In technology markets, complexity, capital intensity, and network effects tend to drive consolidation around a small number of dominant hardware and software providers. AI compute has centralized around Nvidia, while mobile operating systems are controlled by Apple and Google. The autonomous vehicle (AV) space is poised to follow a similar pattern. The immense technical challenge of full autonomy means only a handful of companies are likely to succeed, but those that do will gain leverage far beyond mobility. Their technology breakthroughs will extend into adjacent industries unlocking new growth opportunities, compounding scale advantages, and strengthening competitive moats.

AV innovation is advancing along two tracks: vertically integrated AV developers and robotaxi operators, and third-party technology enablers that support auto OEMs, truck manufacturers, and robotaxi fleet operators. The latter group includes companies such as Mobileye, Wayve, and Applied Intuition, as well as Nvidia which provides a compute and software platform for developing and building AV systems.

Within this third-party supplier cohort, strategic models diverge. Mobileye offers a turnkey solution, delivering a comprehensive full stack hardware and software AV system. Waymo may eventually commercialize its Waymo Driver platform for third party use, positioning it to compete directly in this segment. In contrast, Applied Intuition provides vehicle intelligence development and simulation tools, along with a vehicle OS and autonomy software stack that enables customers to build, test, and refine autonomous systems tailored to their specific applications and operational requirements.

This modular approach enables Applied to target not just the AV market, but also adjacent industries such as defense, agriculture, and construction & mining. Its work on “collaborative autonomy,” facilitating coordination between systems, further broadens its addressable market. While Mobileye has been less explicit about its plans beyond automotive autonomy, similar growth opportunities may emerge over time.

In this landscape, solving autonomy’s hardest problems is the ticket to long term value creation. High technical and capital barriers to entry mean only a few players will break through and Mobileye is well positioned to be one of them. Backed by a strong product pipeline, growing commercial traction, and a cost efficient platform, Mobileye is executing a strategy designed to give it a competitive edge in the AV and robotaxi markets.

Mobileye’s 2025 partnership announcements further reinforce its leadership position. Volkswagen plans to launch a robotaxi service in Los Angeles in 2026 using ID. Buzz vehicles equipped with an autonomous system developed in partnership with Mobileye, with additional cities to follow. During initial testing and rollout, these vehicles will operate with human safety drivers onboard. Lyft and Marubeni have also announced a partnership to deploy a robotaxi fleet in Dallas powered by Mobileye’s AV systems. Additionally, Mobileye secured a design win with a major global OEM for its proprietary imaging radar, purpose built for autonomy and offering superior resolution and object detection.

Subscribe now

Third-Party Tech Enablers Take the Wheel

The future structure of rideshare and transportation markets will hinge largely on the success of third-party AV technology enablers. Alongside integrated players like Waymo, Tesla, and Zoox, these external suppliers will help determine whether the robotaxi landscape consolidates around a few dominant firms or evolves into a more fragmented AV ecosystem.

Both vertically integrated and third-party suppliers must overcome the complexity of building AV systems that perform safely and reliably across diverse geographies, weather conditions, and when encountering unforeseen edge cases. Those with systems capable of generalizing performance across such unfamiliar environments and responding effectively to unpredictable scenarios will be best positioned to capture an outsized share of industry value.

However, technical excellence alone is not sufficient for long term competitive advantage. Cost efficiency will be critical. To scale effectively, AV systems must be both high performing and at a cost that facilitates favorable economics. In addition to developing a cost efficient system and tech stack, companies that can apply their technology across multiple domains, such as passenger transport, commercial logistics, and defense, stand to benefit from broader addressable markets and improved unit economics. By spreading R&D and capital costs across diverse revenue streams, these companies can achieve greater operating leverage and pricing flexibility.

Mobileye’s Strategic Position in the AV Market

Mobileye is arguably the leading third-party AV system supplier. While aiming to match Waymo’s benchmark for precision and safety, Mobileye’s advanced AV platforms, Chauffeur and Drive, are designed to achieve greater recall across a wider range of geographies, driving scenarios, and unstructured environments, crucial for global scalability. Like Waymo, Mobileye employs a compound AI system that integrates multiple subsystems and sensing modalities utilizing rules-based elements with deep neural networks, resulting in a pragmatic, scalable architecture.

This contrasts with Tesla’s “vision-only,” end-to-end neural network approach, which while promising, still struggles to generalize and faces technical challenges in handling real-world variability and edge case scenarios. This is problematic when continuous adaptation to new environments is required. These concerns are supported by data from the Tesla Full Self Driving (FSD) Tracker which still shows frequent FSD disengagements indicating that the system may still not be ready for a large scale unsupervised, driverless deployment.

While Waymo and Tesla have greater scale and access to captive compute resources, raw compute power and massive datasets alone don’t ensure success in autonomy. Mobileye has ample access to both on-premise and cloud compute resources, as well as extensive real world driving data. Ultimately, AV system performance hinges on overall system design and architecture, and the efficiency and effectiveness of vehicle-level compute, not just model training.

Two core features further strengthen Mobileye’s position: its cost-efficient, camera-centric sensor suite enhanced by its proprietary imaging radar, and its Road Experience Management (REM) mapping system. Unlike Waymo’s lidar heavy approach, Mobileye places cameras at the center of its stack and uses lidar and radar to provide what the company refers to as True Redundancy, two independent subsystems capable of full AV functionality. This architecture enhances fault tolerance and enables cross validation between sensing channels, increasing reliability in edge cases while keeping hardware costs down.

Mobileye’s proprietary imaging radar is a further innovation, it delivers superior object detection and resolution that rivals lidar, without the associated cost or power demands. In contrast, Tesla’s reliance on cameras and elimination of radar from its vehicles has sacrificed redundancy and introduces limitations in low-visibility conditions.

Complementing its hardware stack is Mobileye’s REM mapping platform. Tesla is often credited with a significant data advantage, and indeed its access to real world driving data is substantial. However, it is not alone. Mobileye has harvested over 56 billion miles (as of 2024) of driving data through its REM platform. In contrast to Waymo’s use of high-definition maps, which require mapping vehicles and repeated surveying, REM uses a bottom-up, more cost efficient crowdsourced approach. Mobileye collects anonymized data from millions of vehicles equipped with its Advanced Driver Assistance Systems (ADAS) systems to build and continuously update its mapping platform. Mobileye’s system on a chip ADAS solutions have been deployed on approximately 190 million vehicles worldwide. This method offers significant advantages in cost, speed, and scalability enabling more rapid expansion into new geographies.

AV Cost Set to Move Lower

As Mobileye, Waymo, and others drive down AV system costs and ramp up robotaxi supply density, reduced cost per mile and lower fares will follow unlocking positive price elasticity, expanding the addressable market and accelerating adoption across urban mobility markets.

Mobileye’s AV system architecture delivers enhanced power and compute efficiency, facilitating lower overall system cost. Its Chauffeur eyes-off product is projected to initially cost $4,500–$6,000, with future versions falling to $4,000–$5,000. Although Drive, its driverless robotaxi system, will carry a higher cost, it is expected to remain below that of most non-Chinese competitors.

Waymo is also working to lower robotaxi costs by supplementing its high end Jaguar I-Pace fleet with more affordable vehicles like the Zeekr RT and Hyundai IONIQ 5. However, sourcing from Zeekr, a Chinese OEM, may be complicated by new U.S. Commerce Department rules restricting the use of connected and autonomous passenger vehicle hardware and software developed, manufactured, or supplied by entities controlled by or connected to China or Russia.

If Waymo produces the Zeekr RT domestically at its Arizona facility to bypass tariffs and regulatory hurdles, U.S. production costs will still far exceed those in China, pushing the price well above the $40,000 China market price of the Zeekr Mix, which is based on same platform as the RT. While estimates for Waymo’s AV system cost vary and remain speculative, its reliance on costly lidar and a higher cost compute platform suggests its all in system cost is likely well above that of Mobileye’s more efficient, camera-centric solution.

Volkswagen’s planned 2026 deployment of ID. Buzz vehicles, equipped with Mobileye’s AV systems, on Uber’s platform in Los Angeles provides a compelling cost benchmark. While the Pro S trim of the ID. Buzz retails for nearly $60,000, VW’s internal fleet production cost is likely materially lower. Even after adding the cost of Mobileye’s AV system, the total upfront capital cost of these robotaxis is expected to remain highly competitive with Waymo’s lower priced vehicles.

The OEM Dilemma: Autonomy or Dependence?

As robotaxis gain traction and autonomous features become more mainstream, auto OEMs face mounting pressure to integrate autonomy and participate in the robotaxi market deepening reliance on third-party tech providers like Mobileye. Strategically, Mobileye differentiates itself from competitors such as Waymo through an OEM aligned go-to-market approach. Unlike Waymo, which may eventually commercialize its Waymo Driver platform while also operating its own fleets, Mobileye avoids competing with its customers. This neutrality could prove pivotal, as automakers may hesitate to rely on suppliers that also act as rivals. Tesla may in time offer a lower cost autonomy solution, but lingering questions around its reliability and neutrality as a partner remain.

Growing third-party reliance creates strategic challenges for OEMs and fleet operators. As autonomy becomes less of a differentiator, competition will shift toward price. While falling system costs will expand adoption through price elasticity, third-party supplier dependent operators will still face higher costs and thinner margins than vertically integrated players raising the importance of scale and operational efficiency.

To reduce dependency, many OEMs have invested in internal AV system development, benefiting firms like Applied Intuition. Its “white box” approach offers modular tools that allow OEMs to build and customize autonomy stacks tailored to specific use cases. In contrast, Mobileye offers a turnkey, full-stack solution, though it has introduced more flexible options to meet customer demands for greater system flexibility and customization, giving automakers more room to incorporate features that help distinguish their offerings.

Despite their efforts, most auto OEMs have struggled to scale in-house software and AV capabilities. Volkswagen’s software challenges and the shut down of robotaxi initiatives at GM (Cruise) and Ford (Argo) illustrate the difficulty. As internal efforts falter, reliance on third-party enablers like Mobileye and Applied Intuition is likely to grow reinforcing their strategic importance, even as they limit OEMs’ ability to fully differentiate.

Although they overlap in some areas, Mobileye and Applied Intuition’s specific strengths allow them to serve complementary roles. This is evident in partnerships like Volkswagen Group, where Mobileye’s AV technology and Applied’s suite of modular tools and solutions can meet different customer needs, illustrating how both can potentially coexist within the same AV ecosystem.

Rethinking the Role of Rideshare Marketplaces

The rideshare industry’s business model has long relied on a highly fragmented supply structure of human drivers and vehicles, but that model is poised to change as autonomous vehicle technology advances. A key question is whether third-party AV system suppliers will enable a more fragmented robotaxi supply landscape or whether the market will consolidate around a finite number of fully integrated players and large scale fleet operators. Even if third-party suppliers foster a broader base of fleet operators, allowing platforms like Uber to retain relevance, the market’s underlying structure is still shifting toward greater supply side concentration.

Regardless of whether fleets are vertically integrated AV developers or third-party operated, all robotaxi providers will be laser focused on vehicle utilization to drive unit economics. This may drive collaboration with platforms like Uber and Lyft, but early supply constraints, as seen with Waymo’s direct-to-consumer (DTC) service in San Francisco, limit the need for intermediaries.

Over the longer term, Uber and Lyft’s value may lie in aggregating smaller third-party AV fleets that lack the scale to pursue DTC strategies. However, as competition intensifies, consolidation is likely, raising questions about the durability of incumbent platforms. In a more concentrated supply landscape, will the expansion of the total addressable market be enough to offset pressures from declining gross revenue per mile and the risk of lower take rates?

Another looming risk is disintermediation from new distribution models. Large scale AV system suppliers, incentivized by system sales and potential revenue sharing agreements, could roll out white-label DTC apps for their customers that aggregate supply across their fleet partners. By providing this service at minimal cost, they could boost operator profitability and potentially undercut Uber and Lyft’s economics reshaping marketplace dynamics.

Over time, it will be in the interest of fleet operators like VW (with Uber) or Marubeni (with Lyft) to seek multi-platform distribution or possibly their own DTC strategies to boost utilization. This would require neutral service providers like Moove for fleet management, rather than relying on rideshare incumbents. And while running a robotaxi across multiple platforms may pose technical network challenges today, it opens the door for companies like Applied Intuition to meet this nascent industry’s emerging needs, and develop a cross platform AV routing and integration tool as the ecosystem evolves.

Market Disconnects and Strategic Optionality

Mobileye’s recent stock performance underscores a growing divergence between market perception and underlying strategic progress. While Intel’s recent secondary offering prompted a pull back in the stock, Mobileye’s upward revision to second quarter revenue guidance ($502–506 million) and a string of high profile AV partnerships point to improving trends in its core ADAS business and growing momentum in capturing future autonomous system revenue opportunities. While some investors attribute the stronger outlook to demand pulled forward ahead of tariffs, Mobileye’s full year guidance already embeds a sequential revenue decline in the second half, suggesting a conservative baseline and potential for further upside.

