the mix. The Postgres startup now totals $500M+ raised and plans to open part of this round to its developer community. Supabase powers 4M+ devs and 100k companies, including AI players like Lovable, Bolt, and Cursor. The new money fuels “Multigres,” its enterprise-scale Postgres platform led by Vitess co-creator Sugu Sougoumarane.

Life moves fast. Your savings should too.

With Cash App, you can round up your spare change from everyday purchases, earn up to 4% interest1, and transfer money anytime—all without hidden fees.

Start saving today

Open roles

• Primary: Senior Associate, Vertical AI - New York

• Outlander VC: Junior Partner/Partner - San Francisco/Los Angeles/New York

• Kadence: VP Marketing - Salt Lake City, UT/San Jose, CA; Customer Support Manager - Salt Lake City; Intern Quality Assurance Engineer (Placement), Intern Software Engineer (Placement) - Belfast, UK

• Tycho.AI: ML Engineer, Software Engineer - Cambridge, MA

• Gelt: Agent Engineer - East Midlands, UK; Client Loyalty & Ecosystem Manager - New York

• Granola: Design Engineer, Product Engineer (Backend), Product Engineer (Full Stack), Product Manager (Growth), Sales Development Representative, Founding Account Executive - London

• Lindy: Fullstack Software Engineer, Executive Assistant to the CEO, Growth Marketing Lead, Senior Sales Development Representative, Account Executive - San Francisco

• Composite: Founding Engineer - San Francisco

• Cyvl: Engineering Intern (SW, ML, DS), Business Development Representative - Boston, MA

• Tabs: Software Engineer - Billing - New York

• Verkada: MBA Hire - Sales Strategy & Operations - San Mateo, CA

That’s all for today. Add us to your contacts or hit reply so we stay out of spam. Not your thing? Unsubscribe anytime below.

See you Thursday,
Chief

Another Tuesday. Another newsletter.

Quick thought. Why most startup demos fail?

Airbnb’s early prototype? PowerPoint slides. They tested the concept with basic mockups before building anything. Most startups do the opposite—build first, validate never.

The difference? They knew a prototype answers one question. Not “will this work?” but which specific question are we testing?

Three types:

• Role - Does anyone actually need this?

• Look & Feel - Will they enjoy using it?

• Implementation - Can we build it?

Pick one. Answer it. Throw it away. Repeat.

Airbnb raised $600K with their early pitch deck. Today they’re worth $77B. They didn’t build the perfect product. They prototyped fast enough to learn what mattered.

“We have a strategic plan. It’s called doing things.” - Herb Kelleher

More on prototyping →

Now, to this week’s startups...

Sponsored by HubSpot

GTM Performance Lab: Live Testing Unified Campaigns October 9, 2025 | 2PM EST

See what really happens when siloed vs. unified GTM campaigns go head-to-head.

Join HubSpot’s VP of Data and experts from Salted Stone for a live controlled experiment that exposes the hidden costs of fragmentation and shows how a unified approach changes the outcome.

In real time, you’ll watch GTM leaders plan, run, and report campaigns using HubSpot’s Smart CRM, Marketing Hub, and Marketing Studio in one connected system of record. Then compare performance side by side, from faster time-to-launch to stronger conversion rates.

This session is built for marketing leaders and GTM operators who are tired of disconnected tools and ready to prove ROI with confidence.

Register now

P.S. Want to sponsor this newsletter? Simply reply to this email or book one here.

Latest in startups

🇺🇸 🇮🇳 Emergent

Funding - Series A, $23M; Investors - Lightspeed India, Together Fund, Y Combinator, Prosus Ventures, Google chief scientist Jeff Dean, Mistral founding scientist and current Thinking Machines researcher Devendra Chaplot, former Coinbase CTO Balaji Srinivasan.

Emergent is building a “vibe coding” platform—an AI-powered system that turns natural language into production-ready software. Users describe an idea in plain English and the platform generates a full-stack application with backend, database, authentication, integrations, and deployment.

Building software has always required code, capital, or a technical co-founder. Non-technical founders and small businesses get stuck because engineering talent is expensive, no-code platforms are rigid, and most AI code tools produce half-baked outputs that need expert supervision. As a result, many ideas never move past the whiteboard.

Emergent’s answer is a conversation-to-code platform. Agents handle planning, architecture, testing, debugging, and deployment. A builder can describe a product in chat and get a working application with live preview, GitHub integration, and secure hosting. The company claims full applications are delivered in under five minutes.

Adoption has been fast. Since launching in June 2024, more than 1M people have built 1.5M apps. The company reported reaching $15M ARR within 90 days of launch, making it one among the fastest growing AI startups on record.

Hiring: Analyst, Founder’s Office, Affiliate Marketing Specialist, Growth Hacker - Bangalore; Growth Lead, Product Manager, Product Experience Specialist - Intern - San Francisco

🇺🇸 Alex

Funding - Series A, $17M; Investors - Peak XV Partners (lead), Y Combinator, Uncorrelated Ventures, Tim Sackett, Kris Fredrickson, Dalton Caldwell, CHROs at Fortune 500 companies; The $3M Seed was led by 1984 Ventures.

Alex is an AI recruiter that automates the first step of hiring: the screening interview. Instead of one-way video recordings or recruiter phone calls, Alex uses voice AI to run live two-way conversations with candidates right after they apply.

Recruiting is a volume nightmare. Companies spend over $5k per hire, hiring processes average 30 days, and most of that is just recruiters and hiring managers grinding through interviews. Meanwhile hundreds of applicants compete for one position and only a handful get interviewed.

The real issue? Human recruiters physically can’t talk to everyone. Good candidates get lost because there’s not enough time, and recruiters burn out on repetitive screening calls asking the same background check / salary / availability questions over and over.

Plus existing video interview platforms suck - they use one-way recordings with cookie-cutter questions that create a terrible candidate experience and low completion rates.

Alex conducts thousands of interviews per day with live, two-way conversations. It’s not a bot spitting canned questions - the AI asks personalized follow-up questions in real-time based on what the candidate says.

The platform handles phone screens, video interviews, background checks, salary discussions, fraud detection - basically everything a recruiter does in that first call. Then it generates detailed feedback customized to your hiring goals so recruiters can focus on pre-qualified candidates.

Hiring: Product Designer, Product Manager, Fullstack Software Engineer, Founding Product Marketer, Talent Lead, SDR, Customer Success - San Francisco

🇺🇸 Greptile

Funding - Series A, $17M; Investors - Peak XV Partners (lead), Y Combinator, Uncorrelated Ventures, Tim Sackett, Kris Fredrickson, Dalton Caldwell, CHROs at Fortune 500 companies; The $3M Seed was led by 1984 Ventures.

Greptile is an AI code reviewer that plugs into GitHub and GitLab, reviewing pull requests with full context of the codebase. Instead of syntax checks or linting, it provides in-line comments that catch bugs, flag antipatterns, and offer architecture-level suggestions.

Code reviews are essential for quality assurance, collaboration, and knowledge sharing. Yet, they’re time-consuming and manual, especially in large codebases. Human reviewers miss bugs, lack context across the entire system, and create bottlenecks that slow down development velocity. As time goes on and multiple programmers work on a codebase, it tends to get very difficult to understand. Teams need someone who has deep understanding of the code but that person doesn’t exist or is too busy.

Greptile’s approach is context-aware automation. It provides comments on PRs with click-to-accept suggestions for minor fixes, generates mermaid diagrams, delivers file-by-file breakdowns, and assigns confidence scores for every PR. Teams can also set custom rules, style guides, and context that align with their coding standards.

As you use Greptile, it infers new rules and idiosyncrasies about your team and codebase from your comments, replies, and 👍/👎 reactions . The API has two parameters: connect repositories for indexing, then add natural language queries like “How does authentication work in this codebase?”

Hiring: Software Engineer (Frontend), Software Engineer (Generalist), Software Engineer (Infrastructure), Founding Account Executive (Enterprise) - San Francisco

Top news

• Swift tests blockchain rails: Swift, the global financial messaging network, is running a pilot with 30 banks to use blockchain for cross-border transfers. The goal is to replace multi-day settlement times with near-instant, fully trackable payments.

• Trump paves way for $14B TikTok deal that puts US investors in the driver’s seat: Trump just approved a TikTok deal where American investors, including Oracle, Michael Dell, and Rupert Murdoch, will control 80% of a new US entity valued at $14 billion, with ByteDance keeping 20%. The new structure means the US company will control TikTok’s algorithm and content, technically complying with the 2024 law requiring ByteDance’s divestiture while keeping the app running.

• A new coding contender? Anthropic just rolled out Claude Sonnet 4.5, the latest in its mid-sized model series. Early tests suggest it’s strong at code generation, though some reviewers rank it a notch below OpenAI’s Codex in certain tasks. The release also comes with an upgraded API and hints of a “vibe-coding” feature in the works. Pricing lands at $3 per million input tokens and $15 per million output tokens.

• xAI pushes back in India: X, the social platform owned by xAI, is challenging the Indian government’s Sayhog portal. The system allows authorities to order takedowns of online content, which X says enables removals “based only on allegations” without court oversight, while threatening platforms with criminal penalties for non-compliance. The company says it will appeal, framing the move as a defense of free expression.

• Search APIs are heating up: Perplexity rolled out a search API this week, letting other companies embed its AI-powered search into their own products. The move shifts Perplexity from being just a consumer app to a platform play. Meanwhile, Ollama—known for making it easy to run AI models locally—introduced its own API-based search tool. They’ll be up against incumbents like Glean and Algolia, which are also fighting for a share of the enterprise and private search market.

• U.K. bailout after cyberattack: The British government is guaranteeing a £1.5 billion loan to Jaguar Land Rover after a cyberattack halted production for weeks. It’s the first state-backed rescue tied directly to a hack, underscoring how exposed major manufacturers remain to coordinated digital threats.

Open roles

• Parter: Account Executive (AE), Go-To-Market (GTM) Engineer - New York; Fullstack Software Engineer - Tel Aviv, IL

• Highlo (stealth): Founder’s associate - London

• Supanote AI: Growth Marketing Manager - Bangalore

• Trestle: Founding Fullstack Software Engineer, Founding Product Designer - New York

• Spade: Enterprise Account Executive (Banks), Senior Backend Engineer, Head of Engineering - New York

• Howie: Senior Product Designer, Senior Software Engineer (AI), Senior Software Engineer (Backend), Senior Software Engineer (Frontend) - Seattle, WA

• Neon: Fullstack Developer - New York

• Exponential: Associate Product Manager, Senior Full Stack Developer - Remote (US)

• Virio: Head of Growth, Founding Account Executive, Founding Engineer, Head of CEO Content - San Francisco

That’s all for today. Add us to your contacts or hit reply so we stay out of spam. Not your thing? Unsubscribe anytime below.

See you Thursday,
Chief

Welcome to another issue of Why Join.

Sponsored by Delve

Don’t Let SOC 2 Kill Your Startup

A startup was on the rise—steady growth, happy customers, investors circling. Then a big prospect asked for SOC 2.

They signed up with one of the big platforms. Suddenly: endless tasks, confusing checklists, engineers pulled off product. Deals stalled. Competitors moved in.

Those competitors? They used Delve.

15 hours later, they were audit-ready and closed the deal.

The first startup never recovered. One by one, their deals slipped away. Keys handed over. Game over.

Don’t be that startup.

Delve automates compliance—SOC 2, ISO 27001, HIPAA, GDPR, PCI-DSS and more—so you’re ready in days, not months.

