What I’ve noticed beneath all the ambition, optimism, and occasionally unreasonable confidence is something familiar: fear.

What happens if this doesn’t work? What happens if I leave school, walk away from a good job, disappoint my family, or discover that I am not nearly as capable as I thought I was?

Worse, what happens if everyone else discovers that too?

That last one is particularly common. The good news is that almost nobody knows exactly what they are doing. Some people have simply raised enough money to sound more authoritative about it.

Fear triggers the usual human response: fight or flight. I’ve experienced both.

I Quit

I’d be lying if I said I never thought about quitting my jobs or my own startup. There were plenty of moments when walking away seemed appealing, and in hindsight, there were moments when I probably should have left sooner. I kept going.

Sometimes a mentor gave me the right advice. Sometimes I took a breath and convinced myself that there was one more move to make. Sometimes the thing bothering me seemed fixable, and sometimes stubbornness was simply difficult to distinguish from perseverance while I was living through it.

But as a founder, you cannot always wake up one morning, announce that the vibes are off, and leave. You have responsibilities to your team, your shareholders, your customers, and the people who trusted you enough to join the journey. That does not mean staying forever or tolerating every bad situation, but it does mean trying to fix what can be fixed and leaving responsibly when it cannot.

Dan and I had built something real. We had deployed spacecrafts, developed a working product, assembled an exceptional technical team, and created a level of automation that still makes me proud. We had also accumulated structural complications, difficult relationships, competing agendas, and the usual assortment of people convinced they could run the company better than the people who had built it.

Startup life: come for the innovation, stay for the governance.

The last satellite I worked on at SSL weighed more than 4,000 kilograms. A few years later, we were using textbooks, testing hardware on Cessnas and customer helicopters, and building spacecraft under 100 kilograms. The machines got smaller. The responsibility did not.

Toward the end, I had very little personally left to gain. I kept fighting because I wanted to preserve as much value as possible for the shareholders who had backed us, create continuity for the team, and protect the technology and infrastructure we had spent years building.

But that was near the end of the story. Let’s go back a decade.

One More Meeting

In the fall of 2017, Dan and I had been grinding away and bootstrapping the company. We had raised a little money from friends and family, put down deposits for missions, built prototypes, refined the roadmap, and incorporated every piece of investor feedback we could reasonably absorb without turning the company into something entirely different.

Still, nobody had funded us. We were beginning to ask the existential startup question: was this actually going to happen, or were we simply two unemployed aerospace engineers with an expensive hobby?

We met with a venture capitalist on Sand Hill Road whom we trusted for advice, and asked, more or less, whether we were finished. I’m paraphrasing, but he said:

“I don’t think you’ve reached your come-to-Jesus moment yet. If I thought it was over, I’d tell you.”

It was not a term sheet, and it was not especially actionable, but when you have been rejected enough times, mild encouragement begins to feel like institutional capital. So we kept going.

Dan attended one final pitch event on the East Coast, which led to a meeting with a partner from a Boston venture firm. They came to Dan’s garage and spent three hours diligencing us. There was interest, there was hope, and there was, naturally, no term sheet.

After several weeks of waiting and stewing, Dan and I agreed on a deadline. If we could get a term sheet by the end of 2017, we would continue; otherwise, we would shut the company down on December 31.

We flew to Boston on January 4th 2018, and received the term sheet. Technically late, emotionally acceptable.

It would take another four months to close the round and become income-producing members of our households again. By then, we had spent roughly 2 years without a paycheck, supported by our partners.

We were fortunate to be in a position where we could take that risk, but privilege does not eliminate fear. It does not make uncertainty less consuming or failure less embarrassing. It simply gives you more runway to stare at the ceiling at three in the morning.

The Safe, Boring Option

The startup itself emerged from another frustrating period.

I had been at Space Systems/Loral (SSL/MDA/MAXAR/Lanteris and now, Intuitive Machines) for almost five years. In 2015, I spent roughly eight months looking for another job. I did not have much of a network in the Bay Area, and most applications disappeared into corporate portals, presumably entering the same dimension as missing socks and investor follow-up emails.

A few interviews went nowhere. In others, I got the sense that, at 35, I had already crossed into aerospace middle age.

Eventually, I began to believe I might become a lifer at SSL, assuming I survived the next reorganization, acquisition, renaming, or round of layoffs. At the time, that frightened me.

I started talking to a few like-minded people within the company, Dan among them, and the idea of building something together began to take shape.

Until then, I had kept most of my job search within the Bay Area. I did not want to ask Malika to drop everything and move for an uncertain opportunity, especially after she had painstakingly built a small but vibrant network around her.

Eventually, I expanded my search to Seattle. Two opportunities emerged. One was genuinely exciting but became entangled in turmoil inside the company. The other progressed to the offer stage. It was stable, respectable, and boring.

Taking it would mean moving Malika away from the life and community she had built, starting again in a city where we knew almost nobody, and making a lateral career move for work I already suspected I would dislike.

The alternative was considerably less sensible: stay in the Bay Area, continue exploring the startup idea with Dan, work without a paycheck, and hope we could build something from scratch.

I chose the startup, not because I knew it would succeed, but because I was betting on myself.

Betting on Yourself

After 20 years of working, and after accumulating enough scars to qualify as either experienced or damaged, I believe that this is the more useful way to think about fight or flight.

It is not simply about whether you stay or leave.

Staying can be cowardice, and leaving can be courage.

You can remain in a job because you are afraid nobody else will hire you. You can keep a company alive because you cannot admit it is over. You can tolerate a terrible situation and call it resilience.

Sometimes staying somewhere miserable simply to prove that you can absorb more punishment is not courage. It is poor decision-making wearing a Patagonia vest.

Equally, you can walk away because you believe you are capable of something better. You can shut down an idea because the evidence has changed, or leave the thing you built because remaining would require abandoning what you believe.

The question is not simply whether you stayed or left. The question is whether you made the decision from fear, or from belief in yourself.

When you genuinely bet on yourself, you fight for yourself. You fight to improve the situation, create another option, find the next meeting, make the next call, build the next prototype, or endure another rejection. Occasionally, you fight by leaving.

There were times when I felt as though I was running away from SSL, or later from OSK, but that was not really what I was doing. I was fighting for what I believed was the right thing to do, for the life I wanted, and for what I thought success could look like.

What Did Success Look Like?

Hindsight is useful because it brings context and perspective. You accumulate exposure, collect battle scars, and eventually discover that most decisions were neither as brilliant nor as catastrophic as they felt at the time.

Failure is part of life. You will make mistakes, misjudge people, take risks that do not work, and occasionally walk confidently into a wall that several people had already pointed out to you.

Even the most risk-averse person fails. They simply tend to fail more conservatively.

So what did success look like for me?

One interpretation is straightforward. I started a company, built something from scratch, launched satellites, sold a product, saw some remarkable things, accumulated a decade of stories, and walked away with nothing.

That is one way to look at it.

Another is that I bet on myself and discovered what I was capable of. I built things I had once only imagined, worked with extraordinary people, survived situations that at the time felt unsurvivable, and learned how to raise money, launch spacecraft, build teams, sell products, make painful decisions, and begin again.

Of course, a decade like that leaves evidence. Not just satellites, products, pitch decks, and battle scars, but a photographic record of what building a company, raising children, surviving COVID, and repeatedly staring into the abyss can do to a person. The photograph below probably captures the journey better than a cap table ever could.

The company grew. The children grew. My perspective grew.

My hair chose a different trajectory.

Apparently, the company was not the only thing experiencing burn.

Looking at those photos, success did not unfold exactly as I had expected. There were satellites, fundraising, sleepless nights, children, stress, family, small victories, painful failures, and the gradual realization that my hairline had apparently adopted its own exit strategy.

Still, I would not reduce those years to what I walked away with. The journey changed me, tested me, and showed me what I could build and endure.

Success did not look exactly as I had expected. It rarely does.

The Founders in the Room

That brings me back to the founders I’ve been meeting recently.

The world provides plenty of reasons to be pessimistic: wars, political dysfunction, climate anxiety, the day-to-day grind, and the increasingly popular narrative that artificial intelligence will take everyone’s jobs before moving on to humanity itself.

Then you enter a room and ask someone what they are building, and their face changes.

They begin describing a problem they cannot stop thinking about. Their excitement spills into the conversation, and for a few minutes the world feels less doomed, not because the problems have disappeared, but because someone has decided to do something. That energy is infectious.

Many of these people are first-time founders who are weighing school, careers, finances, relationships, children, immigration, family expectations, and the possibility that the idea simply will not work. Some are still searching for the right problem, while others have found one and are trying to determine whether the solution exists outside a slide deck.

The passion required to build something from scratch, challenge the status quo, and solve extremely difficult problems is not for the faint-hearted. By taking the first step, they have already given themselves an opportunity.

Perhaps they have funding to explore. Perhaps they have only a few months of savings and a suspiciously supportive spouse. Either way, fear is there.

My advice to them is the same advice I wish I had understood earlier: Bet on yourself.

Not blindly, not arrogantly, and not because every idea deserves to survive.

Bet on your ability to learn, adapt, recover, and find another path. Stop assuming that a frightening situation means you are trapped. The situation matters, but your response to it matters more.

Fight to make it better. Fight to create alternatives. Fight for the people who depend on you, and fight for what you believe the outcome should be.

Whether you stay or leave, choosing a harder direction than the one others consider sensible is not defeat. Walking away from something that makes perfect sense to everyone else is not flight. It is choosing your own path, on your own terms, and fighting for the life you believe you should have.

AI disclosure: The thoughts, stories, and scars are mine. ChatGPT—my co-pilot, editor, and occasional over-user of em dashes—helps me turn the raw material into something more entertaining and easier to digest. Pronouns: it/its (determined by itself), pending further developments.

Artemis is back in the cultural bloodstream. Space stocks have decided gravity is optional. Orbital data center companies are raising eye-popping amounts of capital. Growth companies are being valued like they already have docking ports on the Moon. And somewhere in all of this, we are all pretending to be calm, rational adults.

We are not.

The industry is excited again. That part is good. I want that. I want more ambition in space, not less. I want Starship to work. I want launch to become boring infrastructure. I want the price of putting mass into orbit to collapse so hard that future generations look at today’s launch market the way we look at dial-up internet.

But wanting something to happen is different from seeing the mechanism that makes it happen.

This essay is about the first mile of space: getting mass off Earth.

The next ones are about the middle miles, moving things once they are already in space, and the last mile, actually delivering value where it is needed.

Because right now, everyone wants to talk about the destination. Lunar industry. Mars logistics. Orbital data centers. Sovereign constellations. AI in space. Manufacturing in microgravity. Pick your preferred flavor of orbital caffeine.

But before all of that, there is the first mile.

And the first mile is still expensive.

Not because the technology is impossible. Not because SpaceX has not changed the game. It has. Obviously. Aggressively. Almost rudely.

But because low internal cost is not the same thing as low market price.

And that distinction matters.

The dream

The dream is simple.

Launch gets cheap. Really cheap. Not “slightly less painful procurement process” cheap. Not “we negotiated a multi-launch discount and got a nice PDF from business development” cheap.

I mean cheap cheap.

Launch gets cheap. Really cheap. Not “slightly less painful procurement process” cheap. Not “we negotiated a multi-launch discount and got a nice PDF from business development” cheap.

I mean cheap cheap.

The type of cheap where companies stop designing spacecraft around scarcity. And yes, we are less constrained than we were 10 years ago. Satellites are getting heavier. Payloads are getting bigger. Every extra kilogram no longer requires a candlelight vigil.

But the hesitation is still there.

Mass margins still feel existential. Hardware teams still hold their breath. Power budgets still get stupid. Structures still get anorexic. And someone in a design review saying, “What if we just made it bigger?” still risks being escorted out of the building.

This is the world people imagine when they talk about Starship. This is the world behind the $100/kg launch-cost charts. This is the world ARK and others love to show in infographics that make the future look like it arrives via clean exponential curves.

And look, I want the chart to be right. I really do.

I want launch to become so cheap that the space industry finally stops behaving like a bespoke science fair with defense contracts and starts behaving like infrastructure.

But then the annoying operator part of my brain kicks in.

The part that has seen aerospace schedules. The part that has watched customers say they want revolutionary change until procurement asks for a flight heritage chart. The part that has lived through enough “this time is different” cycles to know that sometimes this time is different, but usually the PowerPoint is just better.

So the question is not: can SpaceX get its internal cost down?

I think it has.

The question is: does the rest of the market get that price?

And if so, when?

Internal cost is not market price

A reusable Falcon 9 can put roughly 17,500 kg into Low Earth Orbit. So if we take the dream seriously, and say launch should be around $1,000/kg, then a full Falcon 9 should cost around $17.5M.

That would be lovely.

Please, someone call me when I can buy one at that price. I will not be difficult. I will bring snacks.

But that is not the going market rate.

The advertised price for an entire Falcon 9 is closer to $70M. Based on public estimates, Globalstar paid around $64M for an entire Falcon 9, and AST SpaceMobile appears to have paid around $50M per dedicated Falcon 9 launch through a multi-launch deal. That puts dedicated launch somewhere around $2,850/kg to $4,000/kg depending on the mission, deal structure, and how generous we are being with assumptions.

Rideshare is advertised at around $7,500/kg.

So yes, SpaceX may be achieving incredible internal economics. But commercial customers are not buying launch at internal transfer pricing.

That is not a criticism. That is capitalism. Annoying, effective, undefeated.

SpaceX is not a charity. It is not obligated to pass internal efficiency directly to the market, especially when demand exists at higher prices and competition is still catching up.

And if SpaceX goes public, the incentives become even more obvious. Investors do not care whether launch is cheap in some spiritually uplifting, industry-enabling sense. They care whether the company they own is making money. Cheap launch is great for the space economy. High-margin launch is great for shareholders.

Those two ideas may overlap but aren’t twins.

So the real question becomes: what would force prices down?

Two things usually do that.

Demand and competition.

The 2025 launch manifest tells the story

In 2025, SpaceX flew 165 orbital missions.

Here is the rough breakdown:

That means roughly three-quarters of SpaceX’s launches were internal.

Not commercial customers.

Not the broader space economy.

Not an army of venture-backed startups heroically buying mass to orbit.

Starlink.

SpaceX is not just the dominant launch provider. It is the dominant launch customer.

That is the part people skip over when they talk about falling launch costs. SpaceX’s cadence is not primarily being driven by the market. It is being driven by SpaceX’s own vertically integrated demand.

That is brilliant. It is also not the same thing as broad-based industry demand.

If most of your launch volume comes from your own satellite network, then the economics of that launch system are fundamentally tied to your own internal business case.

In other words, Starlink can get Starlink pricing because Starlink is feeding the machine.

Everyone else is still standing at the counter, looking at the menu.

Competition is not exactly breathing down their neck

Surely other launchers will force prices down.

Maybe. Eventually. I hope so.

But right now? Case and point…..

But, here is the competitive picture:

Rocket Lab is the closest thing to a high-cadence Western launch alternative, and Electron is a genuinely impressive vehicle. But it is not a Falcon 9 competitor on price per kilogram. It is selling dedicated access, schedule control, and mission flexibility. Those things matter. They are valuable.

They are also expensive.

And yes, before someone writes a very thoughtful comment explaining that Electron is not trying to be Falcon 9, I know. That is the point. Different system. Different customer. Different economics.

New Glenn is the most interesting one on this list. If it works at cadence, it can change the pricing conversation, especially for Amazon Kuiper / Leo and larger payloads.

But again, aerospace has an annoying habit of reminding everyone that rockets are not slideware.

Blue Origin’s recent New Glenn pad explosion is a good example. First and most importantly, I am glad everyone is safe. That is the only thing that actually matters in the moment.

But also, I would be lying if I said the operator in me did not raise an eyebrow.