Mobileye’s $13 billion market cap stands in stark contrast to private market valuations. Applied Intuition, for example, raised $600 million in June at a $15 billion valuation, highlighting a disconnect in how public and private markets are valuing leading AV enablers. With expanding commercial traction and a cost competitive, scalable AV platform, Mobileye’s long term positioning remains compelling and the company presents an underappreciated opportunity in the race to define the future of autonomy.

As autonomous vehicle systems promise to drive massive disruption and unlock new growth opportunities across transportation, Uber remains at the center of the conversation about how its current business model will adapt to an industry undergoing structural and competitive change.

Along these lines, Uber is reportedly in discussions with its co-founder and former CEO Travis Kalanick to support a bid to acquire the U.S. subsidiary of the Chinese autonomous vehicle (AV) technology company Pony.ai. The talks are said to be in early stages, with no financial details available and no clear definition of what role, if any, Uber would play.

The potential sale of Pony.ai’s U.S. operations follows a Commerce Department rule introduced earlier this year restricting the use in the United States of connected and autonomous passenger vehicle hardware and software developed, manufactured, or supplied by entities controlled by or connected to China or Russia. This policy is consistent with the notion that the structure of the global robotaxi market will diverge across regions as national security and cybersecurity concerns in both the U.S. and China restrict robotaxi operations to primarily domestic players.

If Uber ends up investing in Pony.ai, it could mark the first step toward reestablishing a more closely aligned autonomous technology effort similar to its former self-driving unit, the Advanced Technologies Group, which was acquired by Aurora in December 2020. Moreover, should Uber become a direct investor in a robotaxi operator, it would arguably signal an acknowledgment that the shift from driver-based to driverless rideshare services will fundamentally reshape industry economics in favor of those who control autonomous vehicle technology.

This context is crucial, as the U.S. robotaxi market is likely to evolve toward an industry dominated by a handful of key technology players given the steep barriers to entry. Since only a few companies are expected to develop autonomous systems that operate safely and reliably across diverse environments, varying conditions, and when encountering unexpected edge cases, the result will be a more concentrated supply structure than today’s driver-based rideshare model. As the utility of Uber’s two-sided marketplace diminishes, economic leverage and profit pools are set to shift toward these core AV technology providers, such as Waymo and Mobileye, who are well positioned to be prime beneficiaries of the transition to vehicle autonomy.

While Uber is pursuing a broad range of partnerships as the AV and robotaxi market evolves, its core rideshare business model remains predicated on today’s dynamics and depends on the industry’s highly fragmented supply structure staying intact, yet that structure is set to change. At the heart of the debate around Uber’s future is whether the supply side of the robotaxi market will consolidate into an oligopoly dominated by a few key players or a more fragmented AV ecosystem.

Even if Mobileye’s open-supplier model gains traction and fosters a broader base of fleet operators, potentially preserving a more meaningful role for platforms like Uber, the market’s underlying structure will still change. Technology enablers that solve the toughest autonomy challenges will be best positioned to capture outsized economic rents, given the industry’s technological and capital barriers to entry.

Subscribe now

Uber’s Strategy for the Robotaxi Era

Uber is essentially hedging its bets by pursuing several different partnership models to prepare for and compete in a rideshare market where driverless robotaxi services will account for an increasing share of rides.

In light of these shifting dynamics, Uber’s adoption of a pragmatic and flexible strategy makes sense. The company seems to be pursuing three primary paths to position itself in the evolving robotaxi landscape. First, Uber is partnering with AV technology companies that directly operate robotaxi services on its platform in select cities, such as its existing tie-up with Waymo and its announced partnership with WeRide. Second, Uber is collaborating with partners to deploy robotaxi fleets powered by third-party AV technology suppliers on its marketplace, for example its cooperation with VW to launch a driverless rideshare service in Los Angeles using Mobileye’s autonomous systems. Finally, if Uber moves forward with an investment in an AV technology developer or robotaxi operator, it would mark a third strategic path aimed at addressing the growing competitive threat posed by autonomous vehicle rideshare services.

A direct investment in an autonomous vehicle technology company represents an approach that would offer Uber potential rewards if the AV technology succeeds, but also expose it to technology risk and, depending on the investment structure, create potential conflicts with other partners. Additionally, any significant investment in a more capital intensive robotaxi business could over time represent a departure from Uber’s asset-light business model and expectations that it will return the majority of its free cash flow to shareholders, factors that suggest Uber will proceed cautiously with such a move.

While agreements like Uber’s partnership with Waymo, where it provides fleet management services in Austin and Atlanta, help create stickier relationships with robotaxi operators, they also carry the risk of disintermediation as these partners gain scale and market leverage. Today’s partner could become tomorrow’s competitor. For Waymo, this partnership primarily accelerates its go-to-market strategy by outsourcing fleet management to a third party, while Uber benefits by staying relevant in the robotaxi space. Moreover, this relationship is not exclusive nationwide; Waymo’s separate partnership with Moove for fleet management in Phoenix and Miami illustrates its broader strategy of working with multiple partners across markets to support its rollout to more cities.

In contrast, fleet operators deploying vehicles using third-party AV technology on Uber’s marketplace could offer a more balanced and mutually beneficial relationship—one that is likely to provide Uber with the most favorable long-term opportunity to capture a greater share of the addressable market and benefit from growth driven by robotaxi services. Given the high fixed costs of owning and operating robotaxi fleets, maximizing vehicle utilization will be essential for achieving better unit economics. However, even if companies like Mobileye emerge as successful third-party AV technology suppliers, questions remain about whether the fleet operator landscape will stay fragmented enough for rideshare platforms to maintain a significant role —or whether a more consolidated supply structure will emerge, enabling robotaxi operators to establish direct-to-consumer relationships and reduce their reliance on third-party marketplaces.

Evolving Competitive Landscape

Ultimately, Uber’s future competitive standing and market opportunity in an AV-driven landscape will hinge on two key factors: the industry’s supply structure and the speed of the transition from a supply-constrained robotaxi market to one where falling prices drive positive price elasticity and market expansion.

Early on, robotaxi supply constraints will limit the need for third-party marketplaces, as demonstrated by Waymo’s direct to consumer service in San Francisco. For Uber to offset inevitable market share losses to direct to consumer robotaxi apps, overall market growth and ride volumes will need to expand enough to compensate for declines in both share and, over time, revenue per mile.

However, in the initial phases of robotaxi commercialization, limited vehicle availability will likely keep prices at or above those of Uber and Lyft, curtailing the potential for price-driven market expansion. Without lower prices to stimulate rideshare growth, combined with the novelty of a driverless ride, market share is likely to shift away from Uber and Lyft. This would be bad news for Uber and Lyft’s stocks as concerns mount that, in an autonomous world, they will become net market share losers. Only once fleets scale and increased supply drives prices down will market expansion become large enough to potentially offset rideshare marketplaces’ reduced share and declining per-mile revenue.

Frequent Swings in Sentiment

Investor sentiment on whether robotaxis represent a threat or an opportunity for Uber has repeatedly shifted with developments ranging from changes in Tesla’s robotaxi expectations, Waymo’s successes, optimism around Uber’s partnership announcements, and reports of Waymo gaining meaningful share in markets like San Francisco.

Nonetheless, the competitive implications of expanding robotaxi services continue to cast an intermittent yet persistent overhang on Uber’s stock and valuation, much like how fears of an AI-driven shift in search have pressured Alphabet’s multiple. Should Uber’s mobility gross bookings growth continue to decelerate while Waymo scales and expands to new cities, concerns over Uber’s terminal value will likely intensify.

On June 19, an executive order was signed to delay what is commonly referred to as the “TikTok ban” for the third time, extending it by 90 days and directing the Department of Justice to take no action to enforce the Protecting Americans from Foreign Adversary Controlled Applications Act. As this issue remains unresolved, it is failing to receive the attention it warrants, instead becoming lost in the daily churn of the news cycle.

Despite the lack of a clear legal basis for these repeated extensions, there have been no legal challenges even though TikTok poses significant risks as a platform ultimately controlled by a foreign adversary.

The app’s widespread popularity and the unfortunate impact that enforcing this legislation would have on content creators who monetize their work through TikTok do not justify allowing China to retain control over what is effectively a digital Trojan horse. TikTok’s algorithm and access to user data can be leveraged to amplify soft propaganda narratives that advance China’s objectives by shaping U.S. public opinion on critical issues, ultimately influencing individual voters and potentially undermining U.S. interests.

While discussions continue about a potential brokered deal to sell TikTok to U.S. interests in compliance with the law, the feasibility of such an outcome remains unclear, as it is questionable whether the Chinese government would approve a sale. Notably, ahead of the original deadline set by the legislation, ByteDance, TikTok’s parent company, made no discernible effort to pursue a sale. This original reluctance to sell to a U.S. entity raises questions about the platform’s underlying purpose and objectives.

Whether this legislation is enforced or TikTok is sold to U.S. interests, the outcome will meaningfully impact the social media competitive landscape. If TikTok exits the U.S. market, platforms like Meta’s Instagram Reels, YouTube Shorts, and Snapchat would gain from an influx of user traffic and advertising dollars. Conversely, a sale to a U.S. buyer that preserves TikTok’s scale could intensify competition, as new owners may invest aggressively to drive growth heightening competitive pressures to the detriment of other players.

Subscribe now

Misplaced Priorities

Prioritizing social media convenience over national security interests reflects a political establishment and the Trump administration’s current approach that lacks a sense of proper priorities. This stance directly contradicts the administration’s own core mantra of putting “America First.” If TikTok is not sold to a U.S. buyer and enforcement continues to stall, it will require uncommon political courage and bipartisan effort from leaders to challenge the administration, educate the public on the importance of making sacrifices for the greater good, and ensure the law is enforced.

At the same time, many Americans remain indifferent to the breadth of China’s digital propaganda campaign, failing to recognize that this threat is fundamentally different from concerns about bias or censorship on U.S. controlled social media platforms, as it is designed to advance the interests of a foreign state actor.

This indifference overlooks a long history of Russian and Chinese efforts to shape and influence U.S. public opinion. For example, Russia and the former Soviet Union have a well established record of using active measures and influence operations to undermine the U.S. socially, economically, and politically, fueling the divisions we experience in the United States today. TikTok is simply a modern, technology enabled version of this playbook.

Confronting the Digital Threat

U.S. social media platforms, while imperfect, differ fundamentally from a platform ultimately beholden to the Chinese government, whose political aims often stand in direct opposition to U.S. interests. Although measures to counter this threat may raise concerns about free speech and market interference, it’s important to recognize that China operates under an entirely different set of rules. The Chinese government exploits America’s openness through digital soft propaganda campaigns designed to advance its interests and influence U.S. public opinion. Beijing understands that controlling information is essential to maintaining its grip on power and is applying this same strategy abroad to expand its influence.

Just as U.S. internet firms face restrictions and limitations in China, the Protecting Americans from Foreign Adversary Controlled Applications Act is, in spirit, consistent with the constraints that have prevented U.S. internet giants from operating in China. These measures mirror broader restrictions China imposes on foreign businesses and go even further given the country’s lack of free speech and strict information controls.

Addressing Free Speech Concerns

The Supreme Court unanimously upheld the Protecting Americans from Foreign Adversary Controlled Applications Act, rejecting arguments that it violates free speech rights. The act specifically targets entities that pose a threat to U.S. national security, not the content of individual users. It limits the ownership and control of certain platforms, rather than restricting the speech of Americans who use them.

It is hypocritical for TikTok to claim that this statute violates free speech when access to the U.S. version of TikTok is itself restricted in China alongside other major U.S. social media platforms.

Moreover, research has shown that TikTok actively boosts and suppresses certain viewpoints. A study by Rutgers University’s Network Contagion Research Institute found that TikTok's algorithms amplify pro-China viewpoints while suppressing content critical of China.

Reframing the Real Issue

Speculation over whether the Trump administration’s delays in enforcing the TikTok legislation are motivated by U.S. investment interests in ByteDance or political considerations, such as a key donor’s stake or TikTok’s perceived electoral value, distracts from the core issue. The real concern is the national security threat posed by China’s ability to use TikTok as a tool for digital propaganda within the U.S.

U.S. restrictions on TikTok’s ownership are neither unprecedented nor extreme; they simply mirror the same type of barriers China has long imposed on American firms. Framing these measures as unreasonable or protectionist overlooks the reality of asymmetric digital policy between the two countries. Instead, they should be seen as a necessary response to level the playing field, protect U.S. national security interests, and limit the ability of a geopolitical rival to exploit open American markets while denying reciprocal access.

Following a year of mixed performance, Nvidia’s stock surged to an all time high this past week. With many of the factors fueling investor concerns largely diminished for now, the stock has room to run in the near term.

China-sourced revenue had been a key concern weighing on Nvidia’s stock, prompting the market to implicitly assign a lower multiple to earnings tied to China and driving much of the stock’s prior valuation compression. However, with this risk now reduced, investor focus is shifting back to Nvidia’s dominant market position, the growth tailwinds from the Blackwell product ramp, and its relative valuation, trading at a discount to large-cap semiconductor peers like Broadcom and Texas Instruments.