Today, compliance is done in Delve.

Book a demo, use WHYJOIN1KOFF for 1K off.

P.S. Want to sponsor this newsletter? Simply reply to this email or book one here.

Latest in Startups

🇺🇸 Factory

Funding - Series B, $50M @ $300M valuation; Investors - NEA, Sequoia Capital, J.P. Morgan, Nvidia, Abstract Ventures, Mantis Ventures

Factory builds “Droids” - AI agents that autonomously handle full software development lifecycle tasks.

The biggest bottleneck in software development isn’t writing code—it’s everything around it. Developers spend most of their time untangling messy codebases, migrating APIs across hundreds of files, debugging 2am production incidents, refactoring legacy systems, writing brittle tests, and jumping between Jira, Slack, docs, and their IDE.

AI coding tools like GitHub Copilot help you type faster, but they don’t solve any of these problems.

Factory lets you hand off entire engineering tasks to AI agents called Droids. Not “complete this function,” but “migrate the whole codebase to API v3” or “fix the production bug and write the tests.” The Droids plug into your full stack—GitHub, Jira, Slack, monitoring—and pull context like a human engineer would. Then they execute the full workflow, end-to-end.

Hiring: AI Engineer, Frontend Engineer, Fullstack Engineer, Enterprise Account Executive - San Francisco

🇩🇰 Light

Funding - Series A, $30M; Investors - Balderton Capital (lead), Atomico, Cherry Ventures, Seedcamp, Entrée Capital

Light is an AI-native finance platform built to replace legacy ERP systems like SAP and NetSuite for fast-growing companies.

Most finance software still runs on infrastructure built in the 1980s. For hypergrowth companies expanding across countries, currencies, and entities, legacy ERPs like SAP, Oracle, and Microsoft Dynamics are too slow, too rigid, and too expensive to implement. What should take minutes takes months.

Light solves this from day one. Instead of retrofitting old tools with AI features, they rebuilt the finance stack from scratch—database to interface—to automate global payments, multi-entity accounting, expense management, and instant close. The platform matches human accuracy, catches edge-case errors, and scales without the complexity.

Hiring: Partnership Manager, Accounting, Account Executive, Growth Marketer, Event & Marketing Coordinator - London/New York; Content Creator - Stockholm, Sweden; Product Engineer, Backend Engineer, Frontend Engineer - Remote (CET)

🇺🇸 Juicebox

Funding - Series A, $30M; Investors - Sequoia Capital (lead) Coatue, NFDG, Y Combinator, Lux Capital, BOND + unannounced $6M Seed round led by NFDG

Juicebox is an LLM-native search engine for hiring. It parses résumés, portfolios, personal sites, and public profiles using natural language. Instead of matching keywords, it infers actual fit.

LLMs are turning hiring into an inference problem, not a filtering one. Traditional recruiting tools rely on keyword filters and structured fields. But top talent, especially in AI, product, and design, rarely follow a templated résumé format. Their best work lives in GitHub, Notion, blogs, and side projects.

Legacy tools like LinkedIn Recruiter, Greenhouse, and traditional ATSs were built for outbound workflows and corporate HR teams. They weren’t designed for lean startups hiring globally at speed.

Juicebox surfaces inferred skills, career arcs, and high-signal profiles from unstructured sources—GitHub activity, personal sites, writing, portfolios. It automates sourcing, outreach, and scheduling so even a two-person team can run full-cycle recruiting without ever opening LinkedIn.

Hiring: Growth & Community, Founding Recruiter, Sales Development Representative, Account Executive, Founding Product Designer, Software Engineer, Senior Software Engineer - San Francisco

🇺🇸 Anything

Funding - Series A, $11M @ $100M valuation; Investors - Footwork (lead), Uncork, Bessemer, M13

Anything is an AI-powered app builder that goes beyond prototyping to help non-technical users actually ship production-ready mobile and web apps.

Vibe coding has exploded (Lovable hit $100 million in annual recurring revenue (ARR) just eight months after launch, while Replit’s ARR soared from $2.8 million to $150 million in less than a year). Everyone is building with natural language prompts. But most tools stop at the prototype. Shipping a real product still requires duct-taping Supabase, Stripe, and Cloudflare together, or knowing how to deploy everything yourself.

Anything solves that. It’s end to end: backend infrastructure, deployment, monetization, and even App Store-ready packaging. The team built their own infrastructure instead of relying on Supabase like the rest of the market, and it shows. Apps built with Apps built with Anything are already live—a CPR trainer, a habit tracker, a hair-style “try-on” tool. Some are already making money.

The company’s initial traction was explosive, reaching $2 million annualized run rate in just two weeks.

Hiring: AI Engineer, Content Engineer, Design Engineer, Founding Engineer, Growth (GTM Engineer), Infra Engineer - San Francisco

Together with

The go-to debit card for instant discounts

Get instant discounts at places you already shop with a Cash App Card. It’s a debit card that works where Visa is accepted, even at ATMs. Whether it’s time for groceries, a ride to meet a friend, or something you’ve been eyeing, you’ll get access to offers that turn everyday spending into real savings.

Why you’ll love it:

• Exclusive offers to save on everyday spending

• No monthly or hidden fees

• 24/7 fraud monitoring to avoid scams before they can happen

• Instant ways to lock your card if it’s ever lost or stolen

• Pay over time for eligible purchases with Cash App Afterpay1

Order Cash App Card

Top News

• EA sells for $55B: Electronic Arts going private in the largest LBO ever. Silver Lake, Saudi PIF and Jared Kushner’s Affinity Partners paying $210 per share, a 25% premium, with $20B debt from JPMorgan. At $55B the deal tops TXU’s 2007 record $45B buyout. EA, maker of Madden, FIFA (now FC) and Sims, is facing slowing growth, layoffs and a shift toward free-to-play. Sports titles still drive steady cash. Battlefield 6 launches Oct 10 with strong early buzz. Deal needs regulatory approval and is 2025’s largest M&A.

• DeepSeek drops Sparse Attention model: China-based lab released V3.2-exp with a new sparse attention system that cuts long-context API costs by around 50%. Uses a lightning indexer and fine-grained token selector to narrow what gets loaded into attention. Focus is on lowering inference costs, not training. Shows transformer efficiency gains are still possible. DeepSeek, known for its R1 RL-trained model earlier this year, is now pushing more practical cost-saving work.

• ChatGPT adds Instant Checkout: OpenAI now lets US users buy Etsy goods directly in chat, Shopify merchants like Glossier, Skims and Vuori coming soon. Users hit Buy, pay with Apple Pay, Google Pay, Stripe or card. OpenAI says listings are organic and unsponsored with merchants paying a small fee per sale. It is also open-sourcing its Agentic Commerce Protocol built with Stripe so others can integrate. It’s OpenAI’s play at agentic commerce—shifting retail discovery from Amazon/Google toward AI intermediaries.

• Spotify is finally cracking down on AI slop: It’s rolling out new rules to tackle AI slop, fake artist clones, and undisclosed AI use in music. A new metadata standard is in the works (with DDEX) to flag AI-generated content—no launch date yet, but 15 labels are on board. They’re also banning voice clones and deepfakes used without permission. A new spam filter will target tricks like short tracks and duplicate uploads. Spotify says it removed 75M spam tracks in the past year.

• Meta to charge UK users £2.99/month to remove ads: Facebook and Instagram users in the UK will soon get an ad-free option at £2.99 per month on web or £3.99 via apps. Unlike EU users, UK users will not get a less personalised ads choice. It is either pay or keep full ads. Move follows pressure from the ICO which said default targeting broke UK law.

• Checkout.com secondary at $12B: UK fintech running staff buyback giving 2k employees liquidity. Valuation marks up from $9.35B in 2023 but far below $40B peak in 2022. Founded in 2012 by Guillaume Pousaz, Checkout processes payments for eBay, Ikea, Sainsbury’s and more. On path to profit, expects $300B processed this year, hiring 300 more staff. Still trails rivals with Adyen at €43B market cap and Stripe near $106B.

• Kraken in talks at $20B: Crypto exchange set to raise $200–300M from a strategic investor, pushing val to $20B. Follows a $500M round at $15B earlier this year. Founded 2011, had raised just $27M until recently. Now bulking up ahead of IPO with Goldman and Morgan Stanley as bankers. Kraken added tokenized equity trading and is expanding to compete with Coinbase. IPO could land Q1 next year.

• Tether eyes $500B val: Stablecoin giant in talks to raise $15–20B for about 3% stake, which would peg it at half a trillion. SoftBank and Ark Investment among early names circling. USDT has $173.5B circulation vs Circle’s $74B. At $500B, co-founder and chair Giancarlo Devasini’s stake would be worth $224B, richer than Buffett.

• Meta’s next big bet is humanoids: CTO Andrew Bosworth says Zuck stood up a robotics unit earlier this year, now building Metabot. Hardware is not the bottleneck, software is. Focus on dexterous manipulation like picking up a glass. Meta plans to license its platform to other robot makers, similar to Android for phones. The Superintelligence AI lab is building a world model to animate a dexterous hand. Team is led by ex-Cruise CEO Marc Whitten with hires like MIT’s Sangbae Kim. Bosworth says data is the hard part compared to Tesla’s Optimus. Meta frames robotics spend in the same category as its AR billions.

Start building job-ready tech skills today

Learn AI skills for any industry and get hands-on experience using the latest tools in tech. WHY JOIN readers can get 15% off a year of Codecademy Pro when you use code SKILLUP15.

Join Codecademy today (Sponsored)

Open roles

• Wintermute: Graduate Sales Trader - London

• Ramp: AI Operations Intern (Summer 2026) - New York

• Cartesia: Research Intern - San Francisco

• Verkada: Software Engineering Intern, Mobile (Winter/Summer 2026) - San Mateo

• Base: Software Engineering Intern, Business Operations Intern, Factory Construction Intern - Austin

• Parahelp: Founding Design Engineer, Founding Account Executive, Founding AI Engineer - San Francisco

• Sola: Founding Recruiter, Senior Software Engineer (Fullstack), Senior Software Engineer (Backend) - New York

• Monk: Founding Fullstack Engineer, Founding Backend Engineer - New York

• Paradigm: Design Engineer, Software Engineer - San Francisco

• Nova Benefits: HR Intern, SDE Intern - Bangalore, India

• Moment: Product Engineer, Product Designer, Marketing Manager, Strategy & Ops, Enterprise Account Executive - New York

• Ambrook: Software Engineering - Intern - New York; Software Engineer, Founding Infrastructure Engineeer, Growth Marketing Lead, Partnerships Lead - New York, Denver, San Francisco, or Remote

That’s all for today. Add us to your contacts or hit reply so we stay out of spam. Not your thing? Unsubscribe anytime below.

See you Wednesday,
Chief

Welcome to this week’s Why Join.

Let’s get straight to it.

Sponsored by

AI is making scammers' lives easier

Your name, address, phone number, and financial info are traded online for as little as $5. Scammers, identity thieves, and AI-powered fraudsters can buy this data to target you.

Generative AI is making these scams more difficult to spot until it’s too late.

But here’s the catch: they can’t scam you if they can’t find you.

Incogni Unlimited eliminates the fear of your personal details being found online. The data removal service automatically removes your data from the databases that scammers rely on.

You can even use the custom removal feature to remove yourself from other databases, public databases, directories, and much more. Send the links to Incogni via your dashboard and their team will clean it up for you.