Back in 2015, someone from Blue Origin once told me, with a straight face, that they did not really think anyone was in competition yet because no one else had the vehicles. They said Blue was moving slowly because they wanted to do it right, avoid explosions, and not rush into the messy failure cycle others were living through.

Fair enough.

Also: oof.

That sentence has not aged like fine wine. It aged more like milk left on the launch pad during a hot fire test.

And to be clear, this is not a dunk on Blue Origin. Rockets are hard. Explosions happen. SpaceX had spectacular failures too. Falcon 9 did not become Falcon 9 by emerging from the ocean fully reusable and wearing sunglasses.

But this is exactly the point.

Competition is not a press release. It is not capacity on a website. It is not a beautiful render of a rocket next to an even more beautiful render of a price curve.

Competition is cadence, reliability, customer trust, manufacturing rate, launch rate, recovery rate, re-flight rate, insurance comfort, and the deeply unsexy ability to do the same hard thing again and again without turning the Cape into a fireworks show.

That takes time.

So yes, New Glenn may matter. It probably will. But it is not yet forcing broad commercial launch pricing down today.

Which brings us back to the boring, annoying, unavoidable question.

Where is the demand?

It’s the volume of satellites, stupid

If launch prices are going to collapse, someone has to fill the rockets.

Not once. Not for a demo. Not for a press release with dramatic lighting.

Again and again and again.

So let’s look at who actually has satellites in orbit.

Starlink has more satellites than everyone else combined.

It has roughly 16x more satellites than OneWeb, its next major Western commercial LEO constellation competitor.

That is not a market. That is a gravitational field.

Now, the obvious counterargument is that everyone is launching more satellites. The industry is growing. More commercial companies are coming. More sovereign constellations are being built. More defense architectures are shifting toward proliferated LEO.

All true.

But let’s actually look at the launch numbers.

The pattern is not subtle.

Starlink dominates.

Kuiper is finally showing up, but it is still early. OneWeb built out a meaningful fleet, but not at Starlink scale. The rest of the market is growing, but not nearly fast enough to create broad launch-price compression by itself.

So when people say satellite launches are booming, the more precise statement is:

Starlink is booming.

A few others are trying to boom.

Everyone else is still somewhere between “strategic growth market” and “please admire our total addressable market slide.”

And that is not unfortunate so much as revealing.

SpaceX’s TAM is allegedly around $25+ trillion, a number so large it stops sounding like a market and starts sounding like a rounding error in the GDP of a Type II civilization.

But the funniest part is not the size of the number.

It is where the number lives.

SpaceX built the company on launch. Falcon 9 was the wedge. Reusability was the unlock. Launch was the thing that broke the industry’s brain and forced everyone else to explain why their rocket was different, special, sovereign, responsive, dedicated, or whatever adjective procurement had approved that quarter.

But in SpaceX’s own TAM framing, launch is not the big prize.

It is the small bucket.

That should be a warning label for the entire launch sector. Launch may be the hardest technical problem in the stack, the fire-breathing, metal-bending miracle that makes everything else possible. And yet, in the TAM framing, it is the “worst” business bucket. Everyone wants launch to get to $100/kg. But if launch is already the smallest opportunity, and margins matter most, why would the market race to give away the hardest part of the stack? This is not a moral argument. It is just business gravity being rude again.

The massive opportunity is not just getting things to space. It is what happens when customers start paying to use space-based infrastructure once it exists.

Connectivity. Compute. Data. Sensing. Communications. Defense. Positioning. Maybe one day power, manufacturing, logistics, and whatever else we decide to put above the atmosphere because Earth apparently became too easy.

That is the actual shift.

Launch is the first mile. It matters enormously. But the value is increasingly in the usage layer. The customer does not really want a rocket. The customer wants the thing the space infrastructure enables.

Which is also why SpaceX is so hard to compare against the rest of the industry.

It did not just make launch cheaper.

It used launch to build the thing customers actually pay to use.

But what about tonnage?

Now, before everyone starts yelling about orbital data centers, mass to orbit, tons to orbit, Starship, and the glorious future of launching entire server farms into space, let’s look at this another way.

Instead of counting satellites, let’s normalize SpaceX launches by reusable Falcon 9 capacity.

Assume each Falcon 9 launch represents about 17.4 metric tons of reusable launch capacity to LEO. This is not actual delivered payload mass. It is capacity allocation. It answers the question: where was SpaceX pointing its launch machine?

Again, same story.

Starlink is not just dominating satellite count. It is dominating the tonnage allocation.

The space industry does not have a launch supply problem alone. It has a demand concentration problem.

One company has figured out how to create a real end-market business that requires thousands of satellites. Everyone else is still trying to find enough customers, contracts, capital, and courage to justify similar mass.

And that is the part that should make us both optimistic and nervous.

The $100/kg world needs customers, not just rockets

I am doubtful, at least right now, that commercial launch pricing gets below $1,000/kg in the next decade, let alone $100/kg. Not because the rocket cannot theoretically do it. Maybe it can.

But if SpaceX goes public, the incentives become painfully obvious. Investors do not care whether launch is cheap in some spiritually uplifting, industry-enabling sense. They care whether the company they own is making money. Cheap launch is great for the space economy. High-margin launch is great for shareholders. Those two ideas may overlap, but they are not twins.

The rocket is necessary.

It is not sufficient.

To get there, the industry needs more businesses that behave like Starlink. Not necessarily broadband. Not necessarily consumer internet. But businesses with massive real demand, recurring revenue, and a reason to launch thousands of tons over time.

That is why orbital data centers are interesting.

And also why the current hype around them makes me twitch.

ODCs could be real. Power, cooling, latency segmentation, sovereign compute, AI demand, launch availability, there is a version of this future that makes sense.

But we should not confuse “interesting” with “imminent industrial scale.”

Starcloud is not going to magically put a 1 GW, 12 square kilometer footprint in orbit next quarter because someone made a nice render. Their deal with Starlink terminals tells you what this really is: phased deployment, early revenue milestones, pathfinding, customer discovery, and a lot of capital expenditure.

That is not a criticism. That is how hard things get built.

But 20+ spacecraft to reach early milestones is not enough to bend the global launch curve. Even with $100M raised, or $300M raised, or whatever the next rumored round is, this is still probably a multi-year climb before meaningful operational capacity exists.

K2 has an impressive vision and an order book in the public domain of a few dozen satellites. York is one of the more established high-throughput manufacturers, with stated capacity to build hundreds of satellites per year. But capacity is not the same as launched mass. Factories do not create launch demand by existing. Customers do.

And customers, annoying little creatures that they are, usually want proof, revenue, reliability, and a reason to spend money.

We have seen this movie before

This feels like the second coming of the modern space industry.

The first one was in the mid-2010s.

Back then, it was launch. It was Earth observation. It was OneWeb. Then Starlink. Later Amazon Kuiper. Everyone had a constellation. Everyone had a market. Everyone had a chart that went up and to the right, because charts that go down and to the left apparently do not raise Series B rounds.

Some of those companies survived. Some went public through SPACs and then spent half a decade getting punched in the face by reality. Planet, Rocket Lab, BlackSky all had to endure the market rediscovering that space is hard, customers are slow, and public markets are not especially sentimental.

Now we are seeing another resurgence.

Artemis is back. Defense demand is the driver. Sovereign space is real. SpaceX’s potential IPO is lifting everything with a ticker symbol that can plausibly say “space” in an investor deck. Orbital data centers are the new shiny object. AI has made compute feel infinite and urgent. The Moon is back on the menu.

I love this.

I also remember the last cycle and that is the tension.

The idealist in me is excited because this is exactly how frontier industries begin. Capital rushes in. Some of it is early. Some of it funds things that look ridiculous until they suddenly look obvious.

The pessimist in me, who is unfortunately also the guy who has had to build hardware, asks the less fun questions.

How quickly can these companies execute?

How much capital do they actually need?

How many customers exist outside government?

How many of these business models require launch pricing that does not exist yet?

How many investors will stay patient long enough for the infrastructure layer to mature?

And how many companies are quietly assuming that Starship will arrive, work, scale, and discount the market before their cash balance asks uncomfortable questions?

So how do launch prices actually come down?

I think the answer is simple, but not easy.

We need more Starlink-like businesses.

Not copies of Starlink. The world does not need seventeen broadband constellations all discovering customer acquisition costs at orbital velocity.

But we need businesses with the same structural property:

They generate enough real revenue on Earth to justify massive infrastructure in space.

That is what makes Starlink different. It is not just a satellite constellation. It is a customer-facing business with enormous scale. SpaceX reportedly spent many billions of dollars designing, building, launching, and operating that network. Who else in the industry has raised and deployed that kind of capital against a commercial space business?

That is the bar.

Orbital data centers might be one. Lunar industry might be another. Defense proliferated architectures are already one, though government demand does not always create commercial price compression in the way people hope. In-space logistics, servicing, manufacturing, energy, communications, and compute could all become part of this.

But they need to become real businesses.

Not just payloads.

Not just demos.

Not just “we signed an MOU” confetti.

Real businesses. With customers. With recurring demand. With enough mass moving to orbit that launch providers actually have to compete for volume.

That is when prices come down.

Not when someone makes a chart.

I am still rooting for the cheap launch future

This is the part I want to be clear about.

I am not bearish on space.

I am not rooting against Starship. I am not dismissing ODCs. I am not saying the industry is doomed because the spreadsheet is inconvenient.

I am saying the opposite.

I want the industry to become real enough that cheap launch becomes inevitable.

I want launch to become infrastructure. I want $100/kg. I want a world where the cost of getting to orbit is no longer the first thing that kills the idea. I want space companies to stop designing around scarcity and start designing around abundance.

But hope is not a launch manifest.

The industry does not get cheaper because we put a lower number on a slide. It gets cheaper when enough real customers buy enough real capacity often enough that rockets become infrastructure instead of events.

That is the first mile.

And right now, the first mile is still mostly being paved by one company launching its own network.

That does not make the future impossible.

It just means the next great unlock in space may not be the rocket.

It will be the customer.

Home operations are some of the hardest operations on Earth. Titles mean nothing there. My title may be “Dad,” but I know exactly where I rank in the command structure. Somewhere far below “Mom,” “Mimi,” and occasionally the dog.

I may have multiple engineering degrees, but I still get ordered to stir the daal and make sure the dog is fed. I may have managed spacecraft timelines and mission operations, but my 4-year-old reminds me:

“Make sure you tell the teachers Mimi will pick me up at 3:30pm today. Keep that in your head.”

Space may be the final frontier for humanity.

A home is unconquerable territory.

Image 1 - The belligerents

The Battle!

And nowhere is this more obvious than sleep.

Sleep training. Sleep habits. Bedtime routines. Enter: el diablo.

Now, to be fair, our daughter had a rough start to life. But thankfully, today she thrives regardless of what life throws at her. Fierce, funny, stubborn, unstoppable.

And for a brief moment, we thought we had won.

My wife and I believed in sleep training. We did it with both kids. Armed with field manuals, routines, schedules, and what felt like tactical superiority, we went into battle.

There were early wins. Some retreats. Some counteroffensives. But eventually both kids developed great sleep habits.

My son? A dream. Dinner, fruit, brushing teeth, reading, audiobook, lights out. His head hits the pillow and he’s unconscious before gravity fully takes effect. Somewhere deep in my DNA, the general in me feels proud.

Then came our daughter.

At first, it seemed easier. We knew the process now. We knew the terrain. We knew the enemy. Or so we thought.

Then came the transition from crib to bed.

Freedom

No walls. No barriers. No containment system.

Now my memory is a little hazy here. Mostly because the following years involved sleep deprivation severe enough to qualify as psychological warfare.

But sometime during that first week, around 2 or 3am, I woke up to something cold touching my back.

Tiny feet.

I realized our family had somehow formed an “H” configuration in bed. My wife and I vertical. My daughter horizontal between us. Her head pressed against my wife. Her freezing feet jammed into my spine.

Freud would probably say the “H” stood for “Help.”

I was stunned. The field manuals had not prepared me for this kind of combat.

Naturally, I attempted a counteroffensive. I tried to straighten her out and carry her back to bed.

That’s when she deployed the weapon of mass destruction.

She wrapped her tiny arms around me, buried her nose into my neck, sighed gently, and snuggled closer.

And I surrendered immediately.

There was no defending against this.

Tomorrow, I thought. Tomorrow I retake the hill.

Tomorrow never came.

Every night became a battle.

We marched uphill with explanations:
“You need sleep.”
“You have your own bed.”
“Please don’t come into our room.”

None of it worked.

At first, her attacks were loud. Tiny footsteps stomping down the hallway. Door flying open. Immediate cannonball into the middle of the bed.

Later, she evolved.

Now she operated like Seal Team 6.

Silent infiltration.

Door opens one inch.
Pause.
Enemy assessment.
Slow advance.
Blanket insertion.
Immediate fake sleep upon detection.

We escalated too.

Threats were made:
“If you come in again, I’m carrying you back.”

But the threats were never enforced.

I believe the term these days is TACO.

The enemy adapted quickly.

One night I decided I would finally hold the line and carry her back no matter what.

That night she bypassed me entirely and snuggled directly into my wife.

My counterattack was neutralized before it began.

“Let her sleep,” my wife whispered.

The alliance had fractured.

Then I noticed another tactic.

The Break that never was

Eira sleeps hot. Like, thermonuclear hot.

At bedtime she would kick off her blanket because she was “too warm.” By 2am she would arrive in our room absolutely freezing.

This, I realized, was deliberate strategy.

Radiate heat into open space. Reduce thermal protection. Force emergency migration.

So I adapted.

After she fell asleep, I would sneak into her room and tuck her blanket back over her. Behind enemy lines. Covert thermal operations.

It worked.

For two glorious nights, I slept like a free man.

Then she recalibrated.

To this day, I have not cracked that encryption code.

Eventually, we activated our final remaining weapon.

A Barbie.

Thirty consecutive days sleeping in her own bed would earn her the Barbie.

We felt hopeful again.

The next morning after sleeping in her own bed, she taunted us:

“Did you guys get good sleep?”
“Do you feel rested?”
“Aren’t you proud of me?”

Then she would immediately reset the counter by appearing back in our bed that night.

Day 1… reset.
Day 1… reset.
Day 1… reset.

We got desperate.

Thirty days became seven.
Seven became five.
Five became weekdays only.

And then finally:

Friday, May 15th, 2026.

After more than two years of nightly warfare, she completed five straight weekdays in her own bed.

The Barbie was hers.

We had won.

Victory!

And then a horrifying realization hit me:

Uh oh.

We have no more Barbies.
No more Legos.
No more leverage.
The magazines are empty.

What happens now?

That night, the kids slept with mom.

I slept in her bed.

While everyone is swooning over Earthrise, this is my favorite image. The crescent of Earth from the Moon. A mirror we don’t look into often enough.

It was the summer of 1988. I was five.

Standing outside the terminal, staring at this strange machine; wings, tail, four engines…I fell in love with flight before I even understood it. I was flying from Delhi to San Francisco. Middle seat. My older cousin had the window.

I fell asleep before takeoff.

He woke me up soon after taking off.
“We’re flying,” he said.

I looked around, confused, then asked:

“Did we fly over the moon yet? Is it below us?”

~38 years later.

April 10th, 2026.
“HOUSTON - Integrity, splashdown.”

Artemis II comes home.

And just like that, I’m five again.

For all the criticism of NASA: analysis paralysis, SLS debates, the usual “why are we spending money on this?” chorus, these 10 days showed exactly why the institution still matters.

For 10 days, we forgot the noise. It matters more now than ever before.

The World vs. The Moon

As a dad, it’s hard not to feel like the world is… unraveling.

Everything feels polarized. “Us vs. them”. Except no one can even define who those sides are anymore. It shifts daily. It feels like we’re one bad decision away from something irreversible. You start wondering what kind of a world you’re leaving your kids.