This opportunity comes amid an environment of revived animal spirits in the stock market and stretched valuations. Even so, demand for AI compute infrastructure remains robust, and semiconductor upturns (and cyclical downturns) often last longer than expected.

Prior to Nvidia reaching an all-time high, the AI trade has been alive and well, as demonstrated by the recent strong performance of stocks like AMD and CoreWeave. AMD’s rally has been fueled by growing optimism and anticipation around the company’s 2026 prospects, as customers look to counter Nvidia’s near-monopoly and support a second-source merchant AI chip supplier. This comes as Nvidia continues to dominate the market, capturing the vast majority of AI derived earnings with its expected total gross profit of nearly $140 billion this fiscal year.

In contrast, CoreWeave’s strong stock performance after its first earnings report reflects both speculative enthusiasm for AI plays and technical factors like a small free float and high short interest. However, concerns about its business model raised at the time of the IPO remain valid, as the company’s capital structure and rising debt load leave it vulnerable. If demand growth slows, competition intensifies, or service pricing declines intensify, CoreWeave’s business economics could unravel quickly. In the near term, the upcoming expiration of its IPO lockup period will add more shares to the market, as some pre-IPO investors are likely to monetize holdings at the current elevated valuation.

Against this backdrop, and considering the recent performance of other, less attractive AI stocks, Nvidia stands out with its dominant market position, superior earnings quality, and durable competitive advantages.

Subscribe now

Risks Ease as AI Demand Remains Strong

Over the past year, Nvidia’s stock has faced three primary concerns: risks tied to China-sourced revenue, slowing estimate revisions, and fears of cyclical overinvestment in AI compute capacity leading to oversupply. Currently, these risks have largely eased, with evidence pointing to a more constructive near term fundamental backdrop.

Better Estimate Trends: A key factor previously impacting Nvidia’s share price was the slowdown in the pace of positive estimate revisions. This underlying trend effectively reversed with the company’s late-May earnings report. Although July quarter revenue guidance was roughly in line with expectations, management highlighted that revenue would have been $8 billion higher without lost China sales, underscoring strong demand elsewhere.

At the same time, capex guidance from major cloud providers continues to trend higher, reflecting hyperscalers’ commitment to sustained, aggressive investment in AI. The recent uptick in the war for AI talent illustrated by Meta’s stepped up recruiting efforts further underscores the determination of major AI players to keep investing in ways that should support continued demand for AI compute infrastructure. Additionally, announcements made during President Trump’s recent trip to Saudi Arabia and the Gulf states have renewed focus on these regions as potential sources of strategic sovereign AI demand, investments likely to proceed regardless of cyclical conditions or shifts in the AI compute supply-demand balance.

China Overhang Removed: Until recently, China sourced revenue both direct and indirect via Singapore as a transshipment hub, together accounted for 31% of FY25 revenue and was Nvidia’s most tangible risk. But this overhang has mostly cleared. Last quarter, following the imposition of new export restrictions, Nvidia took a $4.5 billion charge related to H20 inventory and purchase commitments, below the up to $5.5 billion figure announced in April. The company will now exclude China from forward revenue and earnings guidance, removing this uncertainty from estimates.

Additionally, while the Biden administration’s prior Framework for the Diffusion of Advanced Artificial Intelligence aimed to cut off China’s access to advanced GPUs and could have potentially impacted Nvidia’s indirect sales to China, the Department of Commerce rescinded these controls in May, effectively lifting this China related revenue overhang. Although the Trump administration has yet to propose a replacement, signals suggest that access to AI systems may be used as a bargaining chip in foreign policy and trade negotiations, a potentially net positive for demand compared to expectations of the impact under the Biden administration’s policy framework.

AI Compute Capacity Digestion Risk: A cyclical correction in demand from AI compute supply growth outpacing demand isn’t a question of if, but when. However, current evidence suggests such a risk is not imminent, and may still be well into the future. Assessing sustainable AI systems demand growth is challenging, as capacity expands not only through new data center builds but also through efficiency gains in model design, system optimization, and GPU utilization. Algorithmic advances further reduce compute requirements, potentially accelerating AI adoption but clouding near term GPU demand signals.

While these efficiency gains are positive for Nvidia in the long run, they introduce timing risks: transitioning from training-heavy demand to inference-led workloads may not happen smoothly, creating oversupply if buildouts outpace realized demand. Compounding this risk, a notable share of GPU demand is coming from neocloud players, former crypto miners, small cloud operators, and VC-backed startups many of whom are building capacity ahead of solid customer pipelines. This speculative expansion, fueled by equity and debt financing often collateralized by GPUs, leaves them vulnerable if demand falls short, which could in turn curtail future GPU orders.

Additionally, a meaningful portion of AI data center customer demand comes from high cash-burn, AI startups rather than mature enterprises, raising questions about the sustainability of current growth levels. Nvidia CEO Jensen Huang acknowledged this dynamic on the August 2024 earnings call, noting: “the number of generative AI startups is generating tens of billions of dollars of cloud renting opportunities for our cloud partners.”

Signals to Watch

Current industry demand for AI infrastructure remains strong, providing a supportive backdrop for Nvidia and other key players. Looking forward, the resilience of AI demand growth, especially from less established sources of demand like neocloud players and VC backed startups, will be important in determining whether Nvidia can sustain its current growth trajectory or face a cyclical downturn if AI compute supply persistently outpaces profitable demand. Monitoring cloud and neocloud capex trends, along with signs of sustainable revenue pipelines versus speculative overbuilding, and the pace of adoption for inference workloads will be essential for assessing the durability of AI demand growth.

In this context, tracking GPU compute capacity spot pricing will provide early warnings of oversupply. A rapid shift in demand toward Nvidia’s Blackwell generation chips could leave existing AI compute capacity underutilized, pressuring pricing and eroding returns on recently built out infrastructure. Lower utilization rates could also constrain the cash generation needed by neoclouds and other buyers to fund future capex and ongoing procurement of Nvidia systems. Together, these indicators will be essential for assessing whether Nvidia’s growth trajectory remains on track.

The commentary from bulls and bears on the initial launch of Tesla’s robotaxi service is a clear example of the “Rashomon effect” where competing narratives shape different interpretations of the same event. The term "Rashomon effect" originates from Akira Kurosawa's 1950 film Rashomon, where four witnesses offer conflicting accounts of a murder. The film highlights how personal bias, self interest, and emotional state shape perception. The film ultimately resists presenting a definitive truth, emphasizing the subjectivity of human experience.

However, whether an autonomous vehicle meets the performance and safety standards required for an unsupervised, driverless robotaxi is not a matter of perception, it is grounded in objective reality.

Tesla’s recent robotaxi launch, while carefully staged, was limited in scope. It was an “invite-only” event for a small group of Tesla fans and influencers whose strong bias and loyalty to Tesla was evident in predictably enthusiastic reactions to their initial Tesla robotaxi rides. In contrast, Tesla critics seized on any miscues or safety concerns captured in the videos of these rides to make their case that Tesla’s robotaxi service is not ready for primetime.

One example of the differing assessments surrounding Tesla’s robotaxi launch is the presence of a Tesla employee or safety monitor inside the vehicle. Enthusiasts argue that the vehicle is operating without a safety driver, since the employee is seated in the passenger seat rather than behind the wheel. However, this positioning appears to be an attempt by Tesla to create the impression of a fully driverless system despite evidence to the contrary.

In multiple online videos, the Tesla employee is shown with their hand resting on what appears to be either a brake activation button or an emergency shut-down switch. Objectively, the presence of an individual with the ability to override or stop the vehicle is functionally equivalent to having a safety driver. In reality, this remains a supervised, not truly driverless robotaxi service.

This precaution may reflect a phased approach to scaling the service, but it also underscores the gap between the current operational reality and the promotional narrative surrounding Tesla’s robotaxi launch.

These publicly available videos when viewed alongside Tesla’s approach to autonomous vehicle development underscore the gap between Elon Musk’s bold promises and the current capabilities of the system. What is increasingly shaping up to be a muted robotaxi launch, coupled with the likelihood of a slower than expected expansion of Tesla’s robotaxi service, sets the stage for near term downside risk in the share price.

Subscribe now

From Table Stakes to True Autonomy

The launch of Tesla’s robotaxi service within a small, geographically fixed area offers limited insight into the broader capabilities of its autonomous vehicle system, particularly when the vehicles are confined to familiar roads and routes they have driven repeatedly. Like most autonomous vehicle systems, Tesla’s Full Self-Driving (FSD) performs better on roads it has driven many times before, benefiting from repeated exposure. However, this localized performance does not provide a true measure of the system’s ability to operate safely and reliably across diverse environments, varying conditions, or when encountering unexpected edge cases.

This dynamic highlights a fundamental aspect of AI training, repetition. Repeated exposure to data and feedback enables models to improve over time and achieve greater accuracy, something evident in autonomous vehicles repeatedly navigating the same routes. Yet while anecdotal reports from Tesla enthusiasts of consistent, flawless drives, such as commuting from home to the office, may seem impressive they represent only table stakes when it comes to evaluating the robustness of an autonomous driving system.

The more critical benchmark is whether the system can generalize its performance to new, unfamiliar environments and respond safely to unforeseen scenarios. This is where Tesla’s approach to autonomy and current version of FSD likely still falls short.

End-to-End Ambition vs. Proven Performance

As autonomous vehicle companies pursue different technological paths, a central debate has emerged between scalability and safety focused precision. Critics often point to Waymo’s limited operational domain as evidence of a lack of scalability. However, this is a deliberate decision prioritizing safety and reliability. Waymo’s system is not inherently limited to familiar environments; it is optimized for high performance within defined geographies but can also adapt to new and unforeseen scenarios, leveraging advanced neural networks to adapt in real time.

Tesla’s effort to develop a purely end-to-end neural network based autonomous system is aimed at generalizing across diverse driving environments, an approach if eventually successful, could enable rapid and highly scalable deployment. While an end-to-end system may with further technological maturity emerge as the most efficient and high performing solution, current data continues to show that compound AI systems, those that combine engineered elements with extensive use of neural networks, still deliver superior real world performance and significantly lower disengagement rates. As a result, the theoretical scalability of Tesla’s approach becomes largely irrelevant if the system cannot consistently meet safety and performance thresholds.

When Vision Alone Falls Short

Solving the highly complex challenge of developing a fully autonomous vehicle system requires significant technological breakthroughs and a willingness to think creatively, as system performance is influenced by a wide range of factors and variables.

In comparing Waymo and Tesla’s approaches to autonomy, many factors contribute to the performance gap between the two systems. However, two stand out: Tesla’s reliance on a vision only system versus Waymo’s use of a multi-sensor suite, and Tesla’s commitment to an end-to-end neural network approach compared to Waymo’s compound AI system that blends rule-based engineering with the use of neural networks. These foundational differences help explain why Waymo is steadily expanding its robotaxi service, while Tesla may face near term hurdles in scaling its competing platform.

The excerpt below from my post Tesla’s FSD Data Advantage is Overstated outlines the risk of over reliance on a single type of sensor and advantages of a more robust sensor suite. In a camera only system, if visual input is compromised by glare, rain, fog, low light, or visual occlusion there is no complementary sensor data (from radar or lidar) to reconcile erroneous inputs or cross check signals within a vision only architecture. This creates a vulnerability, especially in safety critical edge cases. This was evident when Tesla removed radar from its vehicles, after which many drivers began reporting instances of phantom braking while FSD was engaged.

Beyond sheer data volume, data diversity is important to advancing performance in autonomous vehicle systems, not just the breadth of scenarios, edge cases, and environmental conditions encountered, but also the variety of data sources contributing to the system’s perception. Diverse, high quality inputs enhance signal richness, enabling better model generalization and more robust decision making. This is where Tesla’s perceived data advantage begins to fall short.

Elon Musk’s decision to rely on a narrow sensor suite, cameras only, is based on the belief that because humans drive using vision and a biological end-to-end neural network system (the brain), replicating this model in autonomous vehicles is the most logical and efficient path forward. However, this assumption overlooks still existing challenges faced by an end to end system, and is inconsistent with the idea that not only the volume of data matters in the performance of an AI system, but the richness and diversity of signal is crucial as well.

Unlike human drivers, camera-based systems do not perform reliably in all environments or driving conditions. Adverse weather, glare from the sun or car headlights of an oncoming vehicle, and visual obstructions can degrade camera input and compromise perception quality…..These persistent performance bottlenecks suggest that a richer, multimodal sensor suite incorporating radar and lidar can provide a more comprehensive and robust world model.

A more robust sensor suite helps overcome the limitations inherent to any single type of sensor in an autonomous vehicle system…..Additional sensors not only introduce redundancy and enhance safety but also deliver more accurate and stable perception capabilities. They allow the system to extract crucial environmental data such as the position, velocity, and classification of road agents (vehicles, pedestrians, cyclists) with greater precision…..This improved perception feeds into a autonomous vehicle systems, enabling them to learn more effectively and directly infer appropriate driving actions from higher quality data.