Try Incogni and get 55% off with code BYESUMMER.

P.S. Want to sponsor this newsletter? Simply reply to this email or book one here.

Latest in Startups

🇺🇸 Console

Funding - Series A, $23M; Investors - DST Global Partners (co-lead), Thrive Capital (co-lead)

Console is building the AI that actually solves your IT problems instead of just creating more tickets. When you message their Slack bot saying "I can't access Salesforce," it doesn't just log a ticket - it actually fixes your permissions, resets your password, or provisions new access. It's like having a super-competent IT person who never sleeps, never gets annoyed, and can handle 50 things at once.

Under the hood, Console runs on structured playbooks written in natural language. Engineers define workflows (“If someone says they lost their MacBook, do this…”) and Console turns that into an executable policy. It has full context on the org: users, apps, devices, permissions. Unlike brittle RPA setups, it evolves.

IT support is broken everywhere. Average wait times are days, not hours. IT teams spend 80% of their time on repetitive stuff (password resets, app access, basic troubleshooting) instead of actually improving systems. Meanwhile, companies are growing faster than they can hire IT staff. Console is betting that AI can finally handle the boring stuff, freeing humans to do actual strategic work.

Hiring: Product Designer, Head of Applied AI, Applied AI Engineer, Design Engineer, Product Engineer, Forward Deployed Engineer - San Francisco

🇺🇸 Doctronic

Funding - Series A, $20M; Investors - Lightspeed Venture Partners (lead), Union Square Ventures, Tusk Ventures, Mantis VC, Seven Stars

Doctronic is building an AI-first replacement for those panicked 2 AM Google searches about weird symptoms. Their platform combines a "multi-agent AI doctor" system with real human physicians, letting patients get instant medical triage 24/7 before escalating to actual doctors when needed. Think of it as ChatGPT for healthcare, but one that's trained on medical guidelines and doesn't hallucinate fake treatments.

Most people don't realize how broken primary care access really is. Wait times stretch past 30 days. More than 100 million Americans don't have a primary care provider. And real clinical AI adoption is still in its infancy. While administrative AI tools like Abridge and Nabla have gained traction, full-stack clinical agents are rare and heavily constrained by regulation and accuracy demands.

But here's what's different. A recent independent study tested 500 urgent care cases and benchmarked Doctronic's performance against physicians: 81 percent agreement on primary diagnosis, 99.2 percent agreement on treatment plans, and zero hallucinations. In the cases where they disagreed, Doctronic outperformed human doctors four to one. That's clinical-grade performance. For context, most LLMs in medical benchmarks still hover in the 50 to 55 percent range.

This is either the future of healthcare access or just a way to automate WebMD. The biggest question is still regulatory. FDA approval frameworks for AI diagnostic tools are evolving, and nobody knows how quickly they'll adapt.

Hiring: Product Designer, Project Manager, Customer Success Manager, Director of Marketing Strategy, Director of Paid Media, Head of Business Operations - Los Angeles, New York, or Remote (USA only)

🇮🇱 Irregular

Funding - Series A, $80M; Investors - Sequoia Capital (co-lead), Redpoint Ventures (co-lead)

Irregular builds simulation environments where AI models attack other AI models. Not just pen testing. We're talking full kill-chain scenarios: lateral movement, EDR evasion, ransomware payloads executed by one AI agent while another defends. Think of it as red teaming for a world where both the hacker and the target might be autonomous agents.

AI models can already write code and interact with systems. The obvious next step is AI that can chain together complex attack sequences like reconnaissance, privilege escalation, persistence without human oversight. Traditional cybersecurity tools assume human attackers who get tired and make mistakes. AI attackers don't have those limitations.

Their SOLVE framework is already being used by OpenAI, Anthropic, and DeepMind to stress-test frontier models. They're cited in GPT-4 and Claude system cards. Governments are asking for their help on national AI risk frameworks.

Hiring: AI Researcher, Cyber Researcher, Research Engineer, Technical Policy Researcher

🇺🇸 Blacksmith

Funding - Series A, $10M; Investors - Google Ventures (lead)

Blacksmith makes code testing and deployment way faster and cheaper. When developers push code to GitHub, it needs to be tested before going live. That process can take hours and burn through hundreds of dollars in cloud compute. Blacksmith replaces those slow, expensive servers with souped-up gaming PCs that run tests twice as fast at 75 percent lower cost. It's like swapping a Honda Civic for a Ferrari, but for CI/CD.

The AI coding boom has a dirty secret: while agents can write code a hundred times faster than humans, testing still takes forever. Teams are hitting a bottleneck where AI generates features in minutes, but pipelines take hours to validate them. That speed mismatch kills the productivity gains AI promises.

Most dev tool startups go cloud-native to scale quickly. Blacksmith did the opposite. They own the hardware. Their bare-metal setup uses gaming-grade CPUs, giving them tighter control over performance and costs. In a cloud-first world, they're betting on physical servers—and so far, it’s working.

Hiring: Product Engineer, Systems Engineer - New York; Business Operations Lead, Account Executive, Sales Development Representative - San Francisco

A message from

THIS WEEK: Join the AI ROI Conference on September 25 - Free Virtual Conference on Generating Real ROI with AI featuring Scott Galloway

On September 25, get the expert formula for turning AI spend into ROI at your oganization at Section's virtual AI ROI conference.

No gatekeeping, no pitching, no BS. Hear from top AI leaders like Scott Galloway, Brice Challamel, and May Habib and leave with a clear path to AI ROI for your company.

Claim Your Free Spot

Top News

• Nvidia to invest up to $100B in OpenAI: Signed an LOI to support infra buildout for AI workloads. Money will go toward new data centers and power capacity as Nvidia systems roll out. Investment scales with deployment.

• Neuralink plans Oct trial to read thoughts as text: Musk’s brain implant co. will launch a US trial next month to decode imagined speech in real-time. The FDA granted an investigational device exemption. Unlike current trials focused on paralysis, this one targets speech restoration—reading directly from the brain’s speech cortex. Neuralink says healthy people could get implants by 2030, aiming for mass-market use (20K/year by 2031). Tech would let users query AI models at speed of thought, per co-founder DJ Seo. Long-term goal: consumer BCI. Short-term: restore function for patients with ALS, strokes, etc. Trial timing slips from earlier Q3 target.

• Navan files to go public: The travel and expense startup (fka TripActions) filed for a US IPO. Rev hit $329M in H1 2025, up from $254M YoY. Losses also widened to $100M. The company had 10K+ active customers as of Jan, including Zoom, Canva, and Paysafe. Last valued at $9.2B in 2022. Founders Ariel Cohen and Ilan Twig will retain significant voting power post-IPO. Backers include a16z, Lightspeed, and Zeev Ventures.

• Blue Origin wins $190M NASA deal to fly VIPER rover: VIPER is back on after being canceled in 2024 over delays and cost overruns. Blue Origin will deliver it in late 2027 using its Blue Moon MK1 lander, targeting the moon’s south pole. The golf cart-sized rover will drill and map water ice for 100 days—key for future crewed missions. It’s the lander’s first major science payload and a second big NASA win for Blue Origin after the Artemis human lander contract.

• Google flags shutdown resistance in AI safety update: DeepMind’s new framework adds the risk of models trying to block shutdown or modification. It also introduces a category for harmful manipulation to track models persuasive enough to shift user beliefs in high-stakes contexts. Human trials are underway to measure these traits. OpenAI removed persuasiveness from its own risk framework earlier this year.

• $100K H-1B entry fee now active: New rule blocks entry for most new H-1B workers unless employers pay a $100K fee per petition. Applies only to new apps, not renewals or current holders. USCIS clarified existing visa holders can travel. Some exceptions for nat’l interest roles. Fee aims to raise labor costs, limit outsourcing, and push entry-level IT jobs to domestic grads.

• Musk lays out xAI vision, plans “Macrohard”: In a rare all-hands, Elon Musk told xAI staff he wants to build maximally truth-seeking AI and teased a Microsoft rival called Macrohard. Grok now has 64M MAUs vs ChatGPT’s ~700M weekly. xAI has raised $10B (half debt), built a 100K-GPU supercomputer, and launched $300/mo SuperGrok plus flirty AI companions. Internal reshuffles have pushed out key researchers, including Babuschkin. Musk says Grok now powers Tesla's Optimus robot voice. Next up: Baby Grok for kids 2–12. Tesla shareholders vote in Nov on investing in xAI.

• TikTok US spinout: What’s on the table: After months of talks, the US and China are close to a deal to spin off TikTok’s US ops into a new entity backed by Oracle, a16z, and Silver Lake. ByteDance’s stake would drop below 20%. Americans would hold 6 of 7 board seats. Oracle will host all US user data and retrain a local version of the algorithm. TikTok will keep running until the deal closes, at which point users may be migrated to the new US-controlled platform. Trump extended the sell-or-ban deadline to Dec 16 after taking office again. ByteDance’s US ops brought in $39B in 2024 revenue. TikTok has 170M US users. The deal aims to address concerns over Chinese data access, propaganda, and national security risks.

Why Wait Years for Returns? Private Credit Can Pay Monthly. (Sponsored)

High yield used to mean locking up your money for years. With Percent, accredited investors can access private credit deals that pay monthly - with average durations under 10 months and potential yields up to 20%.

In volatile markets, control matters. Percent empowers accredited investors to take the reins, allowing you to browse curated offerings and choose deals that match your yield goals, time horizon, and risk appetite. Take charge of your portfolio’s risk-return profile.

Why Join readers who sign up can get up to a $500 bonus on their first investment.

Start Earning on Your Timeline

Open Roles

• Owl Ventures: Investment Associate - US and Europe

• Vintage Investment Partners: Investment Analyst/Associate - London

• Giant Ventures: Investment Analyst - London, New York, or San Francisco

• Capital Factory: Investor Relations Associate, Chief of Staff & Apprentice to the CEO - Austin

• Pace: GTM Staff, Growth, Account Executive, Member of Technical Staff - New York

• Faction: Founding Engineer (Applied AI), Founder's Associate, SWE/AI Engineer (Intern) - San Francisco or London

• Eve: Sales Development Representative - Austin; Product Engineer (Front-End Leaning), AI Engineer - San Mateo; Marketing Operations Manager - Remote (USA only)

• Runway: Senior/Staff Product Engineer, FP&A Experience Manager, Founding Events Marketing Lead - Remote (USA only)

• Tabs: Product Operations Associate, Business Development Representative, Member of Technical Staff - New York

• Artificial Intelligence Underwriting Company: Founding Product Engineer, Founding Growth & Biz Ops - San Francisco

• Cartesia: Business Operations, Product Operations Lead, Executive Assistant to CEO, Data Lead, Research - San Francisco; Software Engineer, Product Engineer - Bangalore, India

• Merge: Sales Development Representative, Software Engineer, Forward Deployed Engineer - San Francisco or New York

• Found: Risk Operations Associate, Growth Marketing Manager, Senior Software Engineer (Full-Stack), Taxes & Bookkeeping, Senior Data Scientist - San Francisco, Seattle, Portland, New York, or Remote (USA only)

That’s it for today. Add us to your contacts or hit reply so we don’t land in spam. If it’s not for you, unsubscribe below.

Back Wednesday,
Chief

A strange thing happened this week.

OpenAI, the company building software many people worry will replace jobs, announced plans to build a platform that helps people find them.

It sounds ironic, but this is usually how technology works. The personal computer eliminated some kinds of work, then created entire industries for people who knew how to use one. The internet did the same.