And then something like Artemis happens.

And for one rare moment, it feels like we’re doing something right.

Because the reality is we’ve gotten very good at funding how to destroy things, and far less consistent at funding how to build beyond them.

More goes into preparing for conflict than expanding what’s possible. We need to figure out how to move humanity forward.

I’ve taken my shots at that system, but for all the criticism of the fighting machine, there’s an uncomfortable truth:

Defense and space are deeply intertwined.

Whether it’s Earth observation, communications, or broader space-based infrastructure defense has been, and continues to be, the primary economic engine behind it.

Which is exactly why Artemis matters.

Because beyond SLS vs. Starship vs. New Glenn, the real objective is simple:

Get off the planet.
Get beyond LEO.
Start building elsewhere.

And before someone says, “Why not fix Earth first?”

We can and should do both. Its a false tradeoff that needs to die.

To be curious is to be human.

If curiosity needed approval:

• Galileo doesn’t look up

• The Americas aren’t discovered

• The Earth stays flat (and yes… some still think it is)

Exploration isn’t a distraction from progress.

It’s a driver of it.

The Questions That Matter

I was sitting with a group of parents during the mission.

“Why go to the Moon?”
“Why Mars?”

“Did we even go the first time?”
“That flag was fluttering…”

In case you’re reading and wondering the same, the American flag did not flutter in the wind on the Moon. It appeared to be waving because a horizontal rod holding it up did not fully extend, creating creases and a rippled appearance, while motion from planting it lasted longer in the vacuum.

Meanwhile, my kids:

“Did the rocket make it?”
“Is it back yet?”
“Where do astronauts poop?”
“Why is the toilet broken?”

That’s it.

That’s the whole thing.

Curiosity.

The Moments That Shape You

There are moments in life that quietly shape everything that comes after.

For me, one of them was being woken up at 3am by my dad’s best friend.

“Come outside.”

Comet Hale-Bopp.

A once-in-a-lifetime moment.

He was the kind of person who didn’t lecture you into curiosity, he just nudged you toward it. Star Trek, computers, space… he made it all feel within reach.

He passed away during COVID.
Raj Uncle, I still think about that night, you sparked more than you knew. Wherever you are, I hope you’re exploring…

Aerospace Wasn’t Around Me, I Chose It

This wasn’t something that just happened to be around me.

At some point, aerospace became the thing. Not a phase. Not an interest.

The thing.

Doodling airplanes, the shuttle, and rockets constantly.
Being in Dr. Maughmer’s class and finally understanding aerodynamics.
Working with my advisor Dr. Gandhi, designing and building structures that morph.
Trying to model systems that didn’t quite work.
Attempting wind turbine blades inspired by dragonfly wings.
Trying to understand multiphase flows, converting flue gas to cement with seawater.

Then actually building.

Fueling spacecraft.
Launching payloads.
Working on systems that, even today, feel slightly unreal.

And yes, failing.

Not small failures.

Failures that hit at the worst possible moments.
Failures that forced me to question everything, including myself.
Failures where I genuinely wondered if it was worth continuing.

And in those moments, there wasn’t someone there pulling me out of it.

It was just me and airplanes, shuttles, rockets, flight...

Every time, all I could do was look up at the sky, dream, and stand back up.

NASA had its missions. There were milestones.
It helped me get back up and dream.

What Humans Do

We’re entering a world shaped by AI, robotics, automation.

So what do humans do?

On a planet that’s overcrowded, heating up, economically uneven;

What do we do?

We do what we’ve always done.

We build.

Voyager.
The ISS.
The Burj Khalifa.
The Pyramids.
The Taj Mahal.
USS Gerald Ford.
New York City….

I can go on but we build things that shouldn’t exist, until they do.

And that instinct shouldn’t stop in or just around Earth it should extend beyond it.

That’s what Artemis represents.

Continuation. Expansion. Survival.

Because one day whether it’s climate, conflict, or something coming at us from space; this won’t be philosophical.

It’ll be practical.

And I’d rather we have options.

Earth’s Phases on the Moon - Earthrise, Earthset, Crescent Earth and Eclipse! Image courtesy: NASA

The 10 Days

I spent those 10 days watching Artemis with my kids.

The Moon.
The trajectory.
The return.

The sparkle in their eyes.

And for those 10 days we forgot everything else happening in the world.

Even the random radio chatter between Houston and Integrity that freaked Malika out in the middle of the night.

What NASA Actually Does

NASA does impossible things.

It dares mighty things.

Gemini. Mercury. Apollo.
Skylab. Shuttle. Voyager.
New Horizons. Spirit. Opportunity. Curiosity.
Hubble. James Webb. Letting us look back in time.

And not just space:

Breaking the sound barrier with the X-1.
Pushing hypersonics with the X-15.
Reimagining flight with the X-36 and X-59.
Shaping aviation itself through NACA airfoils.

But beyond all of that beyond the missions, the hardware, the milestones

NASA does something else.

It makes people curious.
It makes people smile.
It creates moments of vulnerability, where things could go wrong, and pulls everyone in.

That’s the magic.

Everything else?

Noise.

Let other agencies handle that.

NASA does the hard, improbable, sometimes irrational things and in doing so,

gives us reasons to look up.

Do what you do best.

Get to the Moon.
Stay there.
Build there.

Then go to Mars.
And beyond.

Because one day, my kids might ask again:

“Did we fly over the moon yet?”

And I hope the answer isn’t just yes.

I hope it’s -

We stayed.

We call it “hard” as if that’s the problem.

It’s not.

The problem is we stop there.

I’ve spent a good chunk of my career building companies where hardware was a major component. Satellites, sensors, things that fly, things that break expensively; The usual suspects.

And over time, after enough conversations with founders, engineers, and operators, a pattern starts to emerge:

A lot of these companies don’t actually have a product.

They have a thing.

And they’re desperately hoping the world will figure out what to do with it.

At some point, it stops being innovation and starts being a very expensive science project.

The Very Expensive Identity Crisis

I recently had coffee with a friend at a late-stage company building unmanned vessels.

On paper: incredible tech.
In reality: mild existential dread.

Why?

Because they couldn’t provide a complete solution. Not because the tech wasn’t good, but because they were anchored to their identity:

“We are a company that builds unmanned vessels.”

That sounds fine… until the market asks for something slightly different.

What they should have been:

“We solve X problem and sometimes that includes unmanned vessels.”

That one shift unlocks everything:

• You integrate other platforms

• You expand your solution space

• You actually solve the problem

Otherwise?

You’re just building a very sophisticated widget.

The Lithium Delusion

This shows up everywhere, not just in startups.

I heard an anecdote about executives in mining who were frustrated their companies weren’t valued like Tesla or Apple.

Their argument:

“Without lithium, none of those companies exist.”

Correct.

Also irrelevant.

Raw lithium is useless to 99.999% of the world.
A battery, a car, a phone; And how they’re used, how people interact with them — those are the solutions.

That’s where value accrues.

Let’s put numbers to it:

• Albemarle (one of the largest lithium producers): ~$20B market cap

• Tesla / Apple: >$1T

Same value chain. Completely different outcomes.

Because one sells inputs.
The other sells solutions people actually care about.

Tesla Was Never Just a Car Company (and That’s the Point)

Ask most people what Tesla is, and you’ll hear:

“An electric car company.”

That’s like calling Amazon a bookstore.

Tesla didn’t win because it built a better car.
It won because it solved the entire problem of owning an EV:

• Range → solved with battery + infrastructure

• Charging → solved with Supercharger network

• Experience → solved with software

• Anxiety → solved with… everything above

The car was just the entry point.

The product was the system.

Image: screen grabs: Silver peak mine, Tesla Car lineup, Tesla Autopilot, Tesla Superchargers

Starlink Didn’t Build Satellites. It Built Internet

Same story in space.

Ask someone what Starlink does, and you’ll hear:

“Satellites.”

But Starlink didn’t set out to launch satellites.
It set out to deliver internet everywhere.

Satellites were just one piece.

So were:

• User terminals (millions shipped, not thousands)

• Routers, cabling, ground infrastructure

• Software orchestration

• Launch cadence

The real product isn’t in orbit.

It’s in your house.

Image: Screen grab from https://starlink.com/stories

Apple: The Best “Hardware Company” That Isn’t One

Perhaps the best example of this and arguably the greatest “hardware company” ever, is Apple.

On the surface, it looks obvious:
Phones, tablets, computers, watches, headsets… hardware everywhere.

Image: The Apple armory, https://www.apple.com/store

But calling Apple a hardware company is like calling a five-star restaurant a kitchen.

Technically true. Completely misses the point.

Take the iPhone.

Yes, it has massive market share, but no one is buying an iPhone just for the hardware.

They’re buying:

• The ecosystem

• The operating system

• The App Store

• The seamless interaction across devices

• The experience

The hardware is just the entry ticket.

What Apple actually sells is control, over the experience, the distribution, and the user.

The iPhone without iOS, without apps, without the ecosystem… is just a very expensive paperweight.

(And yes, AirPods switching between devices still occasionally feels like a beta feature, but we’ll let that slide.)

The EO Industry’s Favorite Lie

This is where it hits closest to home.

In Earth Observation, I’ve heard every version of the same story:

• “We’re a data company”

• “We’re a satellite company”

• “We’re a platform company”

• “We’re verticalizing now”

Translation:

We’re still figuring out what we actually are.

Meanwhile, decks are full of:

• “This sensor can solve X, Y, Z, and 12 other industries”

• “Massive TAM”

• “Revolutionary modality”

I’ve talked about this before and I’ll ask the question again, what’s missing?

A single, clearly defined problem with a buyer who will pay to solve it.

What Actually Happens Instead

If you watch closely, the market quietly forces the answer.

Take a certain public EO company (you know the one):

They started as:

• Data provider → then

• Government contractor → then

• Satellite-as-a-service → now

• Somewhere between infrastructure + sovereign capability

Why?

Because countries didn’t want data.
They wanted control.

So the company evolved, not by design, but by necessity, into a solution provider.

And now you see others following the same path.

The Part I Lived…..

I’ve lived this at OSK.

We had moments where we found real product-market fit.

Not once. Not twice.

Four separate times.

And each time, instead of going all-in…

We hesitated.

Expanded sideways.
Chased new things.
Stayed too attached to what we thought we were.

Why?

For another time. It’ll probably sound like a few seasons of Silicon Valley.

The Takeaway

You can build:

• The best satellite

• The best sensor

• The best hardware anyone has ever seen

And you’ll get some sales.

But if you define yourself by the thing you built…

You’re choosing to be:

• A component

• An input

• A supplier

Instead of the thing that actually matters:

The solution.

It’s harder.
It’s messier.
It forces focus.

But it’s also where all the value is created.

Closing Line

Hardware doesn’t fail because it’s hard.
It fails because too many companies stop there.

And technically, it’s true.

But when you build something from scratch, the lines between work and life don’t stay neat for very long. They blur. Slowly at first. Then completely.

Time stops being compartmentalized. There’s no clean separation between office and home, strategy and stress, ambition and identity. Your startup doesn’t sit in a folder labeled “professional.” It leaks into everything.

You chase the highs; The first working sensor, the first customer, the first wire into the bank account that makes you feel slightly less delusional.

But the cost isn’t just financial.

It’s time.
It’s emotional bandwidth.
It’s presence.

Not because anyone forces it, but because building demands it.

The Beginning, The First Institutional Check!

As Dan and I were closing our seed round, my son was born two months early and spent his first weeks in the NICU.

No founder manual explains how to switch between hospital monitors and investor updates in the same afternoon.

On April 6, 2018, $4.5M wired in.

We sat on the rooftop of our not-so-glamorous Sacramento Street office. It was one of those perfect San Francisco days. We allowed ourselves one drink. A quiet exhale. A sense that maybe this improbable thing had a chance.

That evening, I was back in the NICU learning how to feed my son through a bottle.

He stopped breathing.

Alarms. Nurses rushing in. Blue lips.

That morning was a capital milestone.
That evening was perspective.

Not because the company was “personal.”
But because the boundaries between life and work had already dissolved.

That pattern would repeat itself in different ways over the years.

The Good Investors vs. Inexperienced Ones

There’s something I learned over time.

Capital can be abundant.

Judgment is not.

A seasoned investor understands that companies are built by humans operating under pressure. They know when to push and when to pause. They know the difference between governance and ego. They’ve seen cycles. They’ve seen things break. They don’t panic at the first sign of turbulence.

A novice investor often reacts to volatility as if it’s an anomaly, not the default state of startups.

Experience doesn’t make someone perfect.

But it often makes them steady.

And when the lines between work and life blur, steadiness matters.

What follows are people who showed that steadiness.

The People

Ellen Chang — Customers, Customers, Customers

Ellen ran an aerospace accelerator before aerospace was fashionable.

We were engineers talking about satellites, sensors, spectral resolution.

She kept asking one slightly annoying question:

Who’s paying you?

Value. Pain. Revenue. Repeat.

She forced us to think beyond payload specs and into business reality. Passion is great. Invoices are better.

Brutal clarity delivered kindly.

Matt Thoms; The Blizzard and the Term Sheet

I met Matt in Dan’s garage.

Hours of diligence, real engagement, real curiosity.

But the moment I’ll never forget came later.

It was my anniversary. Dan and I had flown east to meet Matt and his team. Had that call not come through, we probably would have closed shop. We were at that point.

That evening, a blizzard rolled across the East Coast. Dan and I were sitting at a bar, watching snow pile up outside.

Matt called.

He offered terms.

We drank. Then we sprinted back to the Airbnb to go through everything. Dan caught the last flight back to San Francisco before the storm shut everything down. I was stuck for three days in the blizzard.

But we had a term sheet.

More importantly, we had found the right partner, someone who understood us and the journey we were about to take.

After funding us, Matt taught us how to run a board meeting. Not philosophically, but mechanically.

Agenda. Materials. What matters. What doesn’t.

First-time founders don’t know what they don’t know. He did.

Years later, we hugged and talked about our first launch like it was a shared war story. That continuity isn’t transactional. It’s relational.

Shelley Zhuang - “What’s the Product?”

Shelley ran the most grueling diligence of anyone.

Then at our first board meeting, she asked:

“What’s the product?”

Not satellites. Not analytics. Not vision.

What are you actually selling?

It shifted how I thought.

During chaotic stretches, I would call her and ask, “Am I crazy?”

She would answer.

When Malika and I had our daughter, another NICU experience that we hadn’t even had time to prepare. Shelley opened up her own collection of baby supplies before moving out of San Francisco and told me to take whatever we needed.

That gesture had nothing to do with IRR.

It had everything to do with humanity.

Jake Smith, Calm in the Storm

I met Jake at an Indian restaurant in the Castro. I think I was the only founder there and definitely the most sleep-deprived.

He later joined our board during a very stressful stretch.

No theatrics. No baggage. Just steady judgment.

He connected us to other funds. Advocated when he didn’t have to. Assessed situations without drama.

Recently, we shared wine and talked about family and what’s next.

Some relationships begin in boardrooms and evolve into something far more durable.

Hari Ramachandaran, More Than Capital

Hari and I built an investment thesis together over Zoom during COVID.

Four years passed before we met in person.

Singapore. Travel bans. International complexity.

He immediately understood the global angle of what we were trying to build and helped sharpen thinking that had real consequences.

But what began as an investment conversation evolved into something more durable.

Hari is now a close friend.

Our relationship extends well beyond OSK, into life, into ideas, into honest conversations that have nothing to do with quarterly updates. He’s someone I trust deeply, and that trust was built long before we ever shook hands in person.

He didn’t just invest in a company.

He invested in a relationship.

Todd Scruggs, Collaborative Diligence

Todd’s team invested in 2023.

Their diligence felt like collaborative problem-solving, not interrogation.