Challenges for End-to-End Systems in Dynamic Road Scenarios

Relying solely on ever larger datasets and compute power does not guarantee that end-to-end systems can effectively handle unforeseen edge cases. Tesla’s pursuit of a purely end-to-end autonomous solution continues to face technical challenges in managing real world variability, particularly in unpredictable or novel scenarios, which may take considerable time to overcome.

In contrast, Mobileye and Waymo employ compound AI systems, which intentionally introduce engineered bias to reduce generalization error. Generalization refers to a model’s ability to perform well on previously unseen data. The idea is that even with reduced flexibility, this architecture results in a lower overall error rate than a fully end-to-end system.

End-to-end systems route raw sensor data directly into control outputs, avoiding engineered bias but potentially overfitting to the training data leading to higher generalization error and more frequent disengagements when encountering previously unseen edge cases. The performance of these systems may also be constrained by the demands placed on the onboard computer and memory, since compensating for the higher generalization error often exhibited by end-to-end systems requires processing vast amounts of data.

In addition to overfitting, another issue with end-to-end systems is causal confusion, when a model mistakenly interprets correlation as causation leading to flawed behavior and performance in unfamiliar situations. Compounding this is the issue of catastrophic forgetfulness, where newly introduced data causes the system to overwrite previously learned information. This is particularly problematic for autonomous systems that must continuously adapt to dynamic and unpredictable environments.

Will Tesla Change its Approach to Autonomy

At this stage, there is no evidence to suggest that Tesla intends to fundamentally shift away from its end-to-end, vision only strategy. Elon Musk has repeatedly framed this approach as not only the most elegant but also the most scalable, drawing comparisons to how humans drive. Yet if any of the limitations of this architecture hinder the rollout of Tesla’s robotaxi service, pressure to pivot may grow, particularly as rivals such as Waymo continue to scale and expand their services to more cities.

Whether Tesla adjusts its approach will likely depend on a combination of commercial pressure, performance data, and investor sentiment. For now, Tesla remains committed to its current path, but if a gap between narrative and operational reality persists, so too does the question of whether in the near term the current path will deliver a truly scalable and consistently safe robotaxi platform.

Storm clouds continue to gather over U.S., European, and Japanese auto and industrial semiconductor suppliers, as risks to their China related revenue streams become increasingly acute. In the near term, these companies are experiencing signs of a cyclical recovery following the end of a prolonged downturn and period of under shipment relative to end market demand. However, over the medium to long term, intensifying competitive pressure from Chinese firms bolstered by state driven industrial policy will weigh on these companies’ earnings power and valuation multiples. From a stock valuation and exposure standpoint, Texas Instruments and ON Semiconductor stand out as two of the least attractive names in the group.

Last week, Nikkei Asia reported that China’s Ministry of Industry and Information Technology (MIIT) has raised its policy objective, setting a new goal of achieving full self-sufficiency in the development and production of automotive semiconductors by 2027. This marks a notable increase from its earlier target of achieving 25% self-sufficiency.

While full localization is not a mandated requirement and will likely prove overly ambitious in the near term, the policy direction is clear. Chinese automakers, including SAIC, BYD, Geely, FAW, and GAC, are proactively aligning their procurement strategies with national policy, accelerating the displacement of foreign suppliers.

Leading analog, microcontroller, and power semiconductor firms derive approximately 20–35% of their revenue from China. However, the market’s strategic value extends beyond top line exposure. China has outpaced other regions in electric vehicle (EV) adoption and deployment of advanced driver assistance systems (ADAS), making it a key engine of global growth for auto semiconductors.

The auto and industrial semiconductor markets are far from monolithic. These markets encompass a wide range of chips and applications, spanning from highly customized components to more standardized, commodity like products. While this complexity suggests that Chinese suppliers share gains will play out over a number of years, the trend is already underway and the pace of market share loss for foreign players could be faster than many expect.

A common counterpoint is that the rate of Chinese chipmakers share gains will be limited by inferior product quality, especially in segments requiring higher component complexity. However, in a policy driven system like China’s, where official industrial targets carry significant weight, a “good enough” product, particularly if priced below foreign competitors, can be sufficient to gain market share. Moreover, Chinese companies have a history of surpassing expectations in how quickly they can make technical advances and move up the quality curve across other industries.

Mobileye’s recent experience in China illustrates this dynamic. Despite offering a higher performing system on a chip for advanced driver assistance systems, particularly in high speed driving scenarios, the company has lost significant market share to local competitors whose chips were reportedly less advanced but more cost effective. In a market where policy incentives and price competitiveness dominate, even technically less robust products can displace established leaders.

The implications of rising Chinese competition will not be uniform across the semiconductor sector. Gross margin serves as a useful proxy for comparing and contrasting companies’ product differentiation, pricing power, and vulnerability to competition. Firms with high value added products reflected in stronger margin profiles are better insulated from market share loss and pricing pressure. For example, Analog Devices (ADI), with a gross margin of approximately 70%, benefits from significant pricing power and specialized offerings, making it less susceptible to displacement. In contrast, companies like ON Semiconductor, with a gross margin of now less than 40%, are more exposed with a higher mix of lower value added products and weaker pricing leverage placing them at greater risk as Chinese players gain share.

Subscribe now

One Firm’s Growth Is Another’s Decline

Ironically, the competitive rise of Chinese chipmakers has been enabled by U.S., Japanese, and Dutch semiconductor production equipment manufacturers. While U.S. export controls have restricted China’s access to advanced node manufacturing tools, they have not limited the sale of equipment used in “lagging edge” process technologies, the dominant nodes used in the production of automotive and industrial semiconductors.

This regulatory gap has allowed Chinese semiconductor firms to expand their capability and capacity in precisely the areas where companies like NXP, ON Semiconductor, Texas Instrument, and Infineon are most exposed. As a result, foreign suppliers now face intensifying competition facilitated by equipment provided by U.S., Japanese, and Dutch toolmakers as Chinese firms scale production targeting the automotive, industrial, and consumer electronics end markets.

By early 2024, global semiconductor equipment sales to China had more than doubled, accounting for over 40% of total revenue, up from less than 20% in late 2022. Some argue that extending export restrictions to lagging edge equipment would be excessive and forgo valuable sales. However, the revenue gains of semiconductor equipment companies come at the expense of market share and revenue for firms like Texas Instruments, ON Semiconductor, and NXP, as Chinese competitors, backed by state subsidies, expand production not to generate profit, but to fulfill the government’s strategic self sufficiency objectives.

This is a familiar playbook. Across industry after industry, China has prioritized national strategic goals and production output over profits, leading to global overcapacity and ultimately downward pressure on industry economics. Even if Chinese semiconductor capacity remains focused on the domestic market, its global impact will be material. China is projected to account for 39% of mature-node foundry capacity by 2027, up from 31% in 2023.

The Global Ripple Effect of Rising Chinese Competition

While the long term threat posed by China is gaining attention, the full magnitude of the challenge remains underappreciated. As detailed in my post China’s Subsidized Silicon Strategy, Chinese chipmakers benefiting from a debt-fueled industrial policy and state subsidies are rapidly scaling production while operating under a fundamentally different set of rules. This dynamic distorts market pricing, undercuts foreign competitors, and poses a growing threat to U.S., Japanese, and European semiconductor firms.

With China as the world’s largest semiconductor market, even a gradual displacement of foreign suppliers will erode China sourced revenue for these firms and trigger a ripple effect beyond China. As companies seek to offset lost revenue, competition will intensify, further pressuring market dynamics worldwide.

Auto Chip Risk from China’s EV Rise

For U.S., Japanese and European auto semiconductor companies, the challenge extends beyond direct competition with Chinese chipmakers. A second order effect is emerging: the global rise of Chinese EV OEMs, particularly in Europe, poses an additional threat. These vehicles will be increasingly built using domestic Chinese components, including semiconductors, further displacing incumbent suppliers.

While China has developed globally competitive EVs, the sector’s rapid growth has been heavily subsidized and supported by a financial system that prioritizes production over profitability. This has enabled Chinese automakers to scale aggressively, leading to overcapacity and aggressive pricing, conditions that would not be sustainable in market based economies. The result is a structural distortion that amplifies competitive pressure on Western and Japanese semiconductor and automotive firms alike.

Companies like NXP and ON Semiconductor with over 50% of revenue tied to the automotive sector are particularly exposed as Chinese competitors gain market share not through pure market based competition, but with a significant advantage from an economic and financial system engineered to fulfill national strategic policy objectives. As Chinese firms scale, the pattern is consistent: they first displace foreign competitors within the domestic market. Over time, as excess capacity accumulates, a recurring feature of China’s economic model, that production spills into global markets, intensifying competitive pressures and undermining pricing power across the industry.

China Inc.’s Debt Driven Growth Model

To achieve its economic policy goals, the Chinese Communist Party (CCP) relies heavily on credit as its primary lever of economic control, channeled through state-owned banks, local government financing vehicles (LGFVs), and non-bank financial institutions. Under Xi Jinping’s leadership, the financial sector has become even more politically driven, as strategic objectives increasingly take precedence over profitability.

While state directed industrial growth has long been a feature of China’s economy, Xi Jinping’s leadership has intensified this model expanding the supply side policy framework to accelerate technological self reliance and reduce dependence on foreign suppliers. Semiconductors and EVs are both at the center of this strategy. In industry after industry this approach has led to excess investment and structural overcapacity.

Reflecting the consequences of China’s economic policy model, a report from the European Central Bank highlighted that the share of loss making Chinese industrial firms has doubled since 2018, reaching 28%. This trend, coupled with a concurrent sharp rise in the inventory to sales ratio among industrial firms, highlights the deflationary pressures mounting in sectors burdened by severe overcapacity within a financial system where loss making companies are often shielded from typical market-based consequences.

The effects of China’s excess industrial and manufacturing capacity are not limited to its domestic market. Increasingly, China is exporting its surplus production, amplifying the impact on global markets. This export-driven strategy is a critical pillar of China’s economic growth and one that is unlikely to reverse in the near term. As outlined in my post Trade War: China’s Economic Fault Line, this approach is also closely tied to China’s reliance on its trade surplus as a key source of domestic liquidity and economic growth. According to the Mercator Institute for Chinese Studies, China’s global trade surplus reached $992 billion in 2024, equivalent to 5.6% of GDP, with net exports accounting for 30% of GDP growth.

Headwinds on the Horizon

Just as many analysts and investors underestimated the extent to which auto and industrial chip suppliers were over-shipping during the last cyclical upturn, many now appear to be underestimating the competitive threat posed by emerging Chinese players. This risk is being obscured by near term optimism, as management teams offer more constructive commentary while exiting a period of under-shipping relative to end market demand and a likely boost to revenue from demand pulled forward in response to tariff uncertainty.

Once this phase of revenue mean reversion passes, and as Chinese competitors continue to gain traction, the longer term implications will come into sharper focus. A structural shift in the competitive landscape is likely to drive a valuation derating for U.S. and European auto and industrial semiconductor stocks relative to the broader market.

The impact, however, will vary by company. Texas Instruments, for example, trades at more than 30x forward earnings leaving it more exposed to absolute downside risk. In contrast, while NXP is also vulnerable given its significant exposure to China and the auto sector, some of this risk appears to be priced in. The stock currently trades at 15x forward earnings, a discount to the S&P 500, suggesting a greater likelihood of relative underperformance rather than absolute downside in the near term.

Many internet sectors exhibit “winner-take-most” dynamics, where a dominant player captures a disproportionate share of industry profits. These outcomes are typically driven by powerful network effects and scale advantages that compound over time, creating a widening competitive moat for the market leader. Netflix appears to fit this profile.

Netflix remains the undisputed leader in global streaming, underpinned by its scale, content strategy, and highly profitable business model. Its strong financial position enables a self-reinforcing flywheel: higher profitability funds greater investment in content, talent, and platform innovation supporting pricing power and long-term engagement.

However, streaming is structurally different from many other internet sectors. Despite its dominance, Netflix accounts for only 8–9% of total U.S. video viewership, reflecting the fragmented nature of video consumption and the diversity of content preferences. This leaves meaningful room for several winners to emerge.

In this evolving landscape, Disney is well positioned to solidify its place as the No. 2 global streaming platform. As the market continues to transition from the legacy cable TV bundle to a re-bundled, platform driven streaming ecosystem, Disney stands to benefit from industry consolidation and cross-platform bundling partnerships. Its rich portfolio of intellectual property, multi-platform reach (Disney+, Hulu, & ESPN), and growing scale place it in a strong position to capitalize on this shift.

Success in streaming has significant implications for Disney’s stock valuation, with the potential to drive meaningful multiple expansion. The company's direct to consumer (DTC) business is approaching a key inflection point: streaming operating income is expected to surpass that of the linear networks segment in 2026, easing investor concerns around the secular decline of its legacy media business. As streaming becomes a larger share of total earnings, the market will increasingly benchmark the value of Disney’s media business against Netflix.