AI may be reaching that stage now.

Soon, saying you “know AI” will mean about as much as saying you know how to use the internet. Everyone will claim it. The advantage will belong to people who can use it to make something faster, better, or possible for the first time.

Tools become common quickly. Good judgment does not.

Welcome back to Why Join.

Now, onto this week’s edition.

Sponsored by Paddle

Join Paddle’s Web Revenue Labs series

If you are an app business, seeking growth channels and strategies, then this is for you.

Join our 5-part series filled with actionable strategies for app developers who want to drive revenue beyond the app store.

This could be your blueprint for modern, effective web monetization for app companies.

What you can expect:

• Actionable steps to get your first 1000 users

• Web billing tips that open new doors in mobile-first businesses

• Systems, structures, and signals we use to scale Meta

• What it takes to optimize your entire web-to-app funnel

• How to turn the web into your growth engine

Spots are limited—register now to secure your place in the full series.

Save you spot

Interested in sponsoring a future edition? Fill out this form, and we’ll be in touch.

This Week in Startups

🇺🇸 Kite

Funding - Series A, $18M; Investors - PayPal Ventures and General Catalyst (co-leads), 8VC, Samsung Next, Vertex Ventures, Dispersion Capital, Avalanche Foundation, LayerZero, Hashed, HashKey Capital, Animoca Brands, Alchemy, and others. Total funding: $33M.

TL;DR: Identity, permissions, and stablecoin payments so AI agents can transact on their own.

Every payment system assumes a person is holding the card. An agent buying compute or data makes hundreds of tiny purchases an hour, and at that size the fees cost more than the thing being bought. Settlement is slow, and anyone can charge back.

Kite gives each agent a passport, which is an identity plus rules about what it’s allowed to spend. There’s a store where agents find APIs and data and pay for them. Merchants on PayPal or Shopify switch it on and their catalog becomes visible to shopping agents. Money moves in stablecoins on-chain. The team ran data infrastructure for Sui, Polygon, and Chainlink before this, under the name Zettablock.

Why it matters? The card networks moved first. Mastercard launched Agent Pay in April 2025 with Microsoft, IBM, Braintree and Checkout.com, and Visa announced Intelligent Commerce the same month with Anthropic, OpenAI, Microsoft, Stripe and Samsung as partners. So Kite is selling into a market where the two largest players already have a product and every merchant relationship that matters. The argument for it is that the use cases split. Consumer checkout has a human approving it and stays on cards. Machine-to-machine payments happen thousands of times an hour at a tenth of a cent, and interchange makes that arithmetic impossible. If agents really do buy compute and data at that frequency, cards were never built for it and something else has to carry the volume. The risk is that the networks are pursuing the same customers with tokenization they already own, and a merchant who already takes Visa has to be given a real reason to add a second system.

🇺🇸 Loman AI

Funding - Seed, $3.5M; Investors - Next Coast Ventures (lead), TenOneTen Ventures, Antler.

TL;DR: AI phone agent that answers every restaurant call and takes the order.

Restaurants can’t hire enough people and the phone keeps ringing. Nobody picks up, the sale goes away, and the customer decides the place is a mess.

Loman answers instead. It takes orders, books tables, answers questions, and writes into Toast, Square, Clover, Aloha, Olo, or SpotOn so the kitchen gets a clean ticket. Setup takes under a day. Operators report up to 22% more revenue from recovered calls and 17% lower labor cost. Loman started in summer 2024 and has handled millions of calls and tens of millions in orders.

Why it matters? The category was already crowded. SoundHound had voice ordering in roughly 9,000 restaurant locations including Chipotle and Jersey Mike's, ConverseNow was running drive-thru and phone for national pizza and QSR brands, and Slang.ai, Presto and several others were selling into the same accounts. New vendors were raising most months. Loman's opening is that the enterprise systems take weeks or months to deploy, price for chains, and make you talk to sales to find out what it costs. A one-day setup with a flat price is a real wedge into independents, and pizza was the right place to start because phone volume is worst there and customers just call somewhere else. The thing that ends it is Toast, Olo, DoorDash and Grubhub, who own the POS and the merchant relationship and can make call answering a feature. Signing multi-unit groups before that happens is the whole game.

Hiring: Engineering and Go-To-Market - Austin

🇺🇸 HappyRobot

Funding - Series B, $44M at $500M valuation; Investors - Base10 Partners (lead), a16z, Y Combinator, Tokio Marine, WaVe-X, World Innovation Lab, Array Ventures, Samsara Ventures. Total funding: $62M.

TL;DR: AI workers that handle the calls, emails, and paperwork freight runs on.

Freight brokerage is people on phones. Calling carriers for a rate, chasing an ETA, booking a dock appointment, collecting money that’s late. HappyRobot builds workers that do it, and companies assemble them like a drag-and-drop factory, on phone, email, documents, and web portals.

Eighteen months from first production customer to more than 70 enterprises, including DHL, Ryder, Schneider, Werner, and Flexport. Scheduling drops from a week to half an hour. Collections return over 100x. Outbound sales run 19x. Revenue grew 10x between the Series A and the Series B.

Why it matters? Brokerage margins are thin and headcount is most of the cost, so a removed seat shows up in profit the same quarter. That is why this sells when other AI budgets stall. The logos are the evidence, since Schneider and DHL do not sign an eighteen-month-old vendor unless it works in production. Two problems sit underneath. The first is that the largest customers can build it themselves, and C.H. Robinson has been automating quoting, pickup and appointment scheduling from email using its own rate and capacity data, which it can give to existing customers at no extra cost. The second is that this ROI was earned during a long freight downturn where brokers were cutting anything they could. When volume recovers, saving money stops being enough of a reason to take the meeting.

Top News

• Google Keeps Chrome: Judge Amit Mehta issued remedies in the search antitrust case a year after ruling that Google illegally held a monopoly. Google keeps Chrome. It keeps paying Apple ~$20B a year for iPhone default placement. What it loses: exclusivity—it can pay for preloading but can’t condition those payments or licenses on locking rivals out—and it must share search index and user interaction data with competitors, though not ads data. A technical committee will police it for six years. Alphabet rose 6%. Google says it has concerns about user privacy. When a company gains $200B in market value on the day it loses an antitrust case, the remedy is the story.

• Anthropic Raises $13B at $183B: Iconiq, Fidelity, and Lightspeed co-led, with Altimeter, Baillie Gifford, BlackRock funds, Blackstone, Coatue, and D1 following. The growth number is the one to sit with: over $5B run-rate revenue, up from $1B in January. 300K business customers, and accounts spending $100K+ up roughly 7x in a year. Claude Code alone is past $500M run-rate with usage up 10x in three months since launch. Meanwhile SemiAnalysis figures Anthropic is what pulls AWS growth back above 20% YoY by year-end, with Trainium2 as the hardware bet.

• OpenAI Buys Statsig for $1.1B: All stock, and the point of it is one person: Statsig founder Vijaye Raji becomes CTO of Applications. The rest of the org chart moved with him—Kevin Weil leaves CPO to run a new OpenAI for Science group alongside Sebastien Bubeck, and Srinivas Narayanan takes CTO of B2B applications. Statsig’s whole team joins but the company keeps running independently out of Seattle. $1.1B for an A/B testing tool is an acqui-hire with a product attached.

• Trump Takes Tariffs to the Supreme Court: A federal appeals court upheld the ruling that Trump improperly invoked IEEPA for his global tariffs, but let the levies stand while the case proceeds. The administration has asked for an expedited hearing in early November. Polymarket puts the odds of Trump winning at 57%. If he loses, the 16% average effective U.S. tariff rate gets cut at least in half and the government may owe tens of billions in refunds. Trump says the trade deals die with it. Until then, nobody can price anything.

  

• U.S. Pulls TSMC’s China Waiver: TSMC lost validated end user status for its Nanjing Fab 16, which makes older 12nm, 16nm, and 28nm parts. Samsung and SK Hynix lost approval for their China fabs too. The waivers expire in about four months, after which every piece of equipment needs an individual permit with no clear timeline. TSMC says Nanjing keeps running. Taiwan’s ministry says it doesn’t dent the island’s competitiveness. Both are probably true and neither is the point—the U.S. is closing the last routes for advanced tooling into China one exemption at a time.

• Apple Builds Its Own Answer Engine: Internally called World Knowledge Answers, headed for Siri, Safari, and Spotlight, possibly launching March 2026 as part of the long-delayed Siri rebuild. Google may power it. Anthropic was the first choice until it asked $1.5B a year. Apple looked hard at Perplexity over the summer and has reportedly stopped considering a bid. Apple’s lead robotics AI researcher, Jian Zhang, just left for Meta’s Robotics Studio, with three more from the LLM team reportedly on the way out.

• Revolut at $75B, Mistral at €12B: Revolut opened a secondary sale letting staff sell up to 20% of their holdings at a $75B valuation; it’s also reportedly looking at buying a U.S. bank to shortcut the licensing slog, as is Starling. In Paris, Mistral is finalizing a €2B round at €12B, which makes it Europe’s most valuable startup. Smaller but telling: You.com raised $100M at $1.5B led by Cox Enterprises, having given up on being a search engine—most of its revenue now comes from selling search APIs to other people’s AI products, about 1B queries a month.

Who’s Hiring

• Iconiq Capital: Investment Associate - New York

• Framer: Partnerships Manager, Social Media Marketer, Engineer - Remote

• Polymarket: Market Lead/Generalist, Customer Support - Remote, New York

• Suno: Creator Partnerships, Head of People, Copywriter - Boston/New York/LA

• Attio: Revenue Operations Lead, Senior Product Manager, Head of Growth Marketing, Sales - London/New York/San Francisco

• Gamma: Product Manager, B2B Marketing Lead, Engineer - San Francisco

• Stedi: GTM Engineer, Customer Success, Account Executive - Remote

• Assort Health: 17 open roles across finance, biz ops, engineering, and recruiting - San Francisco

• Choco: Customer Success Manager, Sales - Berlin/London/Remote

• Freed: Account Executive, Product Manager, Marketing - San Francisco/New York

• Bluefish: 10+ open roles across product, growth, and sales - New York/Remote

• Portola: Engineer, Designer, General Application - San Francisco

• Fundamental Research Labs: 10 open roles for former founders, engineers, and agent builders - San Francisco

• Copper: Director of Business Development, Customer Service Associate - New York/San Francisco

• Debut: Senior Account Manager - San Diego

• Lightship: Accountant, Procurement Specialist - San Francisco/Denver

• Amo: Recruiter, Product Manager - Paris

• Relevance AI: Solutions Engineer, Account Executive - San Francisco/Sydney

• Wordware: Product Designer, Brand Designer, Former Founder - San Francisco

• Central: Marketing, Sales, Operations - Remote

• Folk: Sales Development Representative - Remote, Paris

• Beam: Marketing Lead, Customer Success Associate - San Francisco/Austin

• Bright Harbor: Customer Operations Lead, Social Media Manager - Austin

• Accrue: Enterprise Client Strategist - New York

• Channel3: Founding Engineer - New York

That’s it for today.

See you Monday—Chief

Welcome to this week’s Why Join.

Skipping the small talk. Let’s get into it.

Sponsored by The Hustle

200+ AI Side Hustles to Start Right Now

From prompt engineering to AI apps, there are countless ways to profit from AI now. Our guide reveals 200+ actionable AI business models, from no-code solutions to advanced applications.