“What if you approached it this way?”
“Have you pressure-tested this?”

Late in my time at OSK, his support was invaluable; Across business, technical, and political terrain.

We grabbed drinks recently. The conversation wasn’t about returns.

It was about how he could help.

That tells you everything.

John Schwarck, The Energizer Bunny

John believes in the product.

Not casually. Religiously.

“The product works.”

He’s the product person’s dream customer; collaborative, candid, invested emotionally and financially.

“How can I help?”
“What do you need?”

Those were his defaults.

Over time, that working relationship became a genuine friendship. John is someone I can call about far more than product strategy.

We shared a recurring phrase during difficult stretches:

“Don’t hate the player. Hate the game.”

We said it often enough that it probably deserved its own board seat.

Sam Dobbyn And Our Cricket Diplomacy

Sam joined in 2022.

At first, it was formal. Quarterly cadence.

Then one London summer evening involving scotch, G&Ts, and selective memory shifted things.

He’s an operator first, investor second. That mattered when decisions became complicated.

We developed an encrypted communication system based entirely on cricket metaphors.

“Is this a cover drive?”
“Are we on a sticky wicket?”
“Is this an Ashes-level collapse?”

Ridiculous? Yes.

Effective? Also yes.

Today, Sam is not just someone I sat across a board table from. He’s a friend. Someone I trust well beyond a cap table.

Eddy Ho, The Right Way

Eddy flew to San Francisco and spent two days in diligence before joining the board.

Astute. Calm. Observant.

With him, it wasn’t just:

“What’s the right thing?”

It was:

“What’s the right way to do it?”

That distinction matters.

He understood that startups blur personal and professional lines. He made space for frustration without amplifying it.

Over time, that steadiness evolved into real friendship. Eddy is someone I hold in enormous respect; not just for how he invests, but for how he shows up as a person.

He didn’t have to.

But he did.

The Thread That Connects Them

The last few years were some of the most complex of my life. Not because work was personal, but because there was no clear line where work stopped and life began.

Board dynamics. Internal politics. Hard decisions.

All of it lives inside the same human container that holds family, fatigue, ambition, and doubt.

These individuals understood that.

They were steady when things weren’t.

They were honest when it was uncomfortable.

They expected integrity.

They showed up.

A seasoned investor understands that they’re not just allocating capital.

They’re stepping into the blurred space between ambition and reality, between spreadsheets and sleepless nights.

They invested in a company.

But more importantly, they understood what it takes to build one.

What I’ve written here are just small snippets from a long journey with each of them. There is so much more to say about how they helped, how they guided, and how they showed up in moments that never make it into board minutes.

I am and will remain forever grateful for their support.

When investors say they invest in the founders and the team, what they’re really saying is this: in the early days, strategy is fragile, technology is unfinished, and markets are theoretical. The only real question is whether a group of humans can align, argue productively, and still ship.

The team at OSK did exactly that — under sustained pressure, much of it outside our control — and delivered something that, historically, only nation-states with deep pockets and long timelines had managed to do.

To put that in perspective, here’s a comparison with other missions in the same sensing class. The table below reflects public information.

If you focus on the organizations here — USAF AFRL, NASA, ASI, DLR — these are government entities with decades of heritage, deep budgets, and thousands of people supporting development and operations. Many of these missions took well over a decade from formal kickoff to launch. EnMAP alone spanned roughly 15.5 years from approval to orbit.

OSK, working with partners including Corning and Astro Digital, built and launched a multi-spacecraft constellation in this same sensing class with a team that averaged about two dozen people.

No ministerial backing.
No infinite runway.
No tolerance for “interesting science” that couldn’t ship.

What the Team Actually Built

Between 2018 and 2026, OSK raised over $50M in equity and an additional ~$20M in non-dilutive funding through competitive government programs — capital that came with delivery expectations, not patience.

With that, the team:

• Designed, built, and launched seven spaceborne spectrometers, five of which formed the GHOSt constellation

• Delivered a fully integrated system spanning spacecraft, ground infrastructure, customer interfaces, and analytics — not a demo, not a science experiment

• Built constellation-aware tasking and ordering that allowed customers to go from request to on-orbit execution in hours, not days

• Implemented automated processing pipelines that converted raw data into usable products with minimal human intervention

• Associated customer tasks and capture strategies before data ever hit the ground

• Enabled near real-time anomaly detection, limited primarily by orbital mechanics and ground contact strategy

• Automatically reduced false positives by 80–90%, allowing analysts to review remaining detections in minutes rather than hours

• Built infrastructure that was scalable by design, where new analytics could be deployed without re-architecting the system

• Delivered operational detection of methane, natural gas, and liquid leaks

• Identified hazardous chemicals, industrial materials, and substances of national-security concern

• Mapped alteration minerals tied to critical resources including lithium, copper, gold, and more

• Delivered highly customized data formats and interfaces for government customers, alongside integrated insight delivery for commercial operators

• Built an integrated front end with API-driven backends that allowed data and insights to flow directly into customer systems

All of this was built and delivered by a team that averaged ~24 people over the company’s lifetime, and never exceeded 50 at its peak.

As a founder, I didn’t want to simply say I worked with the best team. I wanted to be able to say why this was the best team. What we accomplished together — under real constraints, with real customers, and real consequences — is the answer.

Why This Team Could Do What Others Couldn’t

When Dan and I started OSK, we were warned early and often:

“Hyperspectral companies don’t work.”
“There’s a graveyard of them.”
“The physics is hard, the data is messy, and customers don’t know what to do with it.”

They weren’t wrong. There is a graveyard — well funded, technically impressive, and littered with programs that never made it past pilots, papers, or PowerPoint. Hyperspectral imaging doesn’t fail because people aren’t smart. It fails because complexity compounds faster than organizations are willing to own it.

What differentiated this team wasn’t denial of those risks — it was clarity about where hyperspectral efforts typically fail, and a willingness to own those failures upfront.

First, we refused to treat hyperspectral data as the product. Data alone is not useful; decisions are. From day one, the team built systems that tied collection, processing, and interpretation together, so customers weren’t handed terabytes of complexity and a shrug.

Second, we built for operations, not experiments. Many hyperspectral programs optimize for academic performance or perfect imagery. This team optimized for reliability, latency, and integration into real workflows — the unglamorous requirements that determine whether something gets used twice or abandoned after a demo.

Third, constraints were not a bug — they were the forcing function. Limited capital, limited time, and limited margin for error demanded trade-offs. Those trade-offs forced collaboration across disciplines that are usually siloed, and decisions that many organizations postpone indefinitely.

Finally — and most importantly — this team owned the entire problem. Payloads, spacecraft, ground systems, infrastructure, analytics, customer delivery — all under one roof. When something broke, there was no one else to blame. That kind of ownership is uncomfortable. It is also incredibly effective.

And yes — the hyperspectral graveyard is real. But most of it exists because teams optimized for brilliance instead of usefulness. This team chose the opposite.

Values That Made It Possible

This didn’t happen because of culture slogans. It happened because of values that were practiced — sometimes messily, often loudly.

Transparency
We committed to being transparent, even when the answers were hard or incomplete. Startups are emotionally expensive. People deserve enough information to decide whether the ride is still worth taking.

Ownership
Everyone owned what they worked on. Fully. No hiding behind titles, org charts, or process.

Equity
Not equity as in cap tables — equity as in outcomes. Different disciplines, backgrounds, and life circumstances required different support to enable equal contribution. We optimized for that.

The team’s diversity wasn’t just academic or professional. It was ideological. Different ways of thinking, debating, prioritizing, and interpreting the mission coexisted — sometimes uncomfortably — but always with forward motion.

What Was Built, and What Endures

There were arguments. Long nights. Shoestring budgets. Periods of existential threat. And moments where optimism was a conscious decision rather than a natural one.

But progress was constant. And progress compounds.

At this moment, I am no longer a part of the team. I wish the remaining individuals the very best and hope they find their strategy a success. The problems ahead are hard, the stakes are real, and I know firsthand the caliber of the technical team still working to solve them.

And to everyone I had the privilege of building alongside — thank you. Working with this team has been one of the great privileges of my life. You showed up fully, through professional pressure and personal hardship alike, and gave this mission everything you had.

I hope our paths cross again soon.

This was the last photo I took in 2025. I remember that morning clearly. The sun was just coming up, the air sharp and cold, but wrapped in a warmth that felt grounding rather than urgent. As I walked, an unfamiliar feeling settled in—part unease, part calm.

The road itself felt alien at first, yet oddly familiar. It wound gently, split into forks, and offered choice without pressure. At some point, I realized it didn’t matter which direction I took. I wasn’t trying to get anywhere. I was just walking.

Mostly, though, I wasn’t alone. I was with my two constant companions—my mentors, whether they realize it or not.

They go by many names.
Aanik: Poppy, Nik Nik… the negotiator.
Eira: Ra-Ra, Pataka (firecracker), Toophan (storm)… and most recently, El Diablo, a name lovingly assigned by a cousin who knows her well.

They’ll get their own stories someday.

What I’ll say here is that their energy is infectious. Their curiosity, their laughter, and their eagerness to jump into action—figuring things out as they go—has a way of dissolving hesitation. There’s analysis there, especially with the older one, but it doesn’t block momentum. They move first, learn fast, and adjust in motion.

As a parent, one of the hardest things is not passing on your own fears. Not your worries, not your pessimism, not the quiet internal monologue that lists everything that could go wrong. I’m not perfect at this—not even close—but being conscious of it helps.

Take something simple: climbing a tree. I never really learned how. Even now, my mind immediately fills with questions. What’s the safest path up? What if I slip? That branch looks higher than it should. I’m not a big fan of heights. There’s a mild vertigo thing. Before I’ve even tried, I’ve already argued myself out of it.

For them, climbing a tree is just that—climbing a tree. They go.

So my job isn’t to constantly say “be careful,” though every instinct wants to. My job is to stay nearby. To help them up if they ask. To be there if they fall.

We talk a lot about genetics in our family—who got what trait from whom, the good ones and the harder ones. But genetics only sets the starting conditions. Experience decides what compounds.

Watching my kids grow has made something clear: reinvention isn’t reserved for the young. It’s about identifying the traits worth leaning into, and being honest about the ones that need work. Not eliminating fear, but learning how to assess risk. Not standing still because the answer isn’t obvious.

Perspective matters. So does empathy. Letting go of what holds you back, and allowing yourself to be pulled forward.

Through all of this, there’s one person who has seen every version of me—the highs, the lows, the in-between—without pause or reset. My partner in crime, Malika

She’s redefined persistence for me. Sometimes directly, sometimes simply by example. She’s taught me that the smallest things matter more than they appear. Paying attention to details that once felt inconsequential changed more than I expected.

Individually, things like making the bed or cleaning the kitchen can seem trivial. Collectively, they turn a space into a home, and over time became habits I practice every day.

That same micro-to-macro thinking quietly reshaped how I operate professionally too. Systems. Order. Small decisions compounding into clarity. Seeing both the forest and the trees. The micro and the macro aren’t opposites—they’re tightly coupled.

The other defining thread of the last several years has been work—specifically, building.

At OSK, the goal was never to be different for the sake of it. It was to challenge assumptions that no longer held. To push what was commercially and technically possible. To build products people actually wanted to use. To rethink analytics, delivery, and what “from zero to one” really means when there’s real hardware, real customers, and real consequences.

One moment from that journey stands out.

For years, there was an internal debate about a core operational system—something that sat at the heart of how we would plan, task, and operate satellites at scale. Many smart people wanted to buy something off the shelf. On paper, it felt safer. Faster. Less risky.

I didn’t think what we needed existed.

So I blocked it. Repeatedly. Not quietly. I argued the case across teams, through leadership discussions, and at the board level. Even the folks building the surrounding infrastructure pushed back. It would be easier to outsource. Easier to compromise.

But I had a clear picture in my head of what the company needed—not just then, but years out. So we stayed the course and built it ourselves.

When it finally came together, it wasn’t flashy. There were no high fives. No one said, “you were right.” What happened instead was quieter and better: the system worked. It unlocked capabilities we hadn’t been able to touch before. It enabled things that suddenly felt obvious in hindsight.

The same people who had debated it hardest were now using it daily, benefiting from what it made possible.

That was enough.

2025, in many ways, was a year like that.

The last image of the year felt reflective. But the first image of 2025?

That was a damaged door frame—a blown-out mortise, a strike plate falling loose the moment I opened the door. I remember standing there with tools, figuring out how to fix it.

That’s probably the most honest summary of the year.

2025 was about fixing things. Persistently. Sometimes inelegantly.

So when I think about 2026, I anchor myself to a different image. Not the broken frame—but the open road.

I’ve been asked what I want to do “when I grow up”. I’ve asked myself the same question.

The truth is, the core hasn’t changed much since I was five years old. I knew early on that I wanted to be close to aviation and aerospace. That curiosity carried me through formal study, startups, and some unconventional problems—from bio-inspired aerodynamics to energy systems to building space-based hardware and teams that could scale.

Walking that road with my kids, talking loosely about the future, something clarified for me—not in the form of a plan, but as a direction.

When I was younger and finding my way into aerospace, the person I kept coming back to wasn’t the loudest voice in the room or the most visible one. It was Burt Rutan. What I admired wasn’t just the aircraft or the records—it was the way he thought. He questioned assumptions others took for granted. He understood risk, but didn’t let it paralyze him. He built things that looked improbable until they worked.

That mindset stuck with me more than any single achievement.

What pulls me forward now isn’t certainty. It’s exploration—exploring what’s worth building, and then doing the work to build it. It’s working on hard problems in domains that still resist change. It’s building systems, technologies, and businesses that don’t yet have a clean category—but should. Space, in particular, is that frontier: demanding patience, judgment, and a willingness to be early and occasionally wrong.

I don’t need everything mapped out. I’ve learned that waiting for perfect clarity is just another way of standing still. What matters more is being flexible, staying curious, and making decisions—even imperfect ones—when the moment calls for it.

There’s a saying about aiming for the moon and landing among the stars. I’ve never loved it as a goal. But I do like it as a reminder: forward motion matters. Exploration has value even when the destination shifts. And sometimes the path you end up on—the one you couldn’t have planned—is the point.

So I’m choosing the road again.
Not because I know exactly where it leads,
but because it keeps me moving, learning, and building.

Every season starts the same way: optimism, spreadsheets, vibes, and the dangerous thought that this year feels different. It never is.

We all build defensive mechanisms early.
“If we lose one game, I’m done.”
“I won’t watch.”
“I’ll get my time back.”

None of this has ever worked. Not once. You get pulled back in like it’s muscle memory — because it is.

Sports: The Fastest Way to Age Without Actually Moving

Sports give you everything: hope, pain, suffering, grief, joy, stress, release.
It’s wonderful. It’s terrible. And it ages you faster than bad sleep and processed food combined.

I was born in India, went to Penn State, and love aerodynamics. Which means, obviously, my personal holy trinity is:

• Indian Cricket

• Penn State Football

• Formula 1 (specifically Lewis Hamilton)

And man… I’ve aged.

I watched everything. Waking up at ungodly hours for Test matches. Staying up late convincing myself Ferrari strategy might suddenly make sense. Watching Penn State football while mentally preparing for disappointment.

I don’t believe in superstitions — except I tried all of them anyway:

• My trusty cricket ball (juggling it, fake bowling it, pleading with it)

• Loud cussing

• Quiet cussing

• Not watching (while still cussing)

• Standing after second down

• Drinking cold water during critical moments

• Sitting perfectly still in my “lucky” office chair during races

None of it worked.

My rituals failed me completely. Which is obviously not evidence that sports outcomes are random — it just means I need better rituals.

So How Did I Actually Do?

I swore off sports at least five times this year. And yet here I am, watching Penn State play Clemson in a bowl game.