Disney’s stock has posted a solid run since its earnings report in early May, yet zooming out, it remains meaningfully below its 2021 peak. This long stretch of underperformance has been driven by structural declines in linear media and, until recently, mounting losses in streaming. Even as the Parks and Experiences segment has delivered consistent growth compounding operating income at an approximately 10% annual rate from FY22 through FY25, investor sentiment has remained focused on concerns around the durability of Disney’s legacy media assets and the path to sustainable streaming profitability.

Currently trading at approximately 11x 2026 EV/EBITDA and at nearly a 15% discount to the S&P 500’s forward P/E, Disney’s valuation remains undemanding. This reflects a market still anchored in a backward-looking view of the company's earnings profile. But with improved execution, continued DTC subscriber growth, and sustained margin expansion in streaming, a meaningful re-rating becomes increasingly plausible.

Netflix has already demonstrated that scaled, profitable streaming platforms with high engagement and global reach are rewarded with premium valuation multiples. Disney is on the cusp of joining that club. If management delivers, Disney’s streaming success could not only drive upside to earnings expectations, but also justify a structurally higher multiple leading to sustained stock outperformance.

Subscribe now

Earnings Clarity: Results Marked a Turning Point

Disney’s narrative is now beginning to shift. March-quarter earnings showed broad-based progress with EPS coming in ahead of expectations and the company raising its full-year guidance. Importantly its direct to consumer (DTC) segment subscriber growth and profitability were both better than expected. Overall, revenue in the Experiences segment was also encouraging, though the mix differed somewhat relative to expectations. Domestic theme park revenue was resilient with stronger than expected 9% year-over-year growth, while international park revenue declined by 5%.

Notably, the tone of the earnings call reflected a more disciplined approach to investor communications. Following several missteps last year, including lower than expected guidance, and the departure of the investor relations head, Disney’s new CFO has adopted a more conservative and consistent guidance framework. That discipline raises credibility and sets up the potential for positive surprises as execution improves.

DTC at an Inflection Point

The most significant driver of Disney’s evolving investment narrative is the growing profitability and scale of its direct to consumer segment. DTC revenue is now nearly triple that of linear networks, and operating income is on track to surpass linear by next fiscal year, a pivotal earnings mix shift. After years of being a drag on results, DTC is set to become a core growth engine. This shift in earnings mix is central to the bull case for a re-rating and the company’s clearest path to multiple expansion.

Several tailwinds are fueling renewed optimism. Increased bundling of Disney+, Hulu, and ESPN along with any new platform partnerships, such as the one with HBO Max , will drive additional subscriber growth, reduce churn, and boost margins. In the U.S. and Canada, bundling represents an untapped revenue growth opportunity, with approximately two-thirds of subscribers currently taking only one Disney owned streaming service. The fall 2025 launch of ESPN as a direct to consumer service is a significant catalyst to drive additional revenue and bundling opportunities, leveraging one of the most powerful brands in media with preexisting, built in demand. Additionally, although a crackdown on password-sharing does not present an opportunity comparable to Netflix, Disney’s efforts to reduce password sharing across Disney+ and Hulu will also boost subscriber growth and generate high margin incremental revenue. Growing connected TV (CTV) ad revenue and still underpenetrated international markets offer long term upside as well.

Last quarter, Disney’s DTC business featured an operating margin of only 5.5% compared to Netflix’s 31.7% margin in the same quarter. While Disney may not achieve a level of profitability comparable to Netflix, the current gap in margins reflects the opportunity for sustain margin expansion underscored by improved cost discipline and double digit revenue growth. Meaningful operating leverage is poised to drive upside to FY26 DTC operating income, with potential to reach $3 billion—well above current consensus expectations, yet still representing only a 10% operating margin.

Hulu Valuation Overhang Removed

Disney will pay Comcast an additional $439 million for the remaining stake in Hulu, completing a deal initially valued at $8.6 billion. This finalizes the appraisal process and gives Disney full control of the platform. Arguably, it had been in Disney’s interest to manage Hulu’s growth and profitability conservatively to avoid inflating the final purchase price. Any such constraint is now lifted.

With the uncertainty around the valuation resolved, Disney will continue to fully integrate Hulu into its broader DTC strategy. Further integration could unlock additional synergies across tech infrastructure, sales and marketing, and content development, while a unified bundling strategy continues to present an opportunity to boost engagement, reduce churn, and accelerate revenue growth.

Experiences: Cyclical Risk vs. Long Term Growth

The Experiences business segment will continue to represent a meaningful share of Disney’s earnings (nearly 60% of FY25 operating income), however the core narrative driving the stock is set to transition from theme parks to streaming. Although cyclical risks to discretionary consumer spending remain an overhang for domestic parks, international parks should benefit from improved attendance and pricing. In the second half of 2025, year-over-year comparisons will ease as Disneyland Paris laps the 2024 Olympics in the September quarter, while U.S. parks benefit from easier hurricane-impacted comps in the December quarter. Additionally, the cruise ship business continues to expand with one new ship already launched and two more arriving by end-2025.

Parks and Cruises continue to offer a unique and differentiated guest experience and are expected to remain a steady source of long term growth. While cyclical risks persist in the Parks segment, any such risk is transitory. Furthermore, if softer discretionary consumer spending negatively impacts the Experiences segment's earnings in FY26, any such weakness will at least be partially offset by expected strength in the streaming segment.

Even if near-term softness emerges in the Parks business, Disney's current valuation, particularly given its discount to the S&P 500, makes it more likely that the market will look past short term headwinds and focus instead on the company's strengthening position in streaming. This dynamic is not without precedent. In 2020–21 Disney's stock reached an all time high despite widespread park shutdowns during COVID as investors looked through the immediate disruption and awarded the company a higher multiple based on expectations for long term streaming growth, though in hindsight, those expectations proved premature.

Conclusion

Disney’s stock narrative is undergoing a meaningful transformation. The shift from a legacy media operator to a scaled, globally relevant streaming platform is reaching an inflection point. While near term macro uncertainty may persist, the company’s earnings mix is evolving, and its long term trajectory is improving.

With a still undemanding valuation, improving sentiment, and a stock that remains under-owned, the setup is compelling. As streaming becomes a key earnings driver and execution continues to improve, Disney looks increasingly poised for a re-rating, with the risk/reward skewed to the upside.

Tesla's long anticipated robotaxi service launch in Austin is fast approaching. Although the dramatic deterioration in Elon Musk’s relationship with Donald Trump could prove detrimental to Tesla, introducing a new, unpredictable factor that further increases volatility in Tesla's stock, the success or failure of the robotaxi service is a more measurable and crucial long term value driver.

The significance of Elon Musk’s ability to meaningfully shape federal regulations around autonomous vehicles and robotaxis has likely been overstated. Regulatory favor alone is insufficient to drive industry growth or confer a lasting advantage. Ultimately, such influence is irrelevant unless Tesla and its competitors can first overcome technical challenges and successfully develop a scalable, safe, and commercially viable autonomous vehicle system.

As I’ve noted in my previous post Autonomous Vehicles: A Watershed Moment, Tesla’s approach to autonomy still faces technical limitations. Ongoing data showing frequent disengagements in Tesla’s Full Self-Driving (FSD) system underscores that the technology is likely still not ready for large-scale, fully driverless robotaxi deployment. Moreover, some commonly cited advantages in the Tesla robotaxi narrative, such as access to a large volume of data from its vehicles, have been overstated and do not necessarily translate into a superior system. That said, Tesla may still succeed in orchestrating a small scale launch that appears successful on the surface, drawing significant public attention and generating positive commentary from Tesla aligned influencers.

Initial reports suggest the service will include only approximately 10 vehicles operating on well defined, familiar roads, conditions that will help minimize disengagements and improve perceived performance. Tesla is also expected to lean heavily on teleoperators and limit early usage to “invite-only” participants, likely including influencers and Tesla friendly voices who can amplify a positive narrative around service quality and the broader promise of Tesla’s perceived superior position to disrupt personal vehicle ownership.

Given the recent decline in Tesla’s share price, even a tightly managed but visually successful launch will likely fuel a rebound in the stock. However, once the initial excitement fades and EV sales continue to lag, investors will again confront the gap between Musk’s ambitious promises and the operational and technical reality. This sets the stage for renewed downside risk in the stock.

Tesla’s unsupervised Full Self-Driving (FSD) system, built around an end-to-end neural network architecture continues to face inherent technical challenges in handling real-world variability and edge case scenarios. These limitations raise questions about the system’s readiness for broad deployment. Compounding the issue is Tesla’s uncertain go-to-market strategy and, based on limited company disclosures, the potential near-term absence of the operational infrastructure required to support a scalable, multi-city robotaxi service. As these shortcomings become increasingly evident, especially when contrasted with the company’s promotional narrative, the risk of a more meaningful correction in the stock price grows.

Subscribe now

The Misconception of a Data Volume Advantage

There is a common but flawed assumption that simply having more data and superior training compute power is sufficient to develop a successful autonomous vehicle (AV) system. This overlooks an important reality: once an AV system is trained, it is no longer continuously dependent on having access to large AI compute clusters for its everyday operation. At that point, system performance is defined by the efficiency and robustness of the vehicle level tech stack, not access to the cloud or large data center AI compute capacity.

Tesla is often credited with a significant data advantage, and indeed its access to real-world driving data is substantial. However, it is not alone. Mobileye, for instance, has harvested over 56 billion miles (as of 2024) of driving data through its Road Experience Management (REM) platform. In parallel, technical advances in simulation capabilities are making synthetic data an increasingly valuable complement to real-world data in training autonomous systems.

Beyond sheer data volume, data diversity is important to advancing performance in autonomous vehicle systems, not just the breadth of scenarios, edge cases, and environmental conditions encountered, but also the variety of data sources contributing to the system’s perception. Diverse, high quality inputs enhance signal richness, enabling better model generalization and more robust decision-making. This is where Tesla’s perceived data advantage begins to fall short.

Elon Musk’s decision to rely on a narrow sensor suite, cameras only, is based on the belief that because humans drive using vision and a biological end-to-end neural network system (the brain), replicating this model in autonomous vehicles is the most logical and efficient path forward. However, this assumption overlooks still existing challenges faced by an end to end system, and is inconsistent with the idea that not only the volume of data matters in the performance of an AI system, but the richness and diversity of signal is crucial as well.

Unlike human drivers, camera based systems do not perform reliably in all environments or driving conditions. Adverse weather, glare from the sun or car headlights of an oncoming vehicle, and visual obstructions can degrade camera input and compromise perception quality. Although there are some techniques to mitigate the impact of glare from bright light sources, this remains a challenge for a vision only system. These persistent performance bottlenecks suggest that a richer, multimodal sensor suite incorporating radar and lidar can provide a more comprehensive and robust world model.

A more robust sensor suite helps overcome the limitations inherent to any single type of sensor in an autonomous vehicle system. LIDAR offers precise, high-resolution 3D spatial mapping and excels at capturing the shape and position of objects across complex environments. It performs well across varied lighting conditions but remains costly and can experience performance degradation in bad weather conditions. Cameras, by contrast, are inexpensive and provide rich visual data, but are heavily dependent on good lighting and offer limited depth perception, making them less reliable in poor visibility conditions. Radar complements both by delivering performance in a wide range of environmental conditions, directly measuring object distance and velocity.

Additional sensors not only introduce redundancy and enhance safety but also deliver more accurate and stable perception capabilities. They allow the system to extract crucial environmental data such as the position, velocity, and classification of road agents (vehicles, pedestrians, cyclists) with greater precision.

This improved perception feeds into a autonomous vehicle systems, enabling them to learn more effectively and directly infer appropriate driving actions from higher-quality data. Richer sensor data helps reduce reliance on a single sensor by better contextualizing the driving environment, allowing the system to perform more reliably, even in complex or novel scenarios. This approach, combined with the engineered elements of its compound AI system, is Waymo's current strategy for autonomy. It integrates high-definition maps, modular AI frameworks, rigorous safety layers, and extensive use of neural networks to deliver a more robust and deployable platform.

The Consequences of a Vision-Only Approach

In the context of his "vision only" approach to autonomous driving, Elon Musk has repeatedly downplayed the value of sensor redundancy, viewing it not as a means of improving system safety and performance, but as a potential source of signal conflict between sensors. His argument is typically framed around a binary choice: that conflicting data from two different sensors could lead to worse outcomes than relying on a single source. However, the addition of a third, independent sensor can help resolve discrepancies and reinforce accuracy through sensor fusion, an established technique used in many leading AV systems.

Instead of developing a more robust sensor fusion system that supports multiple sensor types, Tesla has chosen a vision-only architecture, risking over-reliance on a single type of sensor. In a camera only system, if visual input is compromised by glare, rain, fog, low light, or visual occlusion there is no complementary sensor data (from radar or lidar) to reconcile erroneous inputs or cross-check signals within a vision only architecture. This creates a vulnerability, especially in safety critical edge cases.