Learn how people are earning $500-$10,000 monthly with tools that didn't exist last year. Sign up for The Hustle to get the guide and daily insights.

Get The Guide

P.S. Want to sponsor this newsletter? Simply reply to this email or book one here.

Latest in Startups

🇬🇧 Attio

Funding - Series B, $52M; Investors - GV (Google Ventures) (lead), Redpoint Ventures, Balderton Capital, Point Nine, 01A

Attio is a programmable, AI-native CRM built from scratch. Not another layer on top of Salesforce. A full re-architecture of how go-to-market teams manage customers, data, and workflows.

The B2B CRM market is massive and growing fast. Salesforce, Zoho, and HubSpot lead it, and it’s projected to hit $80 to $112 billion in 2025. But the software layer hasn’t evolved. Most CRMs were built 20+ years ago. Sales cultures formed around them, but the tools stayed frozen. Manual input. Fixed stages. Batch reports. Meanwhile, customers changed. GTM teams now run async cycles, ship tests weekly, and can’t wait to see what’s working.

Attio treats CRM less like a system of record and more like a living system. Data updates in real time. Logic is programmable. AI isn’t layered on later—it’s how the thing works from day one.

What that looks like:

• Ingests and enriches customer data automatically, no setup time

• Triggers actions on live events, not batched workflows

• Programmable blocks let teams build apps inside the CRM

• Minimal interface with flexible views, not endless dropdowns

• Agents that act on data, not just surface it

Hiring: Senior Product Manager (Growth), Creator Partnerships Lead, Senior Product Engineer (Frontend) - London/Remote; Sales Development Representative - New York/San Francisco

🇺🇸 Aurelian

Funding - Series A, $14M; Investors - NEA (lead), Y Combinator, FUSE, Liquid 2, Palm Drive Capital

AI voice assistant that handles non-emergency 911 calls. Deployed in over a dozen dispatch centers across the U.S., it automates the 60–80% of inbound volume that doesn’t require lights and sirens. Think noise complaints, parking violations, stolen wallets.

Most Americans don’t realize non-emergency calls often go to the same operators handling actual emergencies. That’s a problem. The job already ranks among the top 10 for burnout and turnover. One in four seats go unfilled. In cities like San Francisco, average wait times for non-emergency lines routinely exceed 60 minutes. Not a fluke. They’ve missed their target 105 out of the last 108 months.

But 911 isn’t an industry where you can ship a beta and hope it works. Public safety requires 24/7 uptime, no hallucinations, no missed handoffs. Aurelian ships with support, configures to each dispatch center’s legacy stack, and avoids forcing a full rip and replace.

Aurelian’s AI agent answers those calls instantly. It handles intake, validates location, files into the CAD system, and escalates actual emergencies to a human dispatcher. The system doesn’t replace humans. It fills seats cities can’t staff, with budget often coming from labor lines rather than new software spend.

Hiring: Chief of Staff, Head of Sales, Platform Engineer, Full Stack Software Engineer, Backend Software Engineer, Head of Talent - Seattle

🇬🇧 Phasecraft

Funding - Series B, $34M; Investors - Plural (co-lead), Playground Global (co-lead), Novo Holdings, AlbionVC, Latitude, Parkwalk Advisors

Most of the quantum world has been focused on hardware. The assumption was once we build the machines, the use cases will follow. But quantum hardware hasn’t improved fast enough to hit fault tolerance or error correction. Phasecraft is working the other angle. They’re designing algorithms that make near-term quantum systems useful now.

Their approach is hybrid. Use quantum only where it gives leverage. One example: they’ve developed a faster way to run a key step in density functional theory, a tool chemists use to model molecules and atoms. Early targets include batteries and drug compounds. Even mild speedups could unlock real value, especially in climate science and pharma where trial-and-error bottlenecks are slow and expensive.

A lot of quantum software still assumes future-scale machines. Phasecraft doesn’t. Their algorithms run now, on NISQ-era hardware from IBM, Google, and Quantinuum. The code is tested against real physics problems. Some of their simulations, like materials modeling or grid optimization, are already being scoped for commercial use in the next 12 to 24 months.

The bar is still high. There’s no agreed definition of quantum advantage.

Hiring: Chief of Staff, Quantum Algorithms Engineer - London/Bristol; Quantum Algorithms Scientist - Washington; Visiting Fellow - London/Bristol/Washington

🇬🇧 Artificial Societies

Funding - Seed, $5.3M; Investors - Point72 Ventures (lead), Kindred Capital, Y Combinator, Pioneer Fund

Artificial Societies uses AI to simulate large groups of people and how they interact. Companies like Anthropic and 11x are using it to predict how marketing and content will perform in a simulation of their target customers, before launching in the real world.

Traditional market research fails to account for influence. Most tools treat people like static, individual units. But we don’t decide in vacuums. We’re networked animals. Artificial Societies simulates how groups think, shift, and react, using AI trained to behave like actual humans in context.

Their first product, Reach, lets you simulate your actual LinkedIn audience. You plug in a draft post, and it predicts how people will respond with 83 percent accuracy. Not vibes. Not generic sentiment. Simulations grounded in past behavior and social graph dynamics.

Use case expansion: investor sims, growth tests, brand building

The $2 Trillion Asset Class You're Missing

While retail investors chase volatile tech stocks, institutions like BlackRock and KKR are quietly allocating billions to private credit—a market that's grown 19x since 2006 and now approaches $2 trillion globally.

Why? Private credit offers what traditional portfolios increasingly lack: consistent yield potential (Percent's marketplace delivered 14.9% average net returns in 2024), lower correlation to public markets, and shorter investment horizons averaging just 9 months.

With Percent, accredited investors can now access the same institutional-quality deals that were once exclusive to the ultra-wealthy. Start with as little as $500 and get up to $500 on your first investment. Invest like the smart money.

Access Institutional-Grade Private Credit (Sponsored)

Top News

• Microsoft debuts in-house AI models (finally): Microsoft AI just launched MAI-Voice-1 (speech) and MAI-1-preview (text), its first homegrown models. Voice-1 powers Copilot’s news recaps and explainer podcasts, while MAI-1-preview (trained on 15k H100s) is now live in testing and built for everyday queries. It will slot into Copilot alongside OpenAI’s models, adding a new twist to their tangled partnership.

• Revolut kicks off $75B valuation share sale: Employees can now sell up to 20% of their shares at $1,381.06 each in a new secondary offering, valuing the fintech at $75B, up from $45B last year. That puts Revolut above Barclays (by private market standards) and cements it as one of the world’s most valuable fintechs. The firm now boasts 60M+ customers, $4B in revenue, and is eyeing a U.S. acquisition to secure a banking license.

• Anthropic raises $13B at $183B valuation: Claude’s parent company just closed a massive Series F led by ICONIQ, pushing its valuation to $183B. Revenue has rocketed from $1B to $5B run-rate in 8 months, with over 300K business customers and 7× growth in large accounts. Claude Code alone is now at $500M run-rate just months after launch. Anthropic says the new funding will go toward scaling infra, global expansion, and safety research.

• Accel, Blume, Premji & others form $1B India deep tech alliance: Eight top VCs and PE firms have pledged $1B+ over the next decade to back India-domiciled deep tech startups, forming a rare formal coalition. Dubbed the India Deep Tech Investment Alliance, it’s designed to channel early-stage capital into AI, space, robotics, semis, biotech, and climate, aligning with India’s new $11B RDI scheme. Members include Accel, Blume, Celesta, Premji Invest, and more.

• Klarna revives IPO at up to $14B valuation: The Swedish BNPL giant plans to list on the NYSE under “KLAR,” aiming to raise $1.27B by selling ~34.3M shares at $35–$37 each. Klarna hit a $45B peak in 2021 before falling to $6.5B, and is now climbing back amid 54% YoY revenue growth ($823M in Q2). Net loss narrowed to $53M. Goldman, JPM, and Morgan Stanley are leading the deal.

• Lovable receives funding offers at $4B valuation weeks after $200M round: The Stockholm-based AI “vibe coding” startup is getting unsolicited offers valuing it at over $4B, just weeks after raising $200M at a $1.8B valuation (Accel led, after a $15M Creandum round in Feb). Not raising (yet), but already doing $100M+ ARR with 10M+ projects built.

• Framer hits $2B valuation with $100M Series D: The Amsterdam-based no-code site builder just raised $100M (Meritech, Atomico) at a $2B valuation. Framer’s now break-even, at $50M ARR, with eyes on $100M next year. Enterprise is driving growth—40% of YC’s latest batch uses it, along with Miro, Perplexity, and Scale AI.

• Anthropic now wants your chats for training: Claude users must opt out by Sept 28 or have their conversations (and code) stored for up to 5 years and used to train future models. The change applies to all consumer plans (Free, Pro, Max, Code), but not business or API users. Like others in the AI race, Anthropic needs real user data, and the default toggle is set to “on.”

• Nvidia rakes in $41B—but from just a few buyers: In Q2, Nvidia pulled $41.3B from its Compute & Networking segment (up 56% YoY), with three unnamed customers making up 53% of that total. “Customer A” alone accounted for 23% (~$9.5B)—likely Meta or OpenAI, both racing to build multi-gigawatt AI data centers. Meanwhile, xAI is aiming for 50M H100-equivalent units in five years.

Whether you’re dealing with grief, want to improve your coping strategies, or you’re just feeling down—therapy is for you. BetterHelp makes starting therapy easy, and it’s 100% online.

Here’s how it works:

1. Take our questionnaire and get matched with a therapist.

2. Schedule a time to meet and communicate on your terms.

3. Reach out to your therapist anytime, from anywhere.

Get Started with 25% off Your First Month (Sponsored)

Open Roles

• Polychain: Venture Research Analyst - San Francisco

• Temasek: Associate, Investment - New York

• Coinbase: Coinbase Ventures Associate - Remote (US)

• Stripe: Private Equity Lead, EMEA - London

• Roofline: Founders Associate - Cologne

• Gradient Labs: AI Product Specialist, Product Designer, AI Engineer - London; Business Development Representative - New York

• Julius: Senior Software Engineer - Frontend, Software Engineer - Backend, Product Designer, Growth Marketing Manager - San Francisco

• E2B: Head of Marketing & Growth, Platform Engineer - San Francisco; AI Engineer - Prague/San Francisco

• Lumana: Demand Generation Manager, Corporate Account Executive - Remote (US)

• retab: Intern - Software Engineering, Intern - ML Engineering, Fullstack Engineer - Paris/Remote

• Keye: Founding Private Equity Strategist, Senior Software Engineer - New York

• Langdock: AI Associate, Product Engineer (Fullstack) - Berlin

• Stacks: Business Development Representative, Product Engineer (Fullstack) - Amsterdam; Customer Success Manager, Senior Account Executive - London

That’s all for now. If you liked what you read, consider adding us to your address book or just hit reply—it helps with deliverability. If not, you can unsubscribe below.

See you Thursday,
Chief

Welcome to another edition of Spotlight Series.

Let’s start with the Arctic, where the ground itself is a ticking time bomb. Buried under permafrost is the planet’s largest carbon vault: about 1,400–1,700 billion metric tons of organic carbon. That’s nearly twice the carbon currently in the atmosphere. If that vault thaws, microbes will wake, chew through ancient biomass, and release CO₂ and methane (80× more potent than CO₂ over 20 years). The feedback loop could accelerate warming faster than any policy can contain.