A bowl game.
In New York City.
In winter.
On a field that was half soggy, half frozen, and entirely unnecessary.

Who is this for?

Can we please stop pretending everyone needs a bowl game? Just give us playoffs and championships. Everyone does not need to be a winner.

The result? Satisfactory.
Which is sports-fan code for: “I didn’t throw anything.”

A Very Scientific Emotional Scoring System

To properly quantify the damage, I tracked my mood across games using the following highly rigorous methodology:

• Ecstatic (+3) — Rare, fleeting, suspicious

• Happy (+2) — Usually followed by anxiety

• Satisfied (+1) — Sports fan code for “I survived”

• Neutral (0) — Didn’t emotionally invest (lies)

• Anxious (-1) — Default state

• Annoyed (-2) — Coaching decisions

• Frustrated (-3) — Refs, umpires, strategy

• Devastated (-5) — Overtime losses, collapses, Ferrari

They say sports are a roller coaster. This year felt less like a roller coaster and more like a controlled descent into emotional insolvency.

Figure 1: Mood Score Over Time

Cumulative mood score across the 2025 sports calendar. Note the sharp decline beginning in late September, followed by a complete failure to recover — a textbook case of optimism colliding with statistical reality.

Once it dropped, it stayed down. No rally. No miracle. Just vibes and disappointment.

Figure 2: Game-by-Game Mood Distribution

Distribution of emotional states across games. Positive emotions appear briefly and without commitment. Negative emotions show strong persistence and excellent conditioning.

If sports were a portfolio, this would be a year of catastrophic asset allocation.

When Everything Broke

The crash started in late September.

Penn State’s overtime loss to Oregon didn’t just hurt — it rewired something in me. The losses that followed felt procedural, like paperwork. Eventually the university did the merciful thing and fired the head coach.

Then came cricket. India lost two home Test matches to South Africa.
At home.
The last time that happened was 25 years ago.

And Formula 1? Let’s just say Lewis Hamilton’s first season at Ferrari was… rough. Watching Ferrari after years of Schumacher dominance feels like watching a legacy brand quietly give up on quality control.

Data Doesn’t Lie. My Heart Rate Confirms It.

Sports are objectively bad for health. The stress, the cortisol, the constant near-misses.

According to my heart-rate data, watching sports qualifies as moderate-to-vigorous cardiovascular activity, which feels fraudulent given that I was sitting completely still, staring into the void.

Figure 3: Heart Rate During Some Penn State Games

Heart rate during Penn State football games. Peaks align with third downs, late-game decisions, and moments where “we just need one stop” was said out loud.

No movement. No exercise. Still physiologically dangerous.

Heart rate data shows no meaningful distinction between winning and losing in short-format cricket.

Figure 3: Heart Rate During Some Cricket Matches

ODIs and T20s keep fans in a state of permanent fight-or-flight, where victory simply means you survived the evening without long-term damage. Joy, if it exists, arrives briefly — usually during the highlights, not the match itself.

About Selling Hope

I sell a dream too — I’m a founder, that’s literally the job.
But when I miss, there’s accountability. I don’t get to reset expectations every season and call it “next year.”

In sports, hope is the product. Performance seems optional.

Where We Ended Up

Penn State beat Clemson and hired a new coach. Maybe he’s the one.

India salvaged the South Africa series in other formats, even if the Tests hurt.

Hamilton? Maybe a podium next year. Maybe redemption.

And me?

I’m hopeful again.
Because of course I am.

The definition of insanity is doing the same thing over and over and expecting different results — and I’m already looking forward to the New Zealand series, the T20 World Cup, the Penn State season, and maybe, just maybe, Lewis Hamilton finally winning the eighth championship he was robbed of in 2021.

See you next year.
I’ll bring the data.
And the heart-rate monitor.

Methods (Because This Is Apparently a Study Now)

Data Sources:
All data was collected involuntarily while watching sports I had already emotionally committed to.

Mood Scoring:
Emotions were assigned using gut instinct, pacing frequency, and muttering intensity.

Heart Rate Analysis:
All cardiovascular stress was emotional in nature. No physical activity occurred.

Statistical Rigor:
No smoothing was applied. Emotional instability was preserved.

Limitations:
Results may not generalize to emotionally well-adjusted individuals.

Conflict of Interest

The author is emotionally compromised.

It’s not.

It’s Vicki Hollub (CEO of Occidental) saying it on a World Economic Forum stage. Which is… notable. Because Occidental isn’t exactly a boutique artisanal methane studio. If the people producing the stuff are saying “get aggressive,” it’s worth pausing the hot takes and asking:

What is methane in our system, actually?

This is Part 1 of a 2-part series.

• Part 1 (this one): value, the journey, who “owns” methane, and why the whole thing feels weirdly unaccountable.

• Part 2: how to be constructive — how we encourage efficiency instead of doom-and-gloom compliance theater.

The Journey of Methane (aka: how it gets to your thermostat)

Methane doesn’t magically show up at your house wearing a PG&E name tag. It has a supply chain — a real one — and it’s not that exotic.

Here’s the path:

1. Wellhead (upstream): raw gas comes up (sometimes because you’re drilling for gas, sometimes because you’re drilling for oil and gas is the “bonus prize” you didn’t ask for).

2. Gathering: smaller lines collect gas from multiple wells.

3. Processing plant: gas gets cleaned to meet pipeline specs (water out, impurities managed, liquids separated).

4. Transmission pipelines: long-haul transport across regions (compressors, meters, tariffs, paperwork).

5. City gate: the handoff point into the local utility network.

6. Local Distribution Company (LDC): neighborhood pipes → service line → meter → home/office.

And just to head off the inevitable email: not all gas goes to homes. A lot goes directly from transmission to power plants and industrial facilities. Homes are one branch, not the whole tree.

FIGURE 1 — “Methane Journey to Customers

The physical path: well → gathering → processing → transmission → city gate → distribution → customers.

The Value… or no, the price

Wholesale methane (natural gas) is priced in $/MMBtu. Pick a typical benchmark number and it’s… cheap relative to what it enables.

Which leads to my favorite dumb-but-useful metric:

The Croissant Index™ 🥐

A decent croissant in SF is around $4.50.

At wholesale gas pricing, that’s roughly ~1.2 MMBtu of methane.

That amount of energy can either:

• bake an absurd number of croissants (if you batch them like a sane bakery), or

• heat a well-insulated apartment for days.

So yes: methane is cheap. The molecule is cheap.

The system is not cheap.

Because households don’t buy methane at Henry Hub. They buy:

• pipelines

• compressors

• distribution networks

• maintenance

• emergency response

• regulation

• and the privilege of not living in the “oops, that’s a gas leak” timeline.

FIGURE 2 — “Methane Facts / Why Methane Matters”

Methane is a small molecule with a big climate footprint — and it’s also a meaningful piece of the U.S. energy system.

Ownership vs Chain of Custody (aka: stop trying to fingerprint a methane molecule)

Here’s the core confusion:

People talk about methane like it’s a physical object you can tag and claim:

“That methane is mine.”

But once methane enters gathering/transmission systems, it becomes commingled. It mixes. It’s fungible. You do not get “your molecule” back.

So what does the industry actually track?

It tracks title and quantities at meter points

• Chain of custody = who physically controls the gas in their facilities.

• Title/ownership = who owns the scheduled quantities under contracts, settled at receipt/delivery points.

This is why the unit of commerce is energy (MMBtu/Dth), not “that specific methane molecule that left Midland at 2:07pm.”

FIGURE 3 — Title vs custody / chain

Custody moves physically. Title moves contractually. The molecule is not keeping receipts.

So who owns it?

Before production: mineral rights

In places like Texas, “who owns the minerals” is a patchwork:

• private mineral owners (families, ranches, trusts),

• University Lands,

• the state (GLO / school-related lands),

• municipal entities (airports, counties, districts),

• mineral/royalty companies,

• and sometimes producers.

If you own the mineral estate, you generally own the right to lease and develop oil/gas under your tract (subject to law/regulation). But — again — this is a rights framework, not molecule custody.

After production: leases + royalties

If you lease to an operator (Exxon/Chevron/Pioneer/etc.), you typically get:

• a bonus (upfront),

• and royalties (ongoing), based on your share of production value.

And yes: if a well produces oil + gas + NGLs, royalty owners generally get paid on each product stream per the lease terms.

But if this feels like a lottery sometimes, here’s why:

• reservoirs don’t respect property lines,

• modern horizontal wells don’t care about your 10-acre parcel,

• and your payout depends on unitization/pooling, your decimal, and whether wells actually get drilled.

It’s not a lottery ticket.
It’s a pooled lottery with engineers, contracts, and accountants standing between you and the jackpot.

Pipelines: the part where reality shows up as a “loss factor”

Transmission pipelines usually get paid like a toll road (reservation/transport charges, often in Dth/MMBtu).

And they apply something that quietly destroys the “molecule ownership” fantasy:

Fuel + LAUF (Lost and Unaccounted For)

• Fuel: compressors often burn gas to move gas

• LAUF: measurement tolerances, small leaks/venting, linepack changes, timing mismatches, entropy

Translation:

You put X in, you don’t necessarily get credited X out. You get net of system use + accounting reality.

That’s not politics. That’s operations.

Why methane gets treated like a nuisance (Permian edition)

In the Permian, a lot of gas is associated gas — produced because oil drilling is happening. Oil drives the drilling schedule. Gas infrastructure has to keep up.

So methane ends up in a weird position:

• it’s valuable,

• it’s cheap,

• it’s sometimes a byproduct,

• it’s hard to map accountability cleanly across custody handoffs,

• and it’s easy for responsibility to diffuse into “someone else’s problem.”

Which is why so many methane efforts get stuck as pilots: the value is real, but the incentives are fragmented.

The Corn Example (why I keep coming back to it)

Corn is boring, which is precisely why it’s useful.

Corn has:

• clear ownership (land + crop),

• visible inventory,

• measurable yield,

• and an entire industry obsessed with minimizing waste.

Methane has:

• underground uncertainty,

• commingled logistics,

• contract-defined title,

• and a tendency to be treated as a byproduct until it becomes a headline.

So when we ask, “what is the value of methane?” we’re really asking:
Is it being treated like an asset… or like a nuisance that happens to be flammable?

Conclusion (not doom, just… come on)

My Part 1 takeaway is simple:

For the U.S., methane should be treated like a finite national asset — and handled with the same care we reserve for anything valuable, measurable, and economically useful.

Not because we’re running out tomorrow.
Not because we need more doom.
But because leaking cheap energy at scale is the dumbest possible self-own.

Part 2 is where I’ll get constructive: what incentives would make “efficiency” the default — not a compliance chore, not a PR slide, and not a pilot program that dies the second budgets tighten.

Also: I’m rooting for it. Not because it’s easy, but because it’s one of the few “space infrastructure” ideas that could pull meaningful revenue from outside the classic circular economy of VC money + defense money + VC money again (Starlink being the loud exception).

Why this idea won’t die (and why it shouldn’t)

For a long time, the space sector has been a tidy loop: VCs fund space companies, those companies sell to defense and to other space companies, and we all clap because the slide deck had an Earth-from-orbit background.

A data center in space is compelling for one boring reason: it’s a path to sell compute—a thing the terrestrial economy already understands, already buys, and already overbuys.

If we want “space economy” to mean something other than government contracts and vibes, we need infrastructure that isn’t capped by “how many satellites can we sell to governments.”

The Economics (aka: it’s mostly a launch-cost story)

If you strip the concept art away, the economics pivot on one knob: $/kg to orbit.

If launch stays expensive, this is mostly a brand exercise with excellent renderings.
If launch gets cheap enough, the math gets… annoyingly plausible.

I used ChatGPT as a co-pilot and sanity-checked against the usual spacecraft sizing instincts (and yes, dusted off old SMAD muscle memory).

The common denominator: IT load (MW)

On Earth, almost all electrical power going into IT becomes heat: CPUs, GPUs, memory, networking—everything eventually turns into warm feelings.

• Earth: 1 MW IT load → ~1 MW of heat, but total facility power is larger due to PUE

• Space: 1 MW IT load → you still must reject ~1 MW of heat, but you do it by radiating it away (no convection fairy to save you)

A reminder: IT load ≠ total facility power

• Earth: 1 MW IT load → ~1.2–1.6 MW total facility power

• Space: 1 MW IT load → ~1.05–1.20 MW total (architecture-dependent)

Quick Earth vs Space napkin math (order-of-magnitude, not a religion)

Earth (1 MW IT load example)

CAPEX (delivered): In the headline table, we used ~$87M per MW of IT load as an all-in “delivered data center” number. That’s intentionally a blended figure:

• part of it is facility CAPEX (site, building, power distribution, cooling plant, commissioning, etc.)

• part of it is IT hardware (servers/GPUs/CPUs/RAM/networking/storage)

So this $87M/MW is not “just GPUs.” It’s “the whole thing shows up and works.”

OPEX (electricity, using the same assumptions throughout):

(Energy-only OPEX here; staffing/maintenance/taxes/network are omitted so we don’t accidentally argue about the wrong thing.)

Space (1 MW IT load example)

In orbit, energy shifts toward CAPEX (arrays + power electronics), while OPEX shifts toward ops (mission operations, monitoring, ground segment). The physics headline is still: heat rejection drives size.

Radiator sizing (simplified):

Using a representative case (~400 K radiator temperature), 1 MW IT load gives:

• Radiator area: ~1,560 m² (ballpark)

• Radiator mass: ~6–10 kg/m² → ~9–16 metric tons per MW IT

Space OPEX model we used (ops-heavy placeholder):

• Staffing / mission ops / monitoring: $5M/yr + $0.1M/yr × (MW IT)

• Ground ops (TT&C / gateways placeholder): $0.30M/yr

• Stationkeeping / attitude control placeholder: $0.15M/yr

So:

• Space OPEX/yr ≈ (5.45 + 0.10 × MW) million USD

• For 1 MW IT, that’s ~$5.55M/yr → ~$27.8M over 5 years

Important correction (the “$60M vs $87M” issue):
On the space side, the original “Space CAPEX” column was effectively platform-to-orbit (radiators/arrays/structure/bus, driven by mass-to-orbit and launch cost). That does not automatically include IT hardware.

So I added a separate IT hardware CAPEX = ~$60M per MW IT for the space case. That $60M is meant to represent the compute payload itself (GPUs + the rest of the server platform). It’s smaller than Earth’s $87M/MW because Earth’s number includes both IT and terrestrial facility infrastructure; in space, the “facility” piece lives in the separate platform-to-orbit cost.

In short:

• Earth total = facility + IT

• Space total = spacecraft platform + IT
  …and the table only makes sense if we keep those buckets separate.

The headline comparison table

Assumptions: 1–40 MW IT load; 5-year window; “delivered space CAPEX” modeled as 3× launch cost (manufacturing + I&T + commissioning); replacement/refresh and insurance ignored to keep this readable.

Two important notes so nobody throws tomatoes:

1. Earth OPEX above is electricity-only, using a single example electricity price. Change that price and this column moves a lot.

2. Space CAPEX is still dependent on the mass model (radiator/arrays/structure). The point isn’t precision—it’s “what dominates” and “how sensitive is this to launch cost.”

The Technology (aka: “648 tons?!” yes, and here’s the awkward part…)

At 40 MW IT, the rough sizing of the Radiators comes out to this:

Yes: ~648 tons.
The ISS is ~450 tons. But the ISS is also: astronaut-safe, bespoke, and assembled via a thousand heroic compromises. A compute platform can be designed for mass production + modularity + redundancy, which is the polite way of saying: fewer handrails, more “swap the whole module.”

“But can we even launch that?”