An example of the consequences of this design decision emerged in 2021–2022, when Tesla initially transitioned all vehicles built for the North American market to its camera only Tesla Vision system, discontinuing the use of radar. Following this change, many drivers began reporting instances of phantom braking when FSD (Full Self Driving) was engaged. While correlation is not causation, these incidents are plausibly linked to the absence of radar, an important sensor for detecting the position and velocity of objects ahead, particularly in conditions where vision alone may be unreliable. Removing radar not only stripped the system of an additional safety layer but also eliminated a key input that could help mitigate false positives generated by vision based perception.

A real life driving experience cited in my post AV Technology: Facts vs. Fiction also provides a simple example of the advantages of utilizing a multi-sensor system -

A recent drive in a Tesla Model Y highlighted a key technical challenge for its autonomous vehicle systems. The drive was through a familiar suburban environment, and overall, the FSD system performed well, except for two disengagements.

One disengagement occurred when the vehicle struggled to navigate out of a parking lot. The second disengagement, however, was more noteworthy and informative. While driving through a suburban village, the vehicle came to a complete stop in the middle of an intersection despite a green traffic light. The vehicle’s halt seemed to be triggered by FSD detecting a pedestrian standing at the corner, waiting to cross the street. However, the pedestrian was stationary, the traffic light was green, and the pedestrian signal showed a red "Don't Walk." Despite these cues, the vehicle stopped unnecessarily. This is a good example of an autonomous system identifying an object in its environment but failing to understand the broader context needed to make an optimal driving decision.

Tesla’s vision only system correctly identified the pedestrian but lacked the contextual understanding to conclude that it was safe to proceed. A more robust sensor suite could have improved performance in this situation. Radar, for example, measures an object’s distance, speed, and relative motion, key data points that would have confirmed the pedestrian was stationary and not about to cross, prompting the vehicle to drive through the intersection safely.

Conclusion

The debate and narrative surrounding Tesla’s autonomous vehicle system development have consistently been dictated by Elon Musk’s optimistic forecasts coupled with effusive commentary and praise from Tesla supporters. However, these sources are not objective, nor do they provide concrete data to support bullish conclusions about the current state and readiness for deployment of unsupervised Full Self-Driving (FSD).

Frequent comments from Tesla owners, claiming their vehicles drive flawlessly with no disengagements in their daily commute, are misleading. A daily commute is a fixed route where many current systems can demonstrate solid performance. The true test lies in assessing a system's capacity to perform reliably in unfamiliar environments and when encountering unforeseen edge cases.

Likewise, the true test of Tesla’s autonomous technology and robotaxi service won’t come in the early days of its limited launch, but rather in the weeks and months that follow as the pace of scaling is likely to lag behind the optimistic timelines set by Elon Musk’s public forecasts.

Alphabet’s recent Google I/O developer conference helped ease investor anxiety about the threat generative AI and new players like OpenAI pose to Google’s core search business, however the broader debate remains far from settled. Simultaneously, ongoing antitrust litigation continues to cast a shadow of uncertainty over Alphabet's long term outlook.

The corrective action and remedies that may arise from these legal proceedings could pose a greater risk to Alphabet than competitive threats to its search business. However, absent a value destructive breakup or severe restrictions on its distribution reach, the company’s current valuation discount to the S&P Index is unwarranted. When considering the implicit valuation multiples investors assign to Google Cloud’s larger peers, Amazon Web Services (AWS) and Microsoft Azure, as well as YouTube’s strong market position and growing share of consumer engagement, Alphabet’s sum-of-the-parts valuation becomes even more compelling. Adding further to this case is the underappreciated optionality of Waymo, which stands at the forefront of physical AI and disrupting personal vehicle ownership. Taken together, the implied discount on Alphabet’s core search business is even deeper than headline figures suggest.

Although the legal outcomes remain uncertain, the competitive threat to Google Search can be assessed with more clarity. Comparisons made by some to past competitive disruptions, such as Netflix’s upending of legacy media, are often invoked to argue that Google faces a similar existential risk. However, these conclusions overlook critical distinctions and inherent competitive advantages that continue to favor Google.

It wasn’t long ago that Meta faced widespread pessimism over the impact of Apple’s iOS privacy changes and mounting losses in its Reality Labs division, yet today Meta is stronger than ever. Google now finds itself in a similar position. Investor sentiment is weighed down by fears of AI driven disruption, but the company is increasingly poised to go on offense. As the AI narrative shifts from infrastructure buildout to application-layer growth, Google is positioned to leverage its scale, technical depth, and ecosystem integration to drive user engagement growth.

Time will tell to what extent Alphabet’s leadership successfully navigate the wave of AI related disruption, but concerns rooted in the idea that Google faces a classic innovator’s dilemma may be overstated. The notion that Alphabet will sidestep certain vectors of innovation to protect its core search business overlooks that AI driven answer engines and agentic interfaces represent a transformation of search not a wholesale replacement.

Rather than cannibalizing its core product, Google is layering Gemini’s new capabilities into its existing platform to drive future growth, much like how Meta acquired and leveraged Instagram to adapt to shifting user behavior, appeal to different demographics, and increase overall engagement beyond Facebook. Gemini offers Google a similarly new brand and interface to adapt to evolving user behavior, launch new products, and unlock monetization opportunities. Google’s AI Overviews already has over 1.5 billion users per month and the company has recently launched AI Mode in Google Search to provide more in-depth and conversational chatbot like responses to queries. Google’s Gemini AI app also appears to be approaching a comparable scale to ChatGPT with 400 million monthly active users.

Moreover, unlike legacy industries such as automotive or traditional media where incumbents have been forced into costly dual track strategies, maintaining legacy operations (ICE vehicles or cable TV) while investing heavily in unprofitable new platforms (EVs or streaming), Alphabet does not face the same structural disadvantage. The company’s AI initiatives, particularly through Gemini, are integrated within its existing business and supported by a scalable infrastructure. While the AI era requires significant industry wide capital investment in compute capacity, Alphabet is positioned to absorb these costs without undermining its aggregate profitability. This strategic position allows Google to deploy and monetize AI applications at scale without the operational drag that burdens many traditional incumbents.

Although Google is positioned to counter emerging threats and capitalize on AI driven growth, the disruption of legacy media by video streaming still offers a useful lens to evaluate the potential impact of AI on search. It underscores the critical importance of scale, distribution, user behavior, and economic moats, factors that continue to tilt the competitive balance in Google’s favor.

Subscribe now

Enduring Distribution Advantage

The disruption of legacy media has been driven by a fundamental shift in distribution, from closed cable networks to the open architecture of the internet. This transition has enabled new business models and elevated the economic value of a broader range of content, expanding the addressable market for consumer attention as viewers are able to watch content anytime, anywhere.

Before streaming, the constraints of linear TV meant limited time slots for content, making premium, tentpole programming the key to success. On-demand and streaming services upended this model, creating an industry with virtually infinite content supply. Success in streaming is not only about a handful of marquee shows but about offering a breadth and depth of content that engages as many household members and demographics as possible.

While major, must-watch series remain important, Netflix has prioritized a large volume of “good enough” programming and more niche content that may not be a cultural phenomenon but keeps a wider range of subscribers engaged.

The anytime, anywhere nature of streaming and internet short form video content has not only driven higher aggregate video consumption, but also allows for a standalone platform to generate a consistently outsized share of total engagement as is the case with YouTube and Netflix.

However, unlike media companies that lost control over distribution when content consumption moved from closed cable networks to the open internet, Google continues to be a primary gateway to information online.

This distribution combined with its breadth and depth of customer touch points gives Google a significant advantage and positions it to harness new avenues of growth. With 15 products each serving over 500 million users and all of these products using Gemini models, Google benefits from an unparalleled ecosystem that can rapidly absorb and integrate new technologies. This strengthens both user engagement and monetization potential.

In sharp contrast, traditional cable TV channels largely lacked direct consumer relationships. When legacy media companies launched streaming services, they faced the costly and time consuming task of essentially building consumer bases from scratch. Google, by comparison, enters the AI era with billions of existing daily touch points across its global user base, giving it a formidable market position. While remedies from ongoing antitrust litigation may constrain its distribution reach, absent a more extreme outcome, Google’s business appears well equipped to absorb and adapt to such changes.

The Next Phase of Engagement Growth

The competitive threat to Google Search is not driven by a change in distribution channels as was the case with legacy media, but potentially by a shift in consumer behavior enabled by emerging technologies such as generative AI and AI agents.

The legacy cable TV bundle once provided a collective economic moat for media networks, sustaining engagement and ad revenue. The unraveling of that bundle through cord-cutting not only fragmented audiences but forced traditional media companies into a costly dual track strategy, maintaining declining linear TV businesses while building initially unprofitable streaming platforms. Outside of long dated sports rights, most legacy players lack durable moats around content production, leaving them increasingly exposed to the shift toward streaming and internet video. These dynamics led to shrinking cable profit pools and rising infrastructure and content costs exacerbated by the strategic misstep of reallocating a significant share of premium content from cable to streaming accelerating declines in the linear TV profit pool.

In contrast, Google benefits from a broad ecosystem of direct user touch points, extensive data advantages, and scaled AI capabilities that will power product development and drive engagement growth. While the composition of search related activity is shifting from traditional query-based search to answer engines and AI agents, the rise of AI presents Alphabet with more opportunity than threat, success is theirs to lose.

Although some cite increased personal use of ChatGPT and similar apps as well as recent data showing slower search traffic and paid click growth as evidence of structural headwinds for Google Search, this reflects a narrow snapshot of a rapidly evolving landscape. AI chatbots and agents are broadening the definition of search and expanding the total addressable market. The more relevant question is not whether Google’s share of this expanded market will be lower than its dominance in traditional search, it will be, but whether the broader market can still support aggregate revenue growth. With greater monetization potential from AI enhanced services and the ability to leverage its core competitive advantages to launch new applications, Google is favorably placed to sustain, if not accelerate, growth as the shift unfolds.

As the leading search engine across Android devices, Chrome, and Safari, Google captures billions of daily queries and remains deeply embedded in user behavior. This habitual use is reinforced by a tightly integrated suite of services including Gmail, Maps, Docs, and Drive that form a self-reinforcing ecosystem. While ChatGPT is growing rapidly, it lacks Google’s broad embedded presence across multiple daily-use applications that provide natural, established entry points for introducing new AI experiences.

Google’s ability to layer interfaces like Gemini across its existing platforms combined with its infrastructure and data advantages, not only strengthens its defense of core products but also unlocks new growth opportunities. Travel is a prime example. With unmatched visibility into user intent across Search, Maps, Gmail, and Google Flights, Google is uniquely positioned to deliver personalized, context-aware travel planning. Unlike standalone OTAs, it has the potential to embed AI agents directly into its ecosystem surfacing trip ideas, optimizing itineraries, and handling bookings across multiple touchpoints. Its data advantage enables more accurate recommendations and pricing visibility, positioning it to capture greater value across the entire travel funnel.

Ownership of YouTube also provides a differentiated advantage in the generative AI context. YouTube is not only a dominant platform for video content but increasingly serves as a discovery and how to engine, especially among younger users. Its rich mix of user-generated video data provides Google with a training edge for multi-modal AI and a highly engaging monetization surface. In a world where generative AI shifts user preference toward more dynamic responses, YouTube uniquely enables Google to meet that demand. YouTube’s comment threads and watch-time signals also feed into Google’s broader understanding of relevance and intent, further improving its AI systems.

Another critical advantage is Google’s superior monetization engine and efficient digital ad product with a proven ability to match high intent queries with highly targeted ads. Other than Meta, there are no comparable global platforms for advertisers to generate a comparable level of reach and return on ad spend. ChatGPT, by comparison, remains early in its monetization journey, with a current focus on a subscription based revenue model.

Conclusion

Taken together, Google’s distribution scale, advertising dominance, infrastructure depth, and ownership of YouTube place it well ahead of ChatGPT in both user reach and monetization leverage. While ChatGPT has introduced a compelling new interface paradigm, Google’s expansive ecosystem enables it to integrate similar capabilities without ceding its core advantages, allowing the company to play both offense and defense.

Importantly, reliability remains a key differentiator. Google’s brand is firmly associated with trustworthy, high quality information, an advantage that matters to both consumers and advertisers. In contrast, ChatGPT and its peers continue to grapple with hallucinations and response accuracy, limiting its suitability for certain use cases and for many advertisers.

Although, ChatGPT has established itself as a formidable first mover and one of the fastest growing consumer tech products in history, the AI landscape remains highly fluid, with competitors regularly leapfrogging one another through successive large language model upgrades. No player has yet developed a durable moat that rivals Alphabet’s scale, integration, and distribution power.

This is not to discount the promise of ChatGPT or emerging platforms like Perplexity. Rather, it underscores that AI is not a zero sum game. The addressable market for user engagement and AI applications is expanding rapidly, creating room for both Google and new entrants like OpenAI to grow.

Autonomy represents the most compelling technology growth frontier in the years ahead. Few other vectors of innovation carry the same potential for profound socio-economic impact as physical AI extends across vehicles, robotics, and defense systems.