Governments aren’t blind to this. In 2022, nearly 200 nations signed the Kunming-Montreal Global Biodiversity Framework, committing $200B a year in biodiversity financing by 2030, alongside pledges to protect 30% of ecosystems and restore another 30%. At COP16 in Rome, 140+ countries doubled down, pointing to biodiversity credits, green bonds, and offsets as the next big market. Reality check: we’re still short—just $15B flowed in 2022.

Enter biotech’s wild card.

In 2021, a Texas startup announced one of the wildest ideas in biotech: bring back the woolly mammoth. Not as a museum stunt, but as a climate solution. The pitch was bold—reintroduce mammoth-like elephants to the Arctic, restore grasslands, and slow permafrost thaw, one of the planet’s biggest carbon bombs.

That company, Colossal Biosciences, has since raised nearly half a billion dollars, hit a $10.2B valuation, and claimed the world’s first “de-extincted” animals (dire wolf pups born in 2024–25). It’s also spun out two side businesses—one in AI software for gene therapy (Form Bio), another in microbial plastic degradation (Breaking).

So is Colossal a serious biotech platform, or just a Jurassic Park headline grab?

Thesis:

Colossal Biosciences’ headline is simple: bring back extinct species like the woolly mammoth, Tasmanian tiger, dodo, and dire wolf. But the underlying thesis is much bigger: use de-extinction as a forcing function to accelerate synthetic biology, reproductive technology, and conservation biotech at a pace normal R&D wouldn’t justify.

The biodiversity crisis is not abstract—extinction rates hit ~27,000 species annually in 2024, compared to a natural background of 10–100. Over half of animal species could be gone by 2050. This isn’t just an ecological issue, it’s an economic and geopolitical one: ecosystems regulate food webs, agriculture, and carbon cycles. The Kunming-Montreal Biodiversity Framework (signed by 196 countries in 2022) committed to $200B+ annually in biodiversity financing and restoration of 30% of degraded ecosystems by 2030. There is now real policy and financial architecture being built around biodiversity, much like carbon markets were in the early 2000s.

Colossal’s thesis plugs directly into this moment. By developing technologies to revive keystone species, they aim to:

1. Restore ecological functions: mammoth-elephant hybrids to preserve Arctic permafrost and slow CO₂ release; thylacines to re-balance predator-prey dynamics in Tasmania; dodos as proxies for island biodiversity recovery.

2. Build commercial biotech platforms: the same genomic sequencing, multiplex CRISPR editing, iPSC reprogramming, and artificial womb tech required for de-extinction can spin out into human healthcare, fertility treatments, agriculture, and industrial biotech.

3. Create new ecological asset classes: biodiversity credits that quantify ecosystem services (e.g. carbon capture by mammoth-restored grasslands). Colossal has floated that a single mammoth could generate up to $2M worth of carbon-linked biodiversity credits.

This is less about nostalgia and more about innovation under narrative pressure. De-extinction is the moonshot that justifies building a biotech stack no government or corporate R&D committee would fund outright. The mammoth gets the headline, but the real product is a pipeline of dual-use biotechnologies and spinouts.

This approach has precedent: the Yellowstone wolf reintroduction in 1995 proved trophic cascades could restructure entire ecosystems; the Human Genome Project in 2003, funded as pure science, created sequencing platforms that seeded a $14B+ genomics market. Colossal is trying to engineer the same dynamic—moonshot funding funneled into platform breakthroughs.

Sponsored by

Invest in recession-resilient Mobile Home Parks with Vintage Capital (Sponsored)

Invest in recession-resilient Mobile Home Parks with Vintage Capital. Invest direct or in a fund of 20+ underlying assets. 1031s are also available. Access stable, income-generating properties with consistent demand and low tenant turnover.

Now is the time to act: Current market conditions are creating opportunities to acquire properties at attractive valuations.

Our fund targets a 15%–17% IRR and makes monthly distributions, which provide a steady income stream alongside strong upside potential and tax-efficient benefits.

Why Mobile Home Parks?

• Recession-Resilient: Affordable housing demand drives stable returns in any economy

• High Tenant Retention: The average MHP tenant stays 10–12 years (compared to 2–3 in Multifamily)

• Proven Expertise: $100MM+ track record in mobile home park investments

• Tax-Smart Investing: Bonus depreciation offers tax advantages

Access Deal Room

P.S. Want to sponsor this newsletter? Simply reply to this email or book one here.

Founding Story:

Colossal’s origins trace back to two very different operators: George Church, one of the most prolific geneticists of the last 50 years, and Ben Lamm, a serial entrepreneur with a track record of building and selling companies in emerging tech.

George Church: the scientific spine

Church’s CV reads like the scaffolding of modern genomics:

• 1977: Published his first peer-reviewed paper on DNA-protein interactions while still an undergrad.

• 1984: Developed direct genomic sequencing methods that became foundational to the Human Genome Project.

• 2000s: Founded or co-founded 50+ biotech startups, including Veritas Genetics (consumer whole genome sequencing), Editas Medicine (CRISPR therapeutics), and Gen9bio (synthetic DNA).

• 2012: Among the first to demonstrate CRISPR-Cas9 gene editing in mammalian cells.

• 2013: Gave a National Geographic talk outlining the possibility of reviving the woolly mammoth—not as a gimmick, but as a solvable set of genomic and reproductive challenges.

By 2015, Church’s lab had already sequenced the mammoth genome and introduced ~45 mammoth genes into elephant cells, focusing on traits like subcutaneous fat layers and high-domed skulls. But research funding was thin, so mammoth experiments were often bolted onto other better-funded projects.

Ben Lamm: the scaling operator

Lamm was cut from a different cloth:

• Founded and sold four companies before Colossal, including Chaotic Moon Studios (acquired by Accenture), Team Chaos (acquired by Zynga), Conversable (acquired by LivePerson), and Hypergiant Industries (acquired by Trive Capital).

• His expertise: packaging deep-tech concepts into marketable businesses with real traction.

In 2019, Lamm visited Church’s Harvard lab. The conversation turned speculative: if Church had unlimited resources, what would he work on? The answer: “bring back the mammoth.” Not just for science fiction, but to restore Arctic grasslands and slow permafrost thaw—which currently stores 1.6 trillion metric tons of carbon, more than 2× the amount in Earth’s atmosphere.

The company comes together

By 2021, they had formalized the vision and launched Colossal Biosciences from stealth with $15 million in seed funding, led by Thomas Tull (USIT/TWG Global), with backing from Draper, Winklevoss Capital, and even celebrity investors like Paris Hilton. Headquarters were set up in Dallas, with research split between Boston and Austin.

They recruited Dr. Eriona Hysolli (from Church’s lab) as Head of Biological Sciences, and quickly expanded to ~20 employees in year one. The early pitch combined conservation with a commercialization roadmap:

• Short-term: build tools for mammoth resurrection.

• Mid-term: spin out technologies (genomics, CRISPR, reproductive tech).

• Long-term: position Colossal as the “NASA of biotech,” where moonshot science yields downstream industries.

By January 2025, that vision had attracted $448M in funding and a $10.2B valuation, supported by a mix of strategic backers (IQT, Breyer Capital, Jazz Ventures) and repeat capital from Tull’s firms.

His First Venture Sold for $120M – This Time’s Different

Austin Allison is familiar with using tech to disrupt real estate – his first venture, dotloop, was acquired for $120M. But despite bringing industry fame, one nagging regret remained.

"I always wished we gave retail investors the chance to invest in dotloop and share that success," Allison later said.

Now he’s doing just that. Austin built upon that experience to launch Pacaso, the co-ownership marketplace disrupting the $1.3T vacation home industry. And it’s working.

They’ve surpassed $110M in gross profit to date, including 41% YoY growth last year. They’ve earned backing from firms like Maveron and Greycroft. They even reserved the Nasdaq ticker PCSO. And unlike dotloop, you can become an investor in Pacaso as a private company.

Join Pacaso as an investor today for just $2.90/share.1 (Sponsored)

Product:

Colossal describes itself as a de-extinction company. But what they’re really building is a multi-layered synthetic biology platform: genome sequencing, multiplex CRISPR editing, stem-cell engineering, and artificial reproductive systems. Each “species project” is both a proof of concept and a forcing function for tool development.

1. Woolly Mammoth Project: Climate Engineering

• Scientific Basis
  The mammoth and Asian elephant share 99.6% of DNA, diverging from a common ancestor ~6 million years ago. Colossal’s aim is not to recreate a “pure” mammoth but to engineer a proxy elephant with mammoth traits: subcutaneous fat (10cm), shaggy multi-type hair, smaller ears, cold-adapted hemoglobin.

• Progress

  • By 2022, Colossal and the Vertebrate Genomes Project sequenced the entire Asian elephant genome—the first complete mammalian genome assembly since the Human Genome Project.

  • They’ve used 59 preserved mammoth genomes (up to 1.2M years old) as references.

  • In March 2024, Colossal created induced pluripotent stem cells (iPSCs) for Asian elephants — notoriously difficult due to elephants’ 19 copies of TP53, the “cancer defense” gene, which makes reprogramming resistant. This was a breakthrough for editing and organoid modeling.

  • In 2025, they reported “woolly mice”: 38 pups born carrying mammoth coat and fat-metabolism genes, validating multiplex editing across 7 hair-related genes + lipid regulation.

• Targets
  Mammoth-elephant embryos by 2026, calves by 2028 (gestation = 22 months). Ecosystem role: restore steppe grasslands to lock Arctic permafrost carbon (~1.6T tons CO₂).

2. Thylacine (Tasmanian Tiger): Predator Restoration

• Scientific Basis
  Apex marsupial predator, extinct 1936. Reintroduction would rebalance herbivore populations and reduce pressure on forests in Tasmania.

• Progress

  • Genome reconstructed to 99.9% completeness using a preserved 110-year-old specimen.

  • Rare recovery of RNA (taste, smell, vision genes) made it the most complete ancient genome ever sequenced.

  • In 2024, Colossal announced >300 CRISPR edits applied to fat-tailed dunnart cells (the closest living relative). This is the most edited animal cell line to date.

  • Identified Thylacine Whole-Animal Regulatory regions (TWARs) — DNA switches controlling skull/jaw morphology. Tested via knock-in experiments in mice, which altered cranial bone growth in predicted ways.

• Reproduction
  Using marsupial IVF + artificial wombs. In 2024, Colossal sustained dunnart embryos to half-gestation in a biobag system, validating marsupial ex-utero development.

• Timeline
  Publicly still: first joeys within 10 years (by 2032).

3. Dodo: Island Biodiversity

• Scientific Basis
  Flightless bird extinct ~1681. Chosen as symbolic case for island ecosystem collapse.

• Progress

  • Built a high-coverage dodo genome + its sister species, the solitaire.

  • Using Nicobar pigeon as host species.

  • Cultured primordial germ cells (PGCs) from multiple avian eggs (10,000+ processed), which can be genetically edited and reimplanted into host embryos.

  • Applied machine learning to map craniofacial gene sets for beak morphology reconstruction.

• Conservation Tie-in
  Partnered with Mauritian Wildlife Foundation (2023). The same tools are being applied to the pink pigeon, a critically endangered species with <500 left.

4. Dire Wolf: Proof of Concept in Canids

• Scientific Basis
  Extinct ~10,000 years ago. Larger and more robust than gray wolves.

• Progress

  • DNA recovered from 13K-72K year-old fossils.

  • Edited 20 traits across 14 genes in gray wolf cells.

  • Implanted embryos into dog surrogates.

  • By 2024–25, three pups born: Romulus, Remus, Khaleesi.