If we’re talking about genuinely scaling this, we’re implicitly betting on heavy lift + high cadence:

Also: we’re living in an era where “impossible” keeps quietly turning into “Tuesday.”
EVs with real range. Rockets landing themselves. Cancer research breakthroughs. AI getting promoted to everyone’s boss. So yes—deploying giant radiators and arrays is hard, but it’s not a physics “no.” It’s a money + engineering + iteration “prove it.”

“Who maintains the servers?”

If the answer is “astronauts,” you’ve already lost.

This has to be designed like spacecraft: automation-first, fault tolerant, modular. And yes—cue the Optimus joke: a squad of robots doing orbital hot-swaps while we all pretend that wasn’t the plan the whole time.

Connectivity (the real boss fight)

This is the biggest practical choke point: 24/7 affordable connectivity.

If the economics depend on moving large volumes of data and serving workloads that interact with Earth, then downlink/uplink pricing and availability can make or break the entire concept.

• Want real-time services? Latency and continuous links matter.

• Want batch compute? You still need consistent throughput.

• Optical links and crosslinks help, but they don’t eliminate the core problem: getting data to the ground, into terrestrial networks, reliably and cheaply.

If “space compute” ends up priced like a boutique novelty ISP, it becomes a niche. If it rides a high-volume connectivity ecosystem, it has a path to scale.

And now we reach the part everyone on Earth-based projects feels in their bones…

Regulation (aka: the final boss is paperwork)

I was going to put lighthearted political memes here. But this is where schedules go to die.

For a 40 MW IT-load data center, regulation isn’t just “fees and forms.” It’s time, and time is a brutal line item. Because when you’re sitting on $3–4B of capital, every year of delay brings:

• Lost revenue (the money printer is still in the box)

• Higher financing costs (capital hates uncertainty)

• Tech obsolescence (hardware plans age faster than milk)

Here’s the tangible comparison:

On Earth, you can have the best design in the world and still get slowed down by process, interconnection queues, and local opposition. Texas is often faster, but the grid and extreme weather are their own genre of plot twists.

Space doesn’t erase regulation—it swaps county hearings for licensing, spectrum coordination, and debris mitigation expectations. Different maze, different dragons.

And that tees up the other “tangible” debate people bring up immediately…

Intangibles (the “yeah but what about…” section)

Space: Orbital debris (the universe’s sandblaster)

Debris isn’t a rare Hollywood event. It’s closer to environmental wear plus occasional “bad day.”

• Lower altitudes benefit from atmospheric drag cleaning things up faster

• Higher altitudes keep debris around longer (sometimes much longer)

• Large-area structures (radiators, arrays) increase exposure simply because… they’re huge

A useful way to think about it is damage tier, not “impact/no impact”:

So the design philosophy has to be: damage tolerant + redundant + degradable.
Lose some radiator capacity? You throttle compute. You don’t brick the whole platform.

Also: going higher isn’t a universal escape hatch. You may trade short-lived debris for long-lived debris (and other unpleasantness like radiation). “Just go above it” is not a strategy; it’s a coping mechanism.

Earth: Environment (the planet tries every year)

Earth isn’t gentle. It’s just familiar—and you can send repair crews.

Once seismic risk is engineered away, weather + grid events become the recurring operational risk. Unlike earthquakes, weather shows up on your calendar with annoying consistency.

Earth’s superpower is repairability. Space’s superpower is sidestepping land constraints and (potentially) some permitting timelines. Pick your poison—then engineer for it.

And that brings us to the only responsible ending…

Conclusion

There are risks everywhere: some high, some low, most annoyingly non-zero. The technical challenges are real in either case. The difference is where the pain lives:

• On Earth, you pay in regulatory time, grid constraints, and weather-driven disruptions.

• In space, you pay in mass, deployment scale, debris environment, and connectivity economics.

If launch costs don’t come down meaningfully, space data centers remain a fascinating thought experiment and a very expensive way to avoid CEQA.

If launch costs do come down (and cadence goes up), then the question stops being “is this stupid?” and becomes “where does compute belong when we’re no longer treating Earth as the only place infrastructure can exist?”

And honestly—if the path for humanity is to become a spacefaring civilization, then this entire debate becomes moot. You’re going to want compute nodes in places spacecraft naturally go, in orbits that are accessible, serviceable, and well-connected.

At that point, “data centers in space” isn’t a hot take.

It’s just… infrastructure. The most adult outcome possible.

Yes, free and open data policies have enormous scientific, societal, and economic benefits. Landsat and Copernicus are extraordinary public investments. But here’s the part that often gets conveniently skipped:

someone paid the bill.
A very large bill.

What “free” actually costs

Before we argue about business models, let’s anchor reality.

Representative public EO program costs:

• Landsat 7: > $850M (development, launch, operations)

• Landsat 8: > $900M

• Landsat 9: > $850M

• Copernicus (EU, 2014–2027): ~ $9.7B

These aren’t rounding errors. These are deliberate, taxpayer-funded programs with a public-good mandate, not a profitability mandate.

That distinction matters — because once a private company builds comparable infrastructure, the economics change completely.

The commercial reality check

Let’s look at three public commercial EO data companies: Planet, BlackSky, and Satellogic.

Total capital raised (VC + public markets):

• Planet: > $700M

• BlackSky: > $450M

• Satellogic: > $300M

Combined, they still don’t come close to the Copernicus budget alone.

Yet these companies:

• manufacture satellites,

• run global mission operations,

• maintain ground networks and cloud pipelines,

• employ hundreds to thousands of people,

• and are expected to compete with data that appears “free.”

So let’s strip this down to first principles.

The only metric that actually matters: revenue per orbit

Forget revisit, resolution, constellation size, or marketing decks.

A typical EO satellite in a polar orbit goes around the Earth roughly 16 times per day — about 5,840 orbits per year.

If you operate a constellation, the core question is brutally simple:

How much money does each satellite earn every time it goes around the planet?

Everything else is commentary.

Actuals and clarified assumptions (important)

Two clarifications up front:

• For Planet, satellite depreciation largely sits in Cost of Revenue, not OpEx. For this analysis, I’m using an operating cost burden concept (OpEx plus constellation-related depreciation burden), because cash still leaves the building either way.

• Planet’s actual customer count at FY24 end was 1,018.

2024 snapshot (order-of-magnitude)

What they need vs what they make

Using the numbers above, the math becomes painfully simple.

Revenue needed per orbit per satellite

(just to cover operating costs — no margin)

• Planet: $225

• BlackSky: $1,019

• Satellogic: $342

Revenue actually made per orbit per satellite (FY2024)

• Planet: $175

• BlackSky: $873

• Satellogic: $74

Shortfall:

• Planet: 22%

• BlackSky: 14%

• Satellogic: 78%

(Planet and BlackSky are in the same general orbit. Satellogic is on a different trajectory.)

The part that really hurts: Planet’s customer math

Planet ended FY24 with 1,018 customers.

Break Planet’s actual revenue down further:

• $175 earned per orbit per satellite

• ÷ 1,018 customers

• = $0.172 per customer per orbit per satellite

Every time a Planet satellite circles the Earth, each customer contributes about seventeen cents toward keeping the constellation operating.

What does Planet need instead?

Planet needs $225 per orbit per satellite to break even.

At $0.172 per customer per orbit, Planet would need:

• ~1,310 customers to cover operating costs

• That’s ~292 more customers beyond 1,018

• Just to break even. No margin.

And that ignores the really uncomfortable part:
a major portion of Planet’s revenue is government — so the “commercial customer” economics are likely even harsher than this blended customer count suggests.

This is the vicious loop:
more customers → more sales & support → higher operating costs → higher break-even bar.

“But Landsat created Billions in value”

Yes, Landsat has created enormous downstream value in agriculture, disaster response, climate science, and research.

But that value was:

1. Never monetized at the data layer, and

2. Never intended to be.

Landsat’s ROI accrues to society, not to NASA’s income statement.

Using Landsat’s societal value to argue that commercial EO companies should give away data is like arguing that because highways enable trillions in commerce, trucking companies should haul freight for free.

Public goods and private balance sheets are not interchangeable.

Closing thoughts: where this actually breaks

There are plenty of businesses today that successfully make use of “free” public EO data — and yes, some of them even generate revenue, and in rarer cases, a small profit. That model works up to a point.

But the moment you try to scale beyond exploratory analytics, dashboards, or one-off insights, the cracks start to show.

To get to the next level — operational monitoring, asset-level decision-making, repeatable value — you need:

• higher revisit rates,

• higher spatial resolution,

• more spectral bands,

• better temporal consistency,

• and often bespoke datasets tailored to a specific problem.

That’s where free data stops being sufficient.

If you read my previous piece you start to see where the system breaks down.

EO providers are stuck in a bind: they need to raise prices or escape pilot purgatory to build sustainable businesses — or transition to a fundamentally different model, where manufacturing lines, integration and test facilities, and mission operations, historically treated as cost centers, become revenue-generating products in their own right. In practice, this is why EO companies continue to drift toward defense and sovereign work — not because they love bureaucracy, but because factories and mission ops finally start paying for themselves.

Meanwhile, analytics and insight providers want cheaper data, but in practice still buy imagery project by project, not as long-term infrastructure — which makes planning, pricing, and scaling nearly impossible for data providers.

So we end up with a structurally misaligned market:

• data providers chasing volume to cover fixed costs,

• analytics companies reluctant to commit to recurring data spend,

• and customers who like the idea of EO-driven insight but balk at paying for the underlying inputs.

Free data enabled the ecosystem to exist.
But it won’t carry it forward.

At some point, the industry has to confront the uncomfortable truth:

you don’t get persistent, high-quality, operational EO for free — you just decide who pays, when, and how indirectly.

And no matter how you slice it…

…it’s still just math.

Share

Aravind’s recent essay, “The Earth Observation Adoption Curve – From Hype to Invisibility”, nails part of the phenomenon. But after years of building hyperspectral missions, analytics stacks, and customer-facing products, I’ve come to believe the issue is deeper, as Aravind suggests, structural, and far less likely to be solved by a new round of satellite renders.

Here’s the blunt version:
EO, in its current configuration, is not naturally designed to escape pilot-program hell.
Not because customers don’t care and not because the technology is primitive, but because the architecture of the industry is mismatched to the structure of the problems it’s trying to solve.

Below are the observations that, to me, explain why.

1. EO Modalities: Excellent Tools, Terrible Generalists

EO consists of a spectrum of sensing modalities:

• SAR – deformation, structure, things that twitch

• High-res RGB – spatial and visual detail

• Thermal – heat signatures (and what’s pretending not to be one)

• Multispectral – broad physical/material indicators

• Hyperspectral – chemical/material identification in high fidelity

• LiDAR – structural geometry (whenever orbital systems become operational)

Figure 1: Different sensor modalities available. Image courtesy:
1. SAR Image - NASA
2. Thermal Image - SatVu
3. Multispectral Image - NASA
4. Hyperspectral Image - NASA
5. LiDAR Image - NASA

Every modality is valuable. Every modality is narrow. Physics doesn’t do “one size fits all,” and neither does EO.

A single modality may be exceptional in its slice of reality but completely blind to the rest. Industrial problems, unfortunately, do not politely restrict themselves to one slice.

That’s the first structural trap:
You can have a sensor that performs beautifully but still cannot solve a full problem. So you end up with a scientifically impressive dataset… and a pilot.

2. Coverage, Revisit, and the Economics of Physics

You can convince investors that your sensor is revolutionary.
You cannot convince orbital mechanics to give you better revisit.

Complex sensors → expensive satellites → small constellations → poor revisit → limited coverage → incomplete solutions → pilots.

This is not pessimism; it’s just geometry.

Industrial customers need reliability, frequency, and persistence. A satellite that revisits every 7–14 days does not “monitor” anything. It simply drops by occasionally to check if the problem is still there.

This is why EO companies cannot scale on performance demos alone.
You can’t operationalize around sporadic data.

3. Scale Comes From Depth, Not Brochures

EO companies love claiming applicability across a dozen verticals.
This is also how you guarantee that you don’t meaningfully penetrate any of them.

Scaling requires depth:
understanding workflows, edge cases, operational thresholds, dispatch protocols, regulatory constraints, and what constitutes “actionable” versus merely “interesting.”

This level of embeddedness is uncomfortable and slow.
Which is exactly why almost no one does it.
Which is also why almost no one scales.

4. The Chasm: Platform → Sensor Suite → Analytics → Delivery

This is where the mismatch becomes impossible to ignore.

Platform

For EO companies, satellites are the foundation.
For customers, satellites are trivial. They don’t care whether the data comes from orbit, a drone, or someone standing on a crate with a handheld. They care whether the problem is solved. They want the black box.

Sensor Suite

Your sensor’s characteristics define the boundary of your solution. They also define all the things you cannot do. No industrial workflow neatly conforms to those boundaries.

And when you look at the landscape in terms of resolution vs. cost (see figure 2), the constraints become even clearer:

• Low-resolution, low-cost systems (e.g., multispectral from Sentinel/Landsat) are excellent for broad trends and large-area monitoring, but cannot provide the granularity needed for event-level decisions.

• Medium-resolution systems (e.g., hyperspectral) begin to unlock real material and chemical insight, but the cost and complexity rise sharply—which makes scaling coverage and revisit difficult.

• High-resolution, high-cost systems (commercial RGB and SAR) do have near-daily global coverage when you aggregate fleets.
  But active tasking and persistent monitoring are prohibitively expensive, and even when you do buy the data, the modality itself cannot solve every industrial problem.

In practice:

Low resolution → economical → limited insight.
Medium resolution → scientifically rich → difficult to deploy at scale.
High resolution → higher cost → active monitoring becomes uneconomical.

No single modality sits in the “Goldilocks zone” of high fidelity, high revisit, low cost, and broad applicability.

Figure 2: High-level overview attempt of different modalities, spatial resolutions available, their costs and coverage.

Which brings us to the next issue: even when a modality can contribute meaningfully, the rest of the system has to support it.

And this is exactly where the industry hits the chasm.

Backend Infrastructure

Most EO companies handle ingestion, calibration, atmospheric correction, and cloud processing competently. That is not the weak point.

The real gap is the lack of analytics architecture built into the backend.
Most “solutions” are still analysts running routines, standalone scripts, or isolated ML workflows bolted on to the pipeline rather than integrated into it.

This works for a demo. It does not work for a product.

Analytics

This is the real bottleneck. Customers don’t want imagery; they want:

• detections

• insights

• confidence

• context

• low false alarms

• repeatability

That requires consistent revisit, multi-modality, QA loops, and robust validation.
Most EO setups cannot deliver all of this simultaneously, so the analytics remain incomplete—and the product remains a pilot.

Delivery

A dashboard is not delivery.
A PDF is not delivery.
Emailing a map and praying is definitely not delivery.

Delivery is the mechanism that inserts the insight directly into the customer’s workflow—whether that is SCADA, an API trigger, a control room, a work-order system, or a mobile alert in a field environment with weak connectivity.

This part receives the least attention and has the highest impact on adoption.

5. The Refinery Analogy: A Systems Problem, Not a Pixel Problem

A refinery is monitored by an orchestra of sensors:

• pressure

• temperature

• flow

• vibration

• gas detection

• particulates

• corrosion

Each modality sees something different. None can run the refinery alone.
But all are integrated, tightly, into SCADA and the operational decision loop.

That is what real multi-modality and real integration look like.

Now look at EO.

Each sensing modality lives in a different company:

SAR over here,
RGB over there,
HSI in a third place,
thermal in a fourth.

No one owns the multi-modal stack.
No one integrates at the system level.
Everyone holds a fragment of the puzzle and insists the fragment is the picture.

This is where data providers and analytics companies both get stuck.

Data Providers

Data providers could go deep. But many don’t, because:

• Building analytics capability is expensive

• Investors prefer “data margins” (illusory)

• Government contracts are easier and steadier

• Going vertical requires patience and depth rather than speed

• “Selling data” feels clean on paper, even though it rarely scales

So most data providers stay horizontal, offering limited datasets across many sectors, none of which fully adopt them.