Yet despite autonomous technology’s long term significance, public market investors still lack a reliable framework to assess the true state of technological progress and evolving competitive landscape. Corporate press releases and commentary are often taken at face value, with insufficient appreciation for the immense technical complexity and challenges involved, particularly in the context of autonomous vehicles. However, in the months ahead, key developments most notably the launch of Tesla’s robotaxi service will play a defining role in shaping investor perception of this opportunity and the trajectory of industry leadership.

Autonomous vehicle systems must solve an open-domain problem marked by real-world complexity, unpredictability, and a near-infinite range of edge cases. Unlike closed systems, they must navigate dynamic environments with variable weather, erratic human behavior, and evolving traffic conditions, making it far more challenging for AI systems to master.

Addressing this requires more than pattern recognition or mathematical function approximation; it arguably requires elements of general intelligence, including abstract reasoning and context aware interpretation of the physical environment to make safe, real-time decisions.

Given the formidable technical hurdles involved in developing truly autonomous vehicle systems, only a small number of companies are likely to succeed. However, for those that do, the upside extends beyond mobility. These companies will be positioned to leverage their technology breakthroughs across adjacent markets, unlocking new growth opportunities, compounding scale advantages, and building durable competitive moats.

Subscribe now

Tesla's Robotaxi Launch: A Moment of Truth

Tesla’s widely anticipated June launch of its robotaxi service in Austin marks a watershed moment bringing into focus the central debates shaping the autonomous vehicle landscape: the optimal technology path, future industry structure, and the timeline for mass deployment. However, there are growing signs that Tesla’s rollout may once again fall short of Elon Musk’s bullish projections and elevated investor expectations.

Concurrently, Waymo continues to execute, expanding its operations with measurable success. The company recently reached 10 million robotaxi trips, doubling in the past five months, a testament to its operational momentum and leadership position.

Beyond the high-profile rivalry between Tesla and Waymo, another player warrants closer attention, Mobileye. Unlike Tesla and Waymo, which are primarily pursuing vertically integrated models, Mobileye is focused on becoming a leading third-party supplier of Level 4 and 5 autonomous systems. Its business model is designed to enable auto OEMs and fleet operators, capturing both upfront revenue from the sale of its systems and potentially a recurring fee based on usage from robotaxi partners. Despite its compelling positioning, Mobileye’s stock valuation has yet to fully reflect its potential as an autonomous technology enabler though that is likely to change as new design wins are announced over time.

Uber, by comparison, faces a more uncertain strategic outlook. While it continues to highlight a range of AV partnerships, the core structural risk to its platform remains unresolved. The shift from a fragmented, driver based supply model to a more concentrated robotaxi supply threatens to erode the utility of a two-sided marketplace.

That said, not all scenarios exclude rideshare platform relevance. If Mobileye’s open-supplier model gains traction, it could enable a more diversified ecosystem of fleet operators potentially preserving a more meaningful role for third-party platforms like Uber. Still, even in that outcome, the underlying market structure will be fundamentally different. Technology enablers that solve the most complex autonomy challenges will be positioned to capture outsized economic rents, protected by the industry's steep technological and capital barriers to entry.

Technology Path: Compound AI Systems vs. End-to-End Ambitions

Waymo continues to maintain the pole position in the autonomous vehicle race, while Tesla remains an untested contender. This competitive view is likely to be reinforced by what is shaping up to be a more tepid than expected launch of Tesla’s robotaxi service in Austin, a rollout that will serve as a clearer test of the platform’s technical maturity and commercial readiness.

This divergence in execution reflects a fundamental contrast in strategy. Waymo and Mobileye have both adopted a compound AI system approach, making pragmatic trade-offs to reduce disengagement rates. While their systems aim to solve for an open-domain problem, they also incorporate closed-domain elements such as fixed operational design domains and high-definition maps to enhance safety and performance.

Waymo’s approach reflects a pragmatic and commercially viable strategy. By combining high definition maps, modular AI frameworks, rigorous safety layers, and extensive use of neural networks, Waymo delivers a more robust and deployable platform. Optimized for specific geographies, these architectures can still adapt to new environments, utilizing neural networks to handle unforeseen edge cases and adapt in real time. This layered system not only reflects the adoption of the currently most viable technology but provides a flexible foundation for future advancements, including the eventual adoption of end-to-end AI systems as the technology matures.

In contrast, Tesla is pursuing a more technically ambitious route. Its autonomy strategy leans heavily on a pure end-to-end neural network system, minimizing reliance on rules-based engineering. Tesla’s system is designed to generalize across diverse driving environments, an approach if successful that could enable rapid and scalable deployment. However, while an end-to-end system may ultimately prove to be the most elegant and efficient solution, current data continues to show that compound AI systems still yield superior real world performance.

Tesla’s pursuit of a purely end-to-end solution still faces inherent technical challenges in handling real world variability and edge case scenarios that could take considerable time to overcome. For example, two challenges for end-to-end systems are overfitting and catastrophic forgetfulness. Overfitting occurs when the system overfits to observed data and fails to generalize when encountering previously unseen, new corner cases. Catastrophic forgetfulness happens when a neural network forgets previously learned information as new data essentially overwrites existing knowledge. This can be problematic when continuous adaptation to new environments is required. These concerns are supported by data from the Tesla Full Self Driving (FSD) Tracker, though an imperfect dataset, still shows frequent FSD disengagements indicating that the system is not ready for a large scale unsupervised, driverless deployment.

To mitigate this risk, Tesla may choose to adopt a more pragmatic near term strategy integrating engineered elements alongside neural networks, similar to Waymo’s approach. While it can continue pursuing a pure end-to-end system long term, this shift could accelerate commercialization, improve reliability, and support broader deployment. Without such an adjustment, if Tesla’s progress continues to lag it risks a slower-than-expected rollout, allowing Waymo to extend its lead in commercial robotaxi deployment.

In parallel, Mobileye is also emerging as an AV contender, taking a similar approach to Waymo. Its strategy is similar to Waymo’s in its adoption of a compound AI system, blending engineered elements with neural networks. However, while aiming to match Waymo’s precision, Mobileye’s Level 4 and 5 platforms, Chauffeur and Drive, are also striving to achieve greater recall across varied scenarios, geographies, and less structured environments which is critical for scaling across diverse global markets.

To further strength its position as a potential go to third-party provider of autonomous driving systems, Mobileye is also building a competitive edge in its sensor suit through its proprietary image radar technology. Purpose-built for autonomy, Mobileye’s image radar offers superior object detection and enhanced resolution.

Notably, Mobileye not only intends to integrate this radar into its own AV tech stack but also plans to sell it to third parties, including Tier 1 auto suppliers. This could open a potential opportunity to capture share in the multi-billion dollar automotive radar market.

Industry Structure: Power Shifting to Tech Enablers

A more concentrated robotaxi market structure presents a strategic challenge for incumbent rideshare platforms like Uber. As the U.S. market evolves, high technological and capital barriers to entry are likely to hinder the emergence of most new competitors. The result likely will be a market dominated by just a few technology enablers.

This shift undermines the highly fragmented supply dynamics that traditional rideshare platforms rely on, eroding their value proposition and long term economic leverage. In the early years of commercialization, the robotaxi market will remain supply constrained diminishing the utility of third-party platforms. Waymo’s direct-to-consumer model in San Francisco illustrates how robotaxi operators may increasingly bypass third-party marketplaces.

In response, Uber has promoted a counter narrative to the competitive threat posed by autonomous vehicle services and robotaxis, asserting that it is well positioned to benefit and capture the opportunity that autonomy will unlock, while also claiming that “even as we see AV technology advancing, we expect AV commercialization will take significantly longer.” However, commercialization is already underway. Waymo is expanding into multiple cities with a superior consumer experience. While still small in scale, each new market that adopts robotaxi services represents a potential loss of Uber's addressable market and a threat to the company’s long term growth.

This growing headwind represents a persistent overhang on Uber’s stock and terminal value, much like how concerns over an AI driven shift in the competitive landscape for search have pressured Alphabet’s valuation. If Uber’s mobility gross bookings growth continues to decelerate while Waymo expands, the pressure on Uber’s terminal value will intensify.

To counter rising concerns about autonomy, Uber continues to spotlight new and existing partnerships with AV developers including recent announcements involving Chinese firms Pony AI and WeRide. However, both companies remain smaller players relative to Baidu’s Apollo Go, the leader in China’s robotaxi market. While Chinese AV firms are advancing rapidly, not all robotaxi services are created equal. Some operational deployments remain limited to fixed route services and rely on teleoperator support highlighting technical limitations relative to the demands of dense, unstructured urban environments.

Moreover, the structure of the global robotaxi market may diverge across regions. In both the U.S. and China, national security and cybersecurity concerns may restrict robotaxi operations to primarily domestic players, further limiting the addressable market for cross-border partnerships like those Uber is promoting.

Uber has also sought to reframe the autonomy narrative by highlighting the technical and commercial hurdles that developers must overcome. Chief among these are the challenges of building systems that are not only significantly safer than human drivers but also cost-competitive with existing rideshare services. Ironically, the very barriers that Uber emphasizes also reinforce its structural vulnerability. As AV supply consolidates around a few dominant players, the utility of Uber’s two-sided marketplace diminishes, shifting both economic leverage and profit pools toward the core technology enablers.

The magnitude of these challenges is already evident in the pullback and, in some cases, full exits by companies like Cruise, Motional, and Argo underscoring the difficulty of developing autonomous vehicle systems that are safe, reliable, and scalable. This difficulty is often understated due to the frequently misunderstood gap between advanced driver-assistance systems and truly unsupervised, driverless autonomy. This gap fuels the mistaken perception that many companies are on the cusp of successfully deploying autonomous systems, when in reality, only a few will likely have the technical and operational depth to cross that threshold. As highlighted in my post AV Technology -Facts vs. Fiction -

Analysts, investors, and the media are overemphasizing corporate press releases with limited, out of context performance data, often overstating the significance of new driver assist systems and misrepresenting them as being close to fully autonomous solutions. This has fueled overly optimistic projections that numerous companies are on the brink of successfully developing fully autonomous vehicle systems.

While Level 2+/3 ADAS systems offer "eyes-on, hands-off" convenience, they fundamentally differ from a fully autonomous system. This misconception is commonly seen in the misrepresentation of new ADAS products, such as BYD’s “God’s Eye” system, as being technologically close to fully autonomous solutions. Such conflation leads to an inflated perception of how close many companies are to deploying driverless systems, while simultaneously underestimating the formidable technical hurdles that remain.

This misperception contributes to an exaggerated sense of progress, downplaying the inherent complexity of achieving true autonomy. The technological gap between Level 3 and Level 4 systems is often understated, while the transition from Level 4 to Level 5 represents an even more significant technological hurdle.

Although the emergence of robotaxi services will profoundly reshape the competitive landscape of the rideshare market, the transition will take time. In the foreseeable future, the robotaxi industry is likely to scale toward meeting average load capacity, rather than peak demand avoiding the risk of excess supply. This implies that driver-based rideshare services will continue to play a role for some time. While there may still be a place for third-party platforms like Uber and Lyft, their role will likely be diminished relative to their current prominence in a driver based rideshare market.

Mass Deployment: Progress vs. Promises

Waymo continues to execute with little fanfare, while Elon Musk continues to make bold forecasts. Yet the results speak for themselves: Waymo’s technology strategy is delivering tangible outcomes, including fully autonomous commercial operations in Phoenix, San Francisco, Los Angeles, and Austin. The company plans to expand into Atlanta, Miami, and has announced testing in 10 additional cities further entrenching its lead. Waymo has also begun testing autonomous vehicles on Los Angeles freeways, showcasing its readiness to handle complex, high speed environments.

Tesla’s stock narrative has shifted back toward a focus on autonomy, fueling a rally in its share price ahead of Tesla’s expected June launch of its robotaxi service. However, expectations are high and a disappointing rollout could trigger a sharp correction in Tesla’s share price.

Details surrounding Tesla’s initial robotaxi launch are already underwhelming. The rollout will reportedly begin with an internal fleet of around 10 Model Y vehicles, with no initial participation from owner deployed cars, contradicting Musk’s long standing vision of a shared autonomous fleet. The initial deployment will be restricted to a fixed geographic area, which makes technical sense given that Tesla’s unsupervised Full Self Driving (FSD) will perform best on familiar roads. Still, this is inconsistent with Tesla’s broader claims of a system capable of operating anywhere, anytime.

The challenges of leaping from supervised FSD to a fully unsupervised robotaxi service should not be underestimated. Reports indicate that Tesla is still using safety drivers just weeks ahead of the planned launch. For comparison, Waymo spent a prolonged period testing its vehicles with safety drivers before transitioning to full driverless operations. This precedent points to a potentially more gradual than expected rollout for Tesla, with a continued reliance on safety drivers and, or remote teleoperators during the early stages of deployment.