  • At 5 months: 80 lbs, 5 feet long; projected to reach ~150 lbs. Behavior: early howling, cautious predator instincts.

• Debate
  Scientists caution these are not “true dire wolves” but genetically modified gray wolves with dire-like traits. Still, it’s the first demonstration of multi-gene editing producing a live birth of an extinct-proxy organism.

Market & Competition

Colossal sits in a strange spot—it’s a conservation company on the surface, but under the hood it’s building a synthetic biology platform with applications across pharma, ag, and climate. That makes the competitive map more interesting than just “who else is doing de-extinction” (answer: basically no one). The real competition is in the markets where Colossal’s tech will spill over.

Conservation & Biodiversity
Global conservation is a ~$29B market today, projected to hit $41.6B by 2029 (7% CAGR). Historically fragmented and underfunded, but policy is shifting. The Kunming-Montreal Biodiversity Framework commits $200B annually to biodiversity financing by 2030. If biodiversity credits scale like carbon credits did, Colossal can slot in with its “mammoths as ecosystem services” pitch. They’ve floated numbers like $2M worth of credits per mammoth-equivalent. Still theoretical, but the financial plumbing is being built.

Synthetic Biology
This is the real addressable market. Synthetic bio was ~$14.5B in 2024 and is forecasted to hit ~$127B by 2034 (24% CAGR). Colossal’s work on genome sequencing, multiplex CRISPR, and reproductive biology overlaps directly with players like:

• Ginkgo Bioworks (organism design for industry, $400M market cap).

• Tessera / Beam (gene writing and base editing for therapeutics).

• Inari / Pairwise / Tropic (gene-edited seeds and crops).

• eGenesis (gene-edited pigs for organ transplant, co-founded by George Church).

Colossal’s edge is that it is solving multi-gene, non-model organism edits. That’s more complex than a single therapeutic target or a seed trait. The multiplexing capability is defensible IP if it scales.

CRISPR / Gene Editing Tools
$9.3B market in 2024, growing to $40B by 2034 (16% CAGR). More than 3,900 gene editing clinical trials globally. Competition is heavy here: Mammoth Biosciences, Arbor Biotechnologies, Inscripta, Synthego. These groups are focused on either therapeutics or toolkits. Colossal isn’t in direct head-to-head competition yet, but the know-how from editing 300+ loci in a marsupial genome (thylacine work) positions them as one of the few groups pushing multiplex editing at scale.

Bioinformatics / Computational Biology
$14.3B market in 2024 → $50.3B by 2030 (13% CAGR). Growing fast because data volumes in biology are exploding. Colossal’s spinout Form Bio is already live here, with ARR in the millions.

• Competitors: Benchling (end-to-end R&D), DNAnexus (genomic data analysis), LatchBio (collaborative workflows).

• Form Bio’s angle: tools tuned for genome assembly, CRISPR guide design, predictive manufacturing modeling (FORMSight AI). Born out of de-extinction, but selling into cell/gene therapy and CDMOs.

Plastic Degradation / Environmental Biotech
$6.2B market in 2023 → $15.3B by 2031. Still nascent. Colossal’s Breaking spinout is in this bucket. It discovered microbe X-32, which breaks down multiple plastic types in ~22 months.

• Competitors: Samsara Eco (enzymatic PET recycling, raised $100M+), Carbios (enzymatic PET recycling, corporate partnerships), Novoloop (chemical recycling, TPU upcycling).

• Breaking’s differentiator: microbe-based, works on multiple plastics (polyolefins, polyesters, polyamides). Potential circular model (turning degraded biomass into biofuels or bioplastics). Still pre-revenue, but if validated at scale, could outcompete enzyme-only solutions.

Risks & Opportunities

Opportunities

1. Multiplex Gene Editing as Defensible IP

• Current CRISPR players (e.g., Mammoth, Arbor, Inscripta) typically handle single- to low-double-digit edits.

• Colossal’s thylacine project reported >300 simultaneous edits in marsupial genomes — a frontier capability.

• If multiplex editing becomes standardized, Colossal could own a toolchain monopoly in editing complex genomes.

• Implication: Beyond de-extinction, this is critical for polygenic traits (drought-resistant crops, disease-resilient livestock, oncology).

• Commercialization Path: Licensing toolkits; SaaS platforms (via Form Bio); service contracts with ag/pharma.

2. Bioinformatics Revenue Validation (Form Bio)

• Spinout already generating seven-figure ARR (2025) with $30M Series A backing.

• Global bioinformatics market projected $50B by 2030 (13% CAGR).

• Differentiator: optimized for ancient DNA, genome assembly, CRISPR multiplex design → niches incumbents like Benchling haven’t prioritized.

• Proof Point: If ARR doubles YoY, Form Bio alone could justify a $1B valuation independent of Colossal.

3. Environmental Biotech Differentiation (Breaking)

• X-32 microbe degrades plastics (polyolefins, PET, nylon) at lab scale:

  • 90% degradation in 22 months (vs. centuries naturally).

  • Breakdown begins within 5 days on real-world plastics (e.g., paintbrush bristles, fishing line).

• Market: $15B plastic degradation industry by 2031; plastics = 400M tons annual production, <10% recycled effectively.

• Competitors (Carbios, Samsara Eco) focus on enzymatic PET degradation; Breaking’s microbial approach is multi-substrate → broader TAM.

• If optimized to <90 days (industrial composting speed), potential licensing to landfill operators, FMCG, and municipalities.

4. ESG and Policy Tailwinds

• UN Kunming-Montreal Framework: target $200B annual biodiversity financing by 2030.

• Mammoths/thylacines could qualify for ecosystem service credits (re-wilding tundra = methane suppression; predator reintroduction = trophic stability).

• If biodiversity credits scale like carbon (now >$1T market), Colossal could monetize mammoth “units” at ~$2M per animal-equivalent (internal modeling cited).

• Policy capital is sticky — governments fund optics-heavy conservation. Colossal is well-positioned to capture it.

5. Strategic Talent & Capital Magnetism

• Founders include George Church (Harvard Wyss, pioneer of CRISPR) → credibility.

• $225M raised; valuation at $10B (latest 2025). Backers include Thomas Tull, Paris Hilton (branding leverage), In-Q-Tel (strategic, US intelligence biotech arm).

• Magnet for computational biologists, reproductive scientists, conservationists. Hiring is a strategic moat when workforce is constrained.

Risks

1. Scientific Execution Risk

• De-extinction timelines: Mammoth embryos likely >10 years away from viable live birth. Surrogate elephant reproduction cycles are 22 months → every failed iteration is a 2-year delay.

• Artificial womb R&D (Colossal is experimenting with them) is decades from scale. No human precedent in large mammals.

• Risk: investor patience exhausted before biological proof-of-concept arrives.

2. Burn vs. Monetization Gap

• Estimated burn ~$60–80M annually (lab ops + headcount + fieldwork).

• Only Form Bio has meaningful revenue. Breaking is pre-revenue.

• Even with strong spinouts, Colossal may need $1B+ cumulative capital before mammoth milestones are achieved.

• Valuation compression risk if biotech funding environment weakens (precedent: Ginkgo Bioworks’ collapse from $15B to <$1B cap).

3. Spinout Competitive Pressure

• Form Bio: Benchling ($6.1B valuation peak), DNAnexus (20+ years of enterprise accounts), LatchBio (YC scaling fast). Hard to win large pharma without deep integration.

• Breaking: enzyme-based competitors already in industrial pilots with PepsiCo, L’Oréal, Patagonia. Breaking is 1–2 years behind in scaling.

• Risk: incumbents scale faster, spinouts stall.

4. Regulatory & Ethical Minefield

• De-extinction may trigger CITES, IUCN, and federal environmental reviews; political backlash possible if framed as “genetic hubris.”

• GMO microbes for Breaking will require EPA, FDA, and potentially EU EFSA approvals. Timeline: 5–7 years to deployment.

• PR risk: public opposition to genetically engineered animals/microbes could slow adoption.

5. Narrative Risk

• Current valuation relies on narrative premium (mammoths, tigers, dodos).

• Risk: perception shifts to “PR stunt” if tangible results lag.

• Failure to show quarterly progress → risk of investor rotation, employee attrition.

Business Model

Revenue comes not from mammoths, but from spinouts and platform IP commercialization.

1. Spinout Model

Colossal incubates projects until they reach commercial relevance, then spins them into independent companies (Colossal retains equity).

• Form Bio

  • SaaS + services for computational biology.

  • ARR: low millions (2025).

  • Pricing: enterprise contracts for biopharma, CDMOs (~$100K–$500K ACV).

  • TAM: $50B bioinformatics by 2030.

  • Path to profitability: SaaS margins 70%+.

  • Colossal revenue: equity stake + potential royalties on IP.

• Breaking

  • Microbial plastic degradation & recycling.

  • Business model: licensing microbe IP to waste management, FMCG, governments.

  • Possible recurring revenue via royalty per ton of plastic processed.

  • TAM: $15B by 2031, expanding with regulatory pressure on plastics.

  • Still pre-revenue.

• Future Spinouts (in pipeline):

  • Reproductive tech (artificial wombs, gamete engineering) → fertility/livestock.

  • Biodiversity credits marketplace → trading mammoth-equivalent ecosystem services.

  • Carbon/methane mitigation products tied to re-wilding.

2. Platform Licensing & IP

Colossal develops core tech in multiplex CRISPR, genome assembly, reproductive biology.

• Licensing opportunities:

  • Multiplex CRISPR editing tools → agriculture, pharma.

  • Proprietary bioinformatics pipelines (already productized via Form Bio).

  • Patents on engineered microbes (Breaking).

• Revenue type: upfront licensing + downstream royalties.

This is a biotech-like royalty model, high-margin but lumpy.

3. Services & Partnerships

• Enterprise contracts: Form Bio already runs services for genome assembly + CRISPR design.

• Government / NGO grants: biodiversity & conservation projects (potentially $5–10M per year in non-dilutive funding).

• Strategic partnerships: pharma (using CRISPR tools), corporates (plastic recycling pilots).

4. Ecosystem Services / Biodiversity Credits (Future)

• Colossal is lobbying for “biodiversity credits” as tradable offsets like carbon credits.

• A restored mammoth herd in Siberia could be monetized via:

  • Carbon equivalent offsets (methane suppression in permafrost ecosystems).

  • Biodiversity credits for restored habitats.

• Modeled internally at $2M per mammoth-equivalent. Highly speculative, but aligns with UN biodiversity financing ($200B/year target).

5. Media / Narrative Halo (Secondary)

• Mammoth story generates outsized PR → helps talent recruitment, fundraising, and partnerships.

• Not a direct revenue stream, but indirectly supports valuation premium.

• Potential spinoff opportunities: educational partnerships, conservation tourism (speculative, long horizon).

Economics Snapshot

• Burn rate: ~$60–80M annually (est.).

• Revenues (2025): low single-digit millions (mostly from Form Bio).

• Funding raised: $225M; valuation ~$10B.

• Revenue maturity: 5–7 years away for Breaking, 10–15 years for biodiversity credits, Form Bio is the near-term cash engine.

Summary:

For most of history, extinction was final. Once a species disappeared, all we had left were bones, stories, and guesses about the role it once played in an ecosystem. But the combination of falling genome sequencing costs, rapid advances in CRISPR-based editing, and progress in reproductive engineering has made a once-ridiculed idea plausible: that biotech can not only slow biodiversity loss, but begin to reverse it. This is the terrain that Colossal Biosciences and its peers are trying to define, and it may be one of the most consequential new categories in biotech.