Analytics Companies

Analytics companies try to be the integrators.
They attempt the SCADA role of EO—multi-modality, fusion, verticalization.

But the economics collapse immediately:

• Multi-modality is expensive without scale

• Data licensing restricts usage

• Revisit gaps break analytics

• Sparse coverage increases false alarms

• Small pilots never unlock better pricing

They are trying to build a command center without owning or governing the sensor ecosystem.

A refinery succeeds because its sensors and analytics live in one integrated system.
EO stalls because its sensors and analytics live in separate companies with conflicting economics.

And that is how you end up with the never-ending cycle of polite, repetitive pilot programs.

6. The Pilot Trap (Otherwise Known as Orbit)

Everything becomes partial:

Partial sensing → partial analytics → partial confidence → partial budgets → partial deployment → repeat pilot.

Customers learn the script:

“Very interesting. Let’s run a pilot.”

It’s corporate for: “We don’t trust this yet.”

And honestly, they shouldn’t.

7. The Only Way Out: Think in Systems

This is the section that matters most.

The way out is not more satellites.
It is not richer pixels.
It is not a prettier dashboard.
It is not more AI.
It is not a new spectral band.

The way out is building a system.

A system that spans modalities, integrates them, validates them, and lands the result inside the customer’s operational workflow.

That requires a shift in mindset:

Platform is just an endpoint.
Sensor is just a viewpoint.
Analytics is the logic.
Delivery is the bridge.
And the product is the system that ties all of it together.

Not the image.
Not the spectral curve.
Not the heatmap.

The system.

Only then does EO stop being “additional information” and start being infrastructure.

8. Pricing: The Quiet, Brutal Bottleneck

EO pricing today reflects EO’s internal economics, not customer value.

Data providers charge per sq km because satellites are expensive.
Analytics companies charge like SaaS because software investors prefer it.

Neither approach aligns with operational reality.

Industrial customers buy outcomes—not pixels.
A refinery doesn’t care about the “per-sensor-log megabyte fee,” and neither do EO customers.

If your system prevents a shutdown, flags a leak, reduces a loss, or avoids a major compliance hit, your value has nothing to do with area or spectral bands.

The only pricing model that unlocks multi-modality and true adoption is value-based pricing.

But you can only price that way once you’re delivering a system, not a dataset.

The way out isn’t more data.
It’s building a system that ties all the data together into something customers can actually use.

That’s the conundrum.
And that’s the opportunity.

Was it to stand apart from the fake, AI-generated accounts creeping into LinkedIn? Maybe. But really, the irony is that in both the digital and physical world, being human is what should matter most.

Even professional platforms are becoming politicized. It’s easy to hide behind anonymity, to troll, to reduce people to “the other.” Each time I think I’ve seen it all, I see something worse. So is this a political post? Perhaps. But my intent is simple: a self-portrait.

I am Human.

Right or left, LGBTQ or straight, male, female, they/them, white, Black, brown — we’ve multiplied our labels. Instead of identities uniting us, they’re dividing us. Online, we’re pushed into corners, forced to define ourselves against others. It feels less about belonging to a country or community and more about drawing lines of difference.

And yet, at the most fundamental level, we are the same. Every human being enters the world equal. The only real difference is where we are born — geography, culture, and circumstance, not essence. Being human is the lowest common denominator, the ground we all share. History reminds us of this too: Jefferson’s words that “all men are created equal,” however imperfectly applied, and Gandhi’s insistence that dignity and equality belong to everyone. His idea of satyagraha — truth-force — was not abstract philosophy but a lived practice, a way to resist division and oppression by holding fast to our shared humanity.

Human isn’t about erasing difference. It’s the base layer, not an eraser — the floor of universal dignity. Our shared capacities—to be harmed, to reason, to err and to repair—are what bind us. Differences matter because they tell us what needs repairing. But when we overweight difference, we put the cart before the horse.

Our earliest ancestors survived not by isolating, but by bonding — sharing tools and food as early as two million years ago in places like Kenya. By around 300,000 years ago, Homo sapiens emerged, forming households and later tribes. Thirty thousand years ago, as populations grew, larger tribal networks took root. Over centuries, those networks evolved into kingdoms and states, and by the mid-1600s, into the modern system of nations. The pattern repeats: as resources shift and definitions multiply, we return to tribes.

The old campfire gave warmth and belonging; the digital campfire often gives heat without light. “Human” isn’t another tribe — it’s remembering there’s one fire we all sit around.

I am Human.

What makes us human is not just biology but speech, reasoning, self-awareness, and the ability to create tools, cultures, and communities. Yet those very strengths are now weaponized. We’re talking at each other, not with each other. Labels can help some feel seen, but they also risk driving us apart.

As a parent, I wonder what kind of world my children will inherit. We’re not only depleting the planet that sustains us — we’re eroding the fabric of our own humanity.

Writing is catharsis for me. If life is a journey, then we are all passengers. I hope to be remembered not just as an entrepreneur, or an Indian, or an American — but as a human. Someone who erred, tried, built, connected, and cared.

That’s why I call myself Human. It is the one truth that makes us equal, the denominator we all share, before anything else is added.

Through the years, I’ve had the opportunity to deep dive into markets within Energy, Mining, Agriculture, and other sectors. It’s fascinating how they work — who the players are, which companies dominate, and how value actually moves through the chain.

The Space sector, by contrast, can feel like an echo chamber. We sometimes convince ourselves we “know the market” just because another space company mentioned it. A cursory check, a hand-wave, and suddenly it becomes gospel. But the truth is, every sector is far more complex once you dig in.

Which brings me to agriculture.

EO & Ag – Does the Math Work?

Everyone in remote sensing loves to talk about agriculture as a “huge opportunity.” But when you dig into the economics, it’s not so simple. Data is expensive, farm margins are razor thin, and the value proposition of satellites remains cloudy.

⚠️ Disclaimer: Pricing is based on publicly available Maxar and PlanetScope numbers, USDA datasets (2021–2024), and farm economics literature. I’ve averaged numbers, not modeled every crop. If I’ve missed something or made a wrong assumption, reach out — I’ll fix it.

EO Pricing: Reality Check

Notes:

Avg. U.S. farm ≈ 438 acres (1 km² = 247 acres).

Cropping season ≈ 6 months.

Maxar “new tasking” = $75/km², but requires a minimum purchase of 50 km².

Planet archive = $2.25/km², but requires a minimum purchase of 250 km².

📌 Key point: Even the cheapest options are out of reach for most farms. A week-old “archive” image may not be of much use when each week of the growing season matters.

Agriculture at Scale: What’s Being Grown?

In 2024, U.S. cropland covered 328 million acres (1.3M km²). Only 303M acres were harvested — the rest failed or sat fallow (“summer fallow” is an intentional moisture-saving practice, not farmer vacation).

And here’s the punchline: 83% of all acreage is just six crops — corn, soybeans, hay/alfalfa, wheat, cotton, and sorghum. Fruits, vegetables, and nuts — the glamorous, high-margin crops — together cover just 10.3M acres.

📌 Callout: 83% of farmland = 6 crops. Everything else (fruits, nuts, veggies) = just 10.3M acres.

If you want an excellent breakdown of what’s actually growing in America, check out Sarah Mock’s Substack: Why Food Companies Can’t Drive Change.

Acreage harvested by major U.S. crops in 2022 (USDA NASS / Census of Ag):

Farms & Farmers: The Income Reality

The U.S. has about 1.9M farms, averaging 463 acres. About 95% are family-owned. Only ~5% are corporate farms, and most of those are still small family LLCs.

Now the kicker: median on-farm income is around $0 — and often negative. Most farm households rely on off-farm income for survival.

• Avg. farm household income: ~$98,000.

• But only ~25% comes from farming. The other 75% comes from off-farm jobs, investments, or outside business.

• Small to mid-size farms (< $250K sales/year) often net losses of ($2,000) to +$10,000 annually.

• Commercial producers (> $350K sales/year) can net $100K–$200K per operator in good years.

📌 Callout: Half of U.S. farmers make little to nothing from farming. Most rely on off-farm income.

U.S. Crop Farmers (excluding dairy & livestock, USDA 2022 Census + ERS):

Net Returns Per Acre (Approximate)

🌽 Major Row Crops

• Corn: $150–$300/acre

• Soybeans: $50–$150

• Wheat: $10–$50

• Hay: $30–$100

• Cotton: $100–$250

• Rice: $100–$200

• Sorghum: $50–$120

🌱 Pulses & Minor Grains

• Chickpeas: $80–$150

• Lentils: $50–$100

• Canola: $100–$200

• Sunflower: $120–$250

🍎 Fruits, Vegetables, Nuts (High Value)

• Almonds: $1,500–$2,500

• Grapes (wine): $2,000–$5,000+

• Blueberries: $3,000–$6,000

• Strawberries: $10,000–$20,000

• Lettuce/Leafy Greens: $4,000–$8,000

• Potatoes: $500–$1,000

📌 Callout: Satellites are optimized for corn & soy. But the real money is probably in strawberries & blueberries — which cover only a sliver of U.S. farmland.

The Margin Mismatch: Why Pricing Breaks the Model

NDVI and other vegetation indices are incredibly useful. They’re simple, proven, and widely adopted. Free sources like Sentinel-2 (10m) and Landsat (30m) already give farmers and agribusiness valuable tools for monitoring stress, timing irrigation, and planning rotations. Agtech companies have had limited success on top of free data.

Where the system breaks is when commercial satellite operators try to sell higher-resolution data at premium prices into an industry that simply doesn’t have the margins to absorb it.

• Row crops: Corn, soybeans, and wheat make up the bulk of U.S. farmland — but they yield just $10–$300/acre. That means a 500-acre farm might net $50K–$100K in a good year. Dropping $30K–$40K on commercial tasking isn’t an investment; it’s a gamble that eats half the year’s profit.

• High-value crops: Strawberries, blueberries, nuts, wine grapes — these could justify higher-res monitoring ($10K–$20K per acre revenue). But they occupy just a sliver of U.S. farmland (~3%–5%). Satellites designed to scan millions of acres at once aren’t optimized for this niche.

• Free vs. Paid: Sentinel/Landsat deliver imagery at zero cost and “good enough” spatial/temporal resolution for many decision-makers. I think Farmers and insurers will happily use free data if it answers 80% of their needs. Paying $75/km² for a marginal improvement isn’t compelling.

📌 Callout: Paid high-res data rarely clears the ROI hurdle.

Until pricing drops dramatically — enough that farms themselves can adopt at scale — the economics remain upside down. Downstream customers will keep funding small pilot projects, but that won’t unlock the real market.

The Bottom Line

Remote sensing was built for scale. Agriculture operates on tight margins. That’s the disconnect.

In my view, for satellite-based data to truly succeed, the end user still has to be the farmer or the farm itself. There’s scale there — millions of acres, millions of decisions — and that’s where the product could be most effective.

But the hurdles are clear:

• Pricing: current costs don’t line up with farm-level margins.

• Adoption at scale: it only works if enough farms buy in to drive volume.

Yes, we can keep going downstream — serving agribusiness, insurers, traders, government programs — but that won’t achieve true scale. Those customers will keep writing $50K, $75K, maybe $150K checks for pilot programs. Useful, but not transformative.

The question is whether EO can break out of pilot project purgatory and actually reach the farms that need it most.

From Numbers to Narratives

Details can be complex, numbers and statistics can be boring, and it’s easy to lose the forest through the trees. One of the best ways to explain what you’re doing is to tell a story — to storyboard the pitch so it resonates. And who doesn’t love a good story?

I’ve even tried my hand at writing a book — about 500 pages of an unfinished fiction manuscript still sitting on my hard drive. Who knows if it will ever see the light of day. But it taught me this: storytelling makes the abstract real.

That’s how I’d like to position this exploration of “farm to table.” Not just economics, but characters and narratives that make it tangible.

Cue in JaCob Rhule (Corn), Cotton Cash (Cotton), and Berryoncé (Strawberries) — alongside the colorful cast of Cargill Brooks and Feed Zeppelin.

🎟️ This is your front-row ticket to the harvest tour ;).

Back in 2016, Earth Observation felt like déjà vu on loop. Thousands of papers, endless demos of “visual change detection,” and still no one could explain what actual value this industry was delivering. A decade later, the question hasn’t changed — and the answers are still thin.

So we asked: why are we spending billions on counting cars from orbit when cheaper, faster, better tools already exist? Why are we squinting at pixels when the real action is invisible to the naked eye?

Earth isn’t a postcard. It’s a cauldron of chemistry — 118 elements, 94 found naturally, and over 70 million known compounds created, destroyed, and recycled daily. Energy, mining, defense, agriculture: every major sector runs on chemistry, leaks chemistry, and hides inefficiencies in chemistry. And yet, the EO industry was still acting like a bunch of visual artists with satellites of all shapes and sizes.

The Hyperspectral “Aha” Moment

Spectroscopy already told us: every material interacts with light in its own unique way. Hyperspectral imaging takes that to the next level: hundreds of bands per pixel, each one a chemical fingerprint.

From space, that means you’re not just looking at what’s there — you’re seeing what it’s made of. Suddenly you can spot methane leaks, mineral deposits, stressed crops, or dodgy emissions without boots on the ground. That’s not counting cars. That’s diagnosing Earth.

The technocrats in us were giddy. Billions of dollars to be unlocked. It looked easy. (Spoiler: it wasn’t. Building something from scratch never is!)

Lesson #1: Don’t Sell Data, Solve Problems

Too many space companies start with a shiny technology and go hunting for problems. When that fails, they default to selling raw data. Easy to do, hard to monetize.

We decided from day one: no. Value comes from solving actual customer pain points — ideally the kind that customers will also pay you for. (A radical idea, apparently.)

Ellen Chang drilled it into us: “Customers, customers, customers.” So we did the calls, did the homework, and found the industrial sectors where chemical-level monitoring wasn’t just nice-to-have — it was critical.

The Race to Space

We bootstrapped, sharpened the go-to-market, and before our first institutional round we already had a product scoped to a specific customer problem, one that scaled with space-based sensors.

Then came the race to orbit. We weren’t building Hubble 2.0 — we didn’t need to. We needed elegant, compact, risk-tolerant sensors that worked for our mission. Seven launches later, seven sensors lit up. First light every time.

Along the way, we got grief from “thought leaders” who swore our system wouldn’t work without 10x budgets and 10x testing. One gave us the lesson of Hubble and if people had listened to this industry sage, there wouldn’t have been a Hubble snafu; (My retort, you led the team, you could’ve just said you’re doing it).

Our satellites were never designed to compete with Hubble. And yet… they work. Funny how that goes.

So What Exactly Did We Build?

Industry analysts still pigeonhole us: “application-specific,” “just another data provider,” “probably defense.”

Wrong. What we built is a platform: a single constellation (GHOSt) that can monitor global assets, detect chemical signatures over them, and feed alerts into command centers. Application-agnostic by design — we just went to market with one application first because you have to start somewhere.

That’s why all the bands matter. It’s not indulgence — it’s universality. If you want to be industry-agnostic, you need to capture everything.

Pixels vs. Chemistry

Figure 1: OSK HSI image from GHOSt over an urban area with an air station

Ask someone what they see in the image above, and you’ll get the usual: vegetation, highways, hangars, maybe planes. Boost the resolution to 10 cm, and what changes? You’ll see the cars, maybe even the guy on the bench reading a newspaper.

But so what? If you really want to know what article they’re reading, buy their browsing history from Google.

Visual imagery can show change, yes — useful for defense, insurance, surveying perhaps. But at the end of the day, it’s still just… what you see. And sometimes what you see can be misleading.

So Much More

That image, captured at 8m GSD with 468 bands per pixel, contains over 3 million embeddings. Each spectrum a natural, chemical-level embedding(TM).