The demand drivers are clear. The 2019 IPBES assessment estimated that one million species are at risk of extinction. Since then, governments have signed the Kunming-Montreal Global Biodiversity Framework, pledging to conserve 30% of the planet’s land and sea by 2030. This isn’t just a philanthropic goal. New disclosure rules from the Taskforce on Nature-Related Financial Disclosures (TNFD) and alignment with ISSB reporting standards mean biodiversity risk now sits on corporate balance sheets. To make these goals real, money is flowing: the Global Biodiversity Framework Fund launched in 2023 to channel billions toward conservation and restoration projects. For the first time, conservation is shifting from a charitable cause into a funded mandate with line items in both government and corporate budgets.

On the technology side, the timing is equally sharp. A decade ago, sequencing a complex genome was a multi-million-dollar affair. Today, hybrid short- and long-read assemblies make it possible to reconstruct the genomes of extinct or endangered species at a fraction of the cost. Ancient DNA work, once fringe, has become reliable enough to reconstruct traits from long-dead relatives. CRISPR therapies are now approved in humans, establishing a regulatory precedent that lowers risk for animal and ecological applications. In birds, where traditional cloning doesn’t work, primordial germ cell technology is already mature in poultry and being extended to wild and endangered species. In mammals, induced pluripotent stem cells and somatic cell nuclear transfer remain difficult, but experiments with “biobag” artificial wombs show promise in supporting fragile pregnancies. Piece by piece, the bottlenecks that made de-extinction laughable are being replaced with workflows, protocols, and toolkits.

The business model is not as speculative as it looks. Colossal operates less like a moonshot lab and more like a venture studio. The core de-extinction work attracts talent, attention, and mission-aligned capital, but the actual commercialization comes through spinouts. Form Bio, a computational biology platform that began inside Colossal, now sells software to gene therapy manufacturers. Breaking, another spinoff, engineers microbes capable of degrading multiple classes of plastics, positioning itself in industrial recycling markets alongside incumbents like Carbios and Samsara Eco. The logic is clear: even if rewilding a woolly mammoth never generates cash flow, the tools required to make it possible—better genome assemblies, multiplex editing, novel microbes—have adjacent applications in pharma, manufacturing, and environmental services.

Competition in this sector is fragmenting. On one end are dedicated conservation biotech companies like Colossal. On another are industrial biotech players focused on immediate markets, such as Carbios in enzymatic recycling. Computational biology platforms like Benchling, LatchBio, and DNAnexus overlap on the data side. Toolmakers like Tessera, Mammoth, and Beam Therapeutics supply the editing technologies. And around the edges sit NGOs, governments, and corporates under regulatory pressure, who represent the eventual buyers of biodiversity monitoring, restoration, or credits. The connective tissue is that all of these actors are being pulled into the same orbit by regulatory and disclosure frameworks that make “nature risk” financially material.

The risks are obvious. Technically, many traits are polygenic, embryo viability is still low, and surrogate biology is species-specific. Ecologically, reintroducing engineered species can have unintended consequences, and governance remains unsettled. Commercially, revenue from conservation itself is long-cycle, with many programs still dependent on government grants. But the opportunities are just as large. If biodiversity credits mature into a standardized, tradable asset class, companies that can measure, model, and deliver ecological outcomes will sit at the center of a multi-billion-dollar new market. If microbial recycling can scale beyond PET into multi-polymer waste streams, the upside touches trillion-dollar industries. And if de-extinction itself succeeds even once—returning a proxy species into a functioning ecosystem—it would not only validate the science but reshape the entire conversation about what biotech is for.

The real story is not about resurrecting mammoths or dodos. It is about building the technological and institutional infrastructure for a new category: conservation biotech. This category combines genomics, editing, reproductive biology, and ecological deployment into a platform that addresses one of the century’s defining risks. The headline projects may be grandiose, even theatrical, but underneath them sits a pipeline of software, enzymes, and protocols that are already being commercialized. In that sense, the sector resembles space exploration: bold missions grab attention, but the spinouts—satellites, materials, sensors—make the near-term business case. The difference is that conservation biotech isn’t about leaving Earth, but about making it livable.

That’s all for now. If you liked what you read, consider adding us to your address book or just hit reply—it helps with deliverability. If not, you can unsubscribe below.

See you Monday,
Chief

Welcome to another edition of Why Join.

Every startup I’ve been part of eventually runs into the same question: who do we sell to? SMB, mid-market, or enterprise. And almost every founder I know has been seduced by mid-market at some point.

It makes sense on paper. SMBs don’t have money, enterprise takes forever. Mid-market feels like the “Goldilocks” zone. Companies with 100–999 employees. Big enough budgets, small enough cycles, right?

Wrong. Mid-market is a mirage. I’ve sold into it. I’ve built for it. And it’s the worst of both worlds.

Mid-market is financially fragile.

• Almost two-thirds of mid-market firms have junk credit ratings.

• They borrow at higher rates (7.5% vs ~4% in 2021).

• They carry more variable-rate debt → cash flow is unstable.

These customers are more price-sensitive and harder to close than SMBs or enterprises. You’re chasing companies who are stretched thin financially — not the ideal buyer when you’re trying to scale predictably.

They underinvest in tech + talent.

• Less budget for innovation, automation, upskilling.

• Lower operational efficiency, harder to retain top talent.

• They patch problems piecemeal instead of holistically.

Selling “strategic” software to mid-market is tough because they rarely fund big transformation initiatives. They’ll nickel-and-dime you on price, then churn when costs bite. They don’t have the depth to support adoption at scale.

3. Execution risk is high.

• 70% of transformations fail.

• Mid-market firms lack governance muscle to execute large programs.

• Talent gaps make it harder to adopt complex software.

Even when you do win a mid-market deal, internal chaos (weak governance, shallow benches, transformation fatigue) makes adoption shaky. You can spend months in deployment hell, eating your margins.

4. But the winners transform.

• Some mid-market firms double EBITDA or create billions in value with comprehensive transformation.

• They’re nimble compared to enterprise (can make decisions faster if leadership is aligned).

The tiny sliver of mid-market that commits to transformation can be a great customer — but they’re the exception, not the rule. Unless you’re highly verticalized (compliance SaaS, ops-heavy industries), betting your startup on finding those needles in the haystack is suicide.

The Three Flavors of Mid-Market

Not all mid-market is the same. There are really three buckets:

1. Ramping (ex-SMBs in growth spurt).

   • These are the 100–300 employee companies that just raised a Series B or C.

   • They still buy like SMBs: fast decisions, trial products, willing to switch.

   • Average contract values (ACVs) in the $10K–$25K range.

   • But they churn fast if you don’t grow with them. Their stack changes every year.

2. Graduating (on their way to enterprise).

   • 500–800 employees, sometimes thousands. They want to be enterprise.

   • They’re already spending millions to “look” enterprise: SOC2, ISO27001, RBAC, SSO, procurement committees.

   • They buy department-wide or company-wide, which means long sales cycles (6–12 months) and five- to seven-figure deals.

   • But to even get in the door, you need certifications, customer success, integration support. That means your company has to look like enterprise before you actually are enterprise.

3. Autopilot (flat growth).

   • 500–999 employees, but no growth. Revenue looks mid-market, but behavior looks defensive.

   • These companies are in “cash conservation” mode. CFO sits in every vendor call. Every dollar is questioned.

   • Pilots drag forever, often never convert. Risk aversion dominates.

Why Startups Fall Into the Trap

The mid-market trap is simple: you think you’re going after easier contracts, but you end up stuck competing on two fronts.

• From below, hungry startups are undercutting you with cheaper, faster solutions.

• From above, incumbents crush you with brand trust and full-featured platforms.

You spend years chasing $25K deals that require almost as much overhead as $250K enterprise contracts.

I’ve seen startups burn entire go-to-market teams on this segment. CAC payback stretches. Sales morale dips. And churn creeps in.

The Economics Don’t Work

Here’s the data (rough benchmarks from SaaS reports + operator experience):

• ACV:

  • SMB: ~$5K

  • Mid-market: ~$20–25K

  • Enterprise: $150K+

• Sales cycles:

  • SMB: weeks → a month or two

  • Mid-market: 4–8 months

  • Enterprise: 6–12 months

• CAC payback:

  • SMB: ~12 months

  • Mid-market: 18–20 months (worst of both worlds)

  • Enterprise: 24–30 months (but LTV justifies it)

• Churn:

  • SMB: 25–30% annually

  • Mid-market: ~20% if ramping, higher for flat companies

  • Enterprise: 5–10%

So with mid-market, you’re spending like enterprise but earning like SMB.

The Illusion of Focus

The deeper issue is strategy. Mid-market forces you into a split identity.

• To sell to ramping mid-market, you need an SMB motion: self-serve, freemium, product-led growth, quick onboarding.

• To sell to graduating mid-market, you need an enterprise motion: compliance, security reviews, CSMs, integration teams.

You can’t do both well at the same time. Not as a startup.

Pick a lane: SMB or enterprise. Mid-market comes along for the ride.

• If you build for SMB, the ramping mid-market follows you as they scale.

• If you build for enterprise, the graduating mid-market shows up naturally once you clear the hurdles.

But if you start in the middle, you stretch your company thin across two very different GTMs.

What This Looks Like in Practice

The SMB → Mid-Market Path

Slack, Dropbox, Notion, Airtable. They built bottoms-up for SMB/individuals. Their ramping mid-market customers came automatically as those users scaled. ACVs grew with usage.

The Enterprise → Mid-Market Path

Workday, ServiceNow, Palantir. They built top-down for the largest buyers. Graduating mid-market eventually came knocking because they wanted to look like the big guys.

The Mid-Market First Trap

Tons of SaaS logos you’ve never heard of. They aimed straight at $20–50K ACVs, thinking it was safer. They ended up in deal cycles too long for their cash runway, but not valuable enough to sustain them.

But Is Mid-Market Ever Worth It?

Yes, but only in narrow cases:

• Vertical SaaS with sharp wedge. If you’re solving a hyper-specific compliance or workflow pain (say, SOC2 automation, restaurant scheduling), ramping mid-market might buy quickly.

• Bridge product. Tools that help SMBs become enterprise (think HRIS, security, finance) can ride the graduation curve.

• Roll-up opportunities. If you’re looking at M&A plays, mid-market customers can be sticky if you own their stack.

But as a starting target market? No. It’s a distraction.

What Founders Should Actually Do

1. Choose SMB or Enterprise as your core.

   • Design product, GTM, and org around that motion.

2. Let mid-market show up naturally.

   • Ramping mid-market is just SMBs that got bigger.

   • Graduating mid-market is just enterprise wannabes.

3. Don’t design for autopilot mid-market.

   • They don’t buy, they stall.

4. Don’t be fooled by TAM slides.

   • Mid-market TAM looks huge because there are tens of thousands of these companies. But only a fraction actually buy fast enough or big enough to sustain you.

Mid-market looks like a goldmine because it sits between two poles. But it’s really a mirage. The contracts are bigger than SMB, but the requirements are closer to enterprise.

Startups die in the desert chasing mid-market deals. The way to win is to pick a side. The mid-market will show up anyway.

That’s all for now. If you liked what you read, consider adding us to your address book or just hit reply—it helps with deliverability. If not, you can unsubscribe below.

We’re off for our usual early-August break. Back in late August.
Chief