After two years of operations, our dataset contains 400 billion+ embedding footprints. Coupled with our spectral library, that’s not a picture of Earth — it’s a chemical ledger of it.

So when I look at that image, I don’t see “fields” or “runways.” I see:

• Concrete vs. asphalt

• Plastics, paints, liquid hydrocarbons

• Vegetation species and their health

• Soil composition

• Gas leaks and emissions

That’s why we built OSK. Because solving chemical problems at scale will always matter more than counting cars from space.

Nyota Uhura, Lieutenant Commander, polyglot, linguist, and the coolest person to ever answer an interstellar phone call. In Strange New Worlds, she speaks 37 languages, decodes alien subtext, and does it all without Slack, Jira, or Google Docs.

She did, however, have a starship full of AI, universal translators, and holographic comms — so, really, she was doing “communication” on what amounts to god mode.

And yet, even with all that tech, Uhura still did the one thing our modern tools can’t: actually listen, process, and respond with the right amount of context.

This past year, personally and professionally, I’ve been reminded that poor communication is like a bad transporter beam — it scrambles everyone involved.

The Message Is the Thing

We’ve solved the medium problem — Silicon Valley gave us enough channels to send a single question 14 different ways and still not get an answer.

The problem now is message ownership.
And message ownership starts at home.

The Annual “Where Are You?” Saga

![Downtown SF map with “We’re Here” marker here]

My father visits from New Delhi every two years. He has a gravitational tether to one square mile of Delhi, so getting him to explore San Francisco is a victory in itself.

One evening, he calls:

Dad: “Tushar, we’re ready to come back. What’s the best way?”
Me: “Where are you?”
Dad: “Here.”
Me: “Where is here?”
Dad: “Downtown, San Francisco.”
Me: “Where in downtown?”
Dad: (frustrated) “How am I supposed to know? Somewhere near this store…”

Fifteen minutes later, I’m still no closer to knowing where “here” is.

“Do you see the Transamerica Pyramid?”
“Yes.”
“Where’s the sun?”
He responds: “Next time I’ll bring a sextant.”

Lesson: Lack of context = you driving up and down California Street like an Uber driver with amnesia. These days, I skip the interrogation and just ask him to share his live location. Privacy invasion tech for the win.

The Cement That Wasn’t

Years ago, I was pulled into a plant operations nightmare: a coal flue gas process gone wrong, cement coming out wrong, and three teams locked in what felt like an endless triathlon of meetings, emails, and “let me walk you through my 72-slide deck.”

The communication problem?

• A critical cleaning step wasn’t documented anywhere.

• Ops never mentioned they’d skipped it.

• Every meeting generated more speculation than answers.

A teammate and I ditched the conference room, went to the plant floor, and asked straight questions in person. Within minutes, we learned about the skipped cleaning — something weeks of email chains and presentation slide transitions never uncovered. We cleaned the dryer. Cement problem solved.

Lesson: The failure wasn’t in the chemistry. It was in the absence of clear, shared facts.

The Dashboard of Discord

In a remote-work project, I asked the team to build a dashboard. I didn’t assign a lead (rookie mistake). People ran with their own interpretations, built competing versions, and discussed it in fragmented Slack threads and side calls.

The noise looked like this:

• Stakeholders excluded from decision points.

• Ad-hoc meetings producing contradictory directions.

• Junior members pulled into tasks that clashed with their other deadlines.

The solution was eventually delivered — on time, even. But the discord remains. There’s still lingering frustration, mistrust, and competing ideas about “how it should have been done.”

Lesson: Delivering the product doesn’t mean you delivered alignment.

The Communication Playbook

If you want to avoid Uhura-less chaos, you need more than just Why, Who, What. Here’s the expanded field guide:

1. Why — Set the stage, not just the task.
Don’t just say what you need — explain why it matters, the bigger context, and the consequence of not doing it. People commit harder when they understand the stakes.

2. Who — Define roles before you hit send.
Not just names, but what each person is responsible for, how they fit into the puzzle, and who makes the final call. Cuts down on turf wars and “I thought you were doing that” moments.

3. What — The actions, in priority order.
List the actions, deadlines, and dependencies — plus how they tie back to the “Why.” Bonus points for flagging what’s not in scope so people don’t wander into side quests.

4. How — The communication format.
Agree on where updates will be given (email, Slack channel, weekly call) so nobody is chasing five different inboxes.

5. When — Timeframes and check-ins.
Specify not only the deadline, but when you’ll review progress, and how you’ll decide if the plan needs to change.

Even Uhura, with her 37 languages and 23rd-century AI backup, occasionally had to hail someone twice.

The goal isn’t perfection.
It’s clarity — so the next time you send a message, the person on the other end knows exactly what you mean, why you mean it, and how to act on it.

Because if you don’t, you’ll end up in one of two places:

• On California Street, wondering where your dad is.

• Or in a meeting, watching a 72-slide deck about why your cement is wrong.

I’ll post questions and updates that come my way, and you can jump into the discussion.

Join chat

How to get started

1. Get the Substack app by clicking this link or the button below. New chat threads won’t be sent sent via email, so turn on push notifications so you don’t miss conversation as it happens. You can also access chat on the web.

Get app

2. Open the app and tap the Chat icon. It looks like two bubbles in the bottom bar, and you’ll see a row for my chat inside.

3. That’s it! Jump into my thread to say hi, and if you have any issues, check out Substack’s FAQ.

And here’s the kicker — I never believed raw data was the answer. Yes, the U.S. government buys it (thank you, NGA), but beyond that? Commercially, EO data doesn’t solve clear pain points. At least not on its own. That’s why I’ve never seen much value in selling data as-is. Does a market exist? Technically, yes. Is it the one we should be chasing? I’ll unpack that in a bit.

First, Let’s Define Some Terms (Before Someone Tries to Sell You a Level 1B “Product”)

In the EO world, we like to throw around terms like data, analytics, insights, and products — often interchangeably, and often incorrectly. So before we dive in, here’s how I define them:

• Data is the raw stuff. Pixels, spectra, lat-long coordinates. It’s what satellites collect.

• Analytics is when you start processing that raw stuff. Maybe you classify some land, detect change, or flag anomalies.

• Insights are when you can look a customer in the eye and tell them something that actually helps them do their job better.

• A product is when that insight shows up, on time, without needing a PhD, a GPU cluster, or 400 hours of training to understand it.

Traditionally, EO "products" have been marketed using legacy terminology like Level 1A, 1B, 2, and so on — a system born in the 1970s by NOAA and NASA. The idea was to standardize data formats for the scientific community. Noble goal. But here's the problem: that same science-first mindset is still at the top of the funnel today.

And that’s where the commercial bottleneck begins.

The entire EO pipeline — from sensor design to capture, processing, and so-called “product” delivery — is optimized for researchers and government scientists. If you want to deviate from that? Good luck. Everything downstream is built on top of these rigid standards. There’s no room to build vertically, no flexibility to pivot toward a real-world solution. You end up navigating a maze of gatekeepers and standards committees, only to land... another proof-of-concept.

This is why EO keeps spinning its wheels commercially. Everyone’s busy perfecting Level 2 data, while the customer just wants to know where their assets are, what’s changed, and what to do about it.

Analytics, Insights… and Still No Obvious Value

Analytics and insights in EO are often just fancy derivatives of the raw data — and let’s be honest, the real value of those derivatives remains fuzzy. Sure, there’s a market for EO data, but what problem does it actually solve for a paying customer? That’s still murky. And worse, most of these “analytics” products are built for specialists, not for the average company that needs actionable, plug-and-play answers.

To be fair, I’m generalizing here. A handful of EO players are genuinely trying to solve problems vertically — and a few have had modest success. But no one has truly scaled into the commercial mainstream.

A Lesson from the Space Industry: The Starlink Effect

Take the communications satellite industry as an analogy. For most of the 2000s, giants like Intelsat, SES, and Telesat all have fleets of powerful geostationary satellites. Their services? Practically indistinguishable. Satellite communications were mostly a government game and the commercial market was fragmented.

Middlemen like Gogo Inflight popped up, buying bandwidth and reselling it to airlines — essentially trying to turn satellite comms into something vaguely useful for regular consumers. But the real game-changer came later.

Starlink. A company barely a decade old now boasts 6 million customers (I’m one of them) and generates an estimated $6.9 – $8.2 billion annually. For context, SES — one of the largest, oldest operators of GEO satellites (founded in 1985) — made $2.1 billion in 2024.

So, why is Starlink crushing it? I see two reasons, and they’re painfully simple:

1. They solved a real problem. Connectivity. People everywhere want reliable internet, and Starlink built a product that works for the masses, not just for a few telecom engineers.

2. They set their own standards. Starlink didn’t get bogged down by decades of legacy processes. They designed what they needed, when they needed it, for their customers.

So... How Does EO Replicate That Kind of Success?

The question we should be asking isn’t “who wants EO data?” but rather: “how do we make Earth Observation invisible?”

Yes, I get it — delivering global internet and delivering EO insights aren’t the same thing. But the principle is: when you build something so intuitive, so embedded in daily life that users don’t even realize it’s powered by satellites, that’s when the EO industry wins.

Until then? We’re stuck in proof-of-concept purgatory — pilots, projects, one-off demos, rinse, repeat.

But this isn’t just a product design problem. There are deeper structural issues:

• The massive gap between EO’s estimated value (cue the World Economic Forum report) and actual realized revenue.

• The tension between free vs paid data.

• The broken feedback loop between who buys EO products (often analysts or governments) and who actually uses the insight (usually not the same people).

• And the most basic question: what information are we even trying to extract, and for whom?

More on that in a bit. But before I go, let me offer one example of a company that did get it right: Google.

Google Maps is arguably the most successful EO-derived product in history. It’s used by billions, generates billions in revenue, and powers everything from food delivery to ridesharing to real estate. You don’t think about where the satellite imagery came from. You just care that it gets you to the nearest pizza place.

And here’s the real kicker: there’s no reason why the EO industry — with all its spectral bands, daily revisits, and advanced sensors — can’t build something better than Google Maps.

So where does that leave us? Somewhere between infinite potential and chronic under-delivery. Earth Observation isn’t broken — far from it. But if it wants to break out of its niche and into the commercial mainstream, it needs to stop selling complexity and start delivering clarity.

In the next post, I’ll dig deeper into the economics: why EO’s perceived trillion-dollar value hasn’t translated into real commercial success, the tension between open vs. proprietary data, and what it would take to build the “Google Maps of Everything” for industry.

Stay tuned. The orbital reality check isn’t over yet.

Here’s a sampler of the responses I got:

• “100s of bands across a spectral range”

• “Contiguous band coverage but also >20 bands”

• “More than 10 bands” (lol, ok)

• “Enough spectral bands for an application” (so… vibes?)

• “USGS says >37 bands” (authority flex!)

• “When processing requires distilling the data down to the purest signals” (sounds like wine tasting, but sure)
  

By these definitions, even Landsat, WorldView-3, and Sentinel could be classified as “hyperspectral.” And yet, most people in the industry would sooner delete their LinkedIn than call those payloads hyperspectral.

So I turned to our future overlords—AI and ChatGPT. Here’s what the machine spat out:

“Hyperspectral data is a type of remote sensing data that captures information across hundreds of narrow and contiguous spectral bands…”

Sounds promising, right? Then it continues:

“…collect data in dozens to hundreds of wavelengths…”

Wait—dozens to hundreds? So we’re back to hand-wavy numbers.🥴

My (biased but correct) definition:

Hyperspectral = A large number of spectral bands + narrow bandwidth + continuity across the spectrum.Oh, and all of the above should be available in the full dataset. Don’t hit me with “but it’s tunable!” Excuses are for underachievers. Obviously, I’m partial to our own OSK system. Did you really expect humility?

But let’s zoom out. The two most under-appreciated heroes in this whole HSI game are GSD and SNR. Everyone loves to flex their GSD, but SNR? Crickets. And yet, SNR is everything. You could have a GSD sharp enough to read street signs from space, but if your SNR sucks, your data’s basically cosmic noise. And how high should SNR be? Simple: high enough to be useful. (Shocking insight, I know.)

So, Who’s Even Doing Hyperspectral?

Now that we’ve totally cleared up what hyperspectral actually is (lol), let’s look at who’s in the game

Figure 1. Current HSI systems operating in space

All these specs—spectral range, available bands, GSD—sound impressive, but what do they actually mean? What’s a “spectral range” beyond a fancy term that sounds like it belongs in a sci-fi movie? What does “available bands” get you (besides a bigger marketing slide)? And GSD—why should you care that one pixel covers a 64m² area instead of 900m²? Let’s unpack the buzzwords and figure out why any of this matters.

Spectral Range

To help decode the chaos, I pulled together a sampler platter of Earth Observation sensors and the data they collect, mapped across chunks of the electromagnetic spectrum. This isn’t the full spectrum (calm down, physics majors), and it’s definitely not a complete list of satellites or constellations—just enough to make sense of the landscape without writing a thesis.

Figure 2. Specific parts of the Electromagnetic spectrum and various sensors

Hopefully, the brilliant (and most likely under-appreciated) infographic gives you a clearer view of where different sensors play along the spectral range

• SAR (Synthetic Aperture Radar) – Capella Space, Umbra, ICEYE, and the usual suspects like NASA/NOAA and ESA.

• High-Res RGB & Multispectral – Planet’s Dove, SkySat, and Pelican fleets, MAXAR’s imaging workhorses, and the classics—Landsat and the Copernicus program.

• Hyperspectral – That’s our lane: OSK’s GHOSt constellation, doing the most in the spectral domain.

Another major flex in the hyperspectral world is band count. Figure 2 gives you a taste of how MSI (multispectral imaging) uses broad, chunky bands, while HSI (hyperspectral imaging) slices things into fine, contiguous layers. But what does it really mean when someone brags about “32 bands” versus “468 bands”? Is more always better?

Enter Figure 3 (pretend it's here) – a simple plot showing how many data points one pixel gives you, each point representing a unique wavelength. Spoiler: more bands = more data = more potential insight = more excuses for using “AI” in your pitch deck

Figure 3. Spectral bands available for each pixel

And let’s not forget: GSD makes or breaks your ability to do anything remotely precise. An 8x8m pixel covers 64m². A 30m pixel? That’s 900m². That’s the difference between spotting a puddle and guessing it might rain.

Figure 4. Landsat 8, 30m RGB Image over Singapore (left), GHOSt 5 Standard collect, 8m, RGB image over Changi Airport, Singapore (center) and a close of the terminal building, runway and taxiway of the GHOSt 5 standard collect (right).

Final Take: Does It Actually Matter?

Let’s wrap this up. HSI data isn’t about just band count or just GSD or just SNR. It’s about how all those things combine to actually solve a problem. You know—real-world utility, not just pretty spec sheets.

Now, who has the best sensor? The internet would have you believe everyone does. Just scroll through Twitter, sorry, X:

• “Most advanced sensor ever!”

• “A new era of EO!”

• “Revolutionizing geospatial intelligence!”

Figure 5- Image credit: ChatGPT

Honestly, it’s like watching kids in a sandbox argue over who has the coolest shovel.

Truth? None of it matters if your fancy sensor doesn’t solve the problem it was built for. The most advanced sensor is the one that works for the job (and it’s all TBD right now). Everything else is just a popularity contest.

And yes—before the PR folks swarm in—this info is based on what’s actually flying. If your amazing, revolutionary, world-shattering hyperspectral system is still in “Phase I,” great! Hit me up when it’s in orbit.

Disclosure:

OSK has always claimed to be the best-in-class. Best tech, best strategy, best applications. And yeah—I think OSK has the best sensor. Don’t believe me? Look at the data.

Authored by me. Enhanced with sarcasm and spectral sass by ChatGPT.