There also remains limited visibility into Tesla’s robotaxi go-to-market strategy or evidence of meaningful investment in the operational infrastructure needed to support a large scale autonomous fleet. In contrast, Waymo has partnered with Uber in Austin and Atlanta, and with Moove for fleet management in Phoenix and Miami illustrating the potential need for strategic partnerships to handle logistics, maintenance, and local market scaling. These moves underscore that technology alone is not sufficient. The successful commercialization of robotaxi services will also depend on the ability to build or outsource the operational backbone required for multi-city deployment.

As I outlined in my post Does Tesla need a Lyft? -

Beyond technological hurdles, Tesla must also build a rideshare platform while simultaneously developing fleet management and logistics infrastructure on a city-by-city basis.

Rather than building the operational capabilities and infrastructure necessary for a nationwide rollout, Tesla could acquire them. Acquiring Lyft would provide immediate access to a proven marketplace, fleet management infrastructure, local market regulatory expertise, and an established rider base that could accelerate the rollout and mitigate initial risks in launching and scaling its robotaxi service.

Conclusion

The autonomous vehicle race is far from over, but current trajectories suggest a concentrated market favoring technology enablers. Waymo’s pragmatic approach and established operational experience give it an advantage. Tesla, while pursuing a more ambitious technological path, in the near term may face some hurdles in achieving reliable and scalable driverless deployment. The near term narrative may favor hype, but long term success will hinge on demonstrable performance and a viable go-to-market strategy.

As investors navigate heightened uncertainty and management teams contend with issuing forward guidance amid numerous unknowns, several recent stock moves appear disconnected from fundamental risks. In some cases, the market is overlooking these risks; in others, there’s an unwarranted complacency about how slowing economic growth and global trade disruptions could impact business performance. Below are a few observations on three companies where recent stock price action diverges from my expectations for the trajectory of their future fundamentals.

Rivian: Miles Away from Profitability

Despite continued cash burn and a projected year-over-year decline in 2025 vehicle sales, 1Q25 deliveries fell 36%, Rivian’s stock has remained surprisingly resilient amid recent market volatility. Some investors view Rivian as relatively insulated from rising tariffs due to its U.S. based manufacturing footprint. While that may provide some protection, elevated tariffs and a weakening consumer backdrop increase the risk of lower than expected sales and higher cash burn, potentially delaying the profitability milestones required to unlock the next $1 billion investment from Volkswagen.

Rivian’s partnership and joint venture with VW bought Rivian time, helping it avoid a near-term equity raise, however this partnership doesn’t resolve Rivian’s fundamental challenges in achieving profitability. Rivian ended 2024 with $7.7 billion in gross cash and is expected to burn $3–3.5 billion in 2025, leaving approximately $4–4.5 billion at year-end (excluding the next $1 billion investment from VW). This remaining cash is barely sufficient to cover 2026 capital requirements, especially considering the $1.25 billion in debt maturing next year.

Beyond the VW investment, Rivian has also secured a $6.6 billion Department of Energy loan under the Biden administration to fund its new Georgia production facility. However, this loan faces potential risk under a Trump administration. Furthermore, like the VW investment, the DOE loan is also contingent on meeting specific profitability milestones. Consequently, Rivian may need to return to capital markets to bolster its balance sheet and fund construction of its Georgia facility, where the more affordable R2 SUV is slated to enter production in 2026. Under Rivian’s agreement with VW to receive the next $1 billion payment, Rivian must achieve either -

1) Two quarters of $50 million or greater of gross profit excluding the accounting impacts of the joint venture (the two quarters are not required to be consecutive) or 2) two consecutive quarters of $1 million or greater gross profit excluding the impacts of the joint venture.

The DOE loan has a similar requirement and an additional need to meet certain vehicle sales metrics -

The Borrower may request advances under the DOE Loan for purposes of funding certain eligible Project costs, subject to the Borrower’s satisfaction of the conditions under the Loan tranche that is designated for the relevant Block. Such conditions include the Sponsor maintaining positive gross margin for certain periods prior to the first Note A Advance, the Borrower achieving certain vehicle sales metrics prior to the first Note A Advance and first Note B Advance.

These layered contingencies, policy risk, execution hurdles, and milestone-dependent capital, underscore the uncertainty around Rivian’s future funding position. While Rivian reported a positive auto gross profit in 4Q24, this was driven by $299 million in regulatory credit revenue. In 2025, dependent upon the size and timing, regulatory credits could help Rivian meet the positive gross margin thresholds required to unlock the next $1 billion from Volkswagen and advance the $6.6 billion Department of Energy loan.

However, it remains far from assured. Higher input costs and weaker than expected U.S. vehicle sales could make it difficult for Rivian to achieve a positive gross margin, increasing the likelihood of a return to capital markets.

Even if Rivian achieves its gross profit milestones and the DOE loan remains secure under a Trump administration, the company's path to profitability remains challenging. For context, Tesla, benefiting from significant cost advantages due to its superior scale ( more than 35 times Rivian’s sales volume) and vertical integration, posted a negative operating margin in 1Q25, excluding regulatory credits, highlighting the immense scale Rivian still needs to achieve to reach sustainable profitability.

Rivian may highlight its technology credentials, but at its core, it's still a capital intensive auto company, burning cash, and facing declining sales. Relative to Tesla, Rivian’s nearly $14 billion market cap may appear modest. But in the context of just $5 billion in 2024 revenue and compared to Ford’s $40 billion market cap and $185 billion in revenue it’s clear Rivian is far from a bargain.

Subscribe now

Carvana - Growing Pressure from Consumer Credit Risk

Carvana is viewed as a potential beneficiary of higher tariffs, under the assumption that rising new car prices will drive more consumers toward the used car market. While this dynamic may support near term demand, particularly through pull-forward effects ahead of anticipated price hikes, it overlooks the fundamental nature of Carvana’s business. At its core, Carvana operates more as a consumer finance business, with significant exposure to higher risk, subprime borrowers. This reliance on credit market conditions introduces meaningful risk, especially in a deteriorating macroeconomic environment.

While the full impact of rising U.S. import tariffs and disruptions to global trade remains uncertain, the dual effect of what amounts to a regressive tax on lower income households combined with broader employment headwinds will likely drive an increase in loan losses, with subprime borrowers disproportionately affected.

This pressure is emerging against the backdrop of an already elevated auto loan 60+ day past due rate for subprime borrowers of 6.6% compared to approximately 5% during the global financial crisis, an ominous signal for Carvana’s business model. A increase in loan losses not only erodes Carvana’s ability to recognize gains on the sale of auto loans, a key driver of gross profit, but more critically, any reduction in credit availability threatens the company’s ability to meet revenue growth expectations.

While Carvana’s 2023 debt restructuring provided financial flexibility, the company has not capitalized on its surging share price to raise equity and strengthen what remains a suboptimal capital structure. For a low margin, cyclical business with significant exposure to riskier consumers, this is a missed opportunity. Trading at a demanding valuation of 28x 2025 consensus EBITDA, the stock faces downside risk as consumer credit conditions deteriorate.

Uber - Navigating Cyclical Headwinds and a Structural Shift

Despite underperforming the market over the past year, with many investors already underweight or short the stock, Uber has proven relatively defensive during this recent period of market volatility. However, three factors are likely to drive further underperformance in the near to medium term.

Uber’s revenue outlook faces two key cyclical risks: exposure to travel and a potentially softer retail media market. Concerns around a slowdown in mobility bookings growth were evident in the reaction to Uber’s 4Q24 earnings results and guidance. Despite strong headline numbers, the company missed 1Q25 booking guidance expectations by nearly 2%, triggering a 7.5% sell-off on the day of the release and cast doubt on its ability to sustain prior growth momentum. These concerns have since been compounded by emerging signs of weakening travel trends.

Airport rides, which account for a mid teens percentage of mobility bookings, now represent a growing source of incremental risk. Combined with existing concerns, such as weakening price elasticity from higher fares and the potential plateauing of new product driven volume gains, these factors heighten the probability of a more pronounced deceleration in mobility gross bookings growth.

In addition, Uber’s delivery business is exposed to potential weakness in retail media spend. While macroeconomic risks to the digital media market have been widely discussed, most of the focus has centered on advertising centric platforms. Less attention has been paid to retail media, which alongside connected TV (CTV) has been a major driver of digital ad growth. This higher margin revenue stream has been a meaningful factor in the improved delivery EBITDA margins for both Uber and DoorDash.

Uber’s advertising revenue run rate now exceeds $1 billion annually, with the majority stemming from its delivery business. In 4Q24, EBITDA from Uber’s mobility segment was 2.4x that of delivery, yet delivery contributed nearly 45% of year-over-year EBITDA growth driven in large part by expanding ad revenue and strong margin improvement.

However, this ad revenue stream is vulnerable in a more challenging macro environment. A slowdown in retail media spending could pressure delivery EBITDA margins and weigh on gross bookings, leading to lower than expected profitability, especially as consumers become more cost conscious.

Concurrently the threat from robotaxi services remains a structural risk. As outlined in my post Mobility Market – Shifting Gears, robotaxi services are set to reshape the rideshare industry, raising terminal value risks for market leaders like Uber. The crux of this transformation lies in the shift of the industry's competitive moat from two-sided marketplaces that connect drivers and riders to the technology enablers that develop and control autonomous vehicle systems.

In the foreseeable future, the U.S. robotaxi market appears to be evolving toward an industry dominated by three or four key technology enablers, with Waymo leading the way. High technological and capital barriers to entry will hinder the emergence of most new competitors. This concentrated market structure poses a challenge for Uber, as leading operators will remain supply constrained for some time and increasingly pursue direct to consumer strategies that bypass third-party platforms. Waymo’s service in San Francisco, where rides are booked directly through its app, highlights this shift. Zoox, backed by Amazon, is also emerging as a contender, now active in six test locations and expected to go direct to consumer as well.

Waymo’s partnership with Uber in Austin and Atlanta serves as a way to accelerate Waymo’s go-to-market strategy while outsourcing fleet management to a third party. The specific financial terms of the arrangement have not been disclosed. While this collaboration makes Uber relevant in the robotaxi space, the exclusive use of Uber’s app in these cities may function more as a form of compensation to Uber than a necessity for Waymo to be on its platform. Waymo's partnership with Moove for fleet management in Phoenix and Miami further signal a flexible, multi-partner approach to scaling its operations.

Investor concern over the competitive threat from robotaxis was evident on Uber’s last earnings call, where four out of six analyst questions focused on autonomy and future robotaxi services. Against this backdrop, the risk of Uber missing mobility bookings expectations over the next couple of quarters combined with more muted EBITDA margin expansion skews the stock’s risk-reward to the downside. While international mobility growth may help cushion a U.S. slowdown, near term trends in the U.S. market are likely to dominate the narrative and weigh on the company’s perceived terminal value.

The complexity of the global economy, combined with the unpredictability of Trump-era policy shifts, makes the fallout from a tariff-induced trade war exceptionally difficult to forecast. Yet while some, notably the Wall Street Journal Editorial Board declaring U.S. Tariffs Make Xi Jinping’s Day, argue that China is better positioned than the U.S. to weather a trade war, the risks of a prolonged conflict are in fact more apparent and acute for China.

Though the U.S. faces real economic and diplomatic risks, China’s financial vulnerabilities run deeper. Its fragile financial foundation and heavy reliance on its trade surplus, a key source of domestic liquidity, leave it ill-equipped to absorb an economic shock. This vulnerability also raises the risk of an unintended response from Xi Jinping, including a potential escalation in Cross-Strait tensions with Taiwan to divert attention from any domestic economic instability.

That said, any reset of U.S. trade policy should prioritize China, not trigger a broader global trade war as is now unfolding. A focused strategy would not only be more effective but would also likely command broader domestic support. For years, the U.S. has tolerated a lopsided relationship with China, marked by restricted market access, intellectual property theft, subsidization of Chinese companies to undercut U.S. competitors, and an ongoing digital propaganda campaign. Simultaneously, China has also aggressively expanded its regional ambitions unchecked, including constructing militarized artificial islands in the South China Sea, threatening key international shipping lanes.

While China will attempt to exploit global frustration with the U.S. and the Trump administration, particularly among allies and trading partners, any fallout from a trade war would be more of a loss for the U.S. than a win for China. Across Europe and Asia, many countries remain wary of China’s intentions and its expansionist behavior. The influx of cheap Chinese goods continues to fuel concerns about threats to local industries and job losses.

Although the U.S. may be retreating from providing foreign aid, reducing its soft power reach, China is not a natural substitute. Its Belt and Road Initiative (BRI), while framed as a development program, has largely focused on infrastructure not on soft power pillars like healthcare and education. Often viewed as “debt trap diplomacy,” BRI is widely seen not as a tool for partnership, but as a mechanism for Beijing to gain economic leverage and advance its strategic agenda.

The Wall Street Journal editorial argues that “China’s authoritarian system means Mr. Xi probably can ride out whatever political or social pain might result from higher unemployment or slower economic growth in a trade war.” But this view overlooks the depth of China’s underlying financial and economic risk: significant economic imbalances, a deterioration in net FDI inflows, elevated youth unemployment, and the lasting damage from its property sector downturn. Simply put, this is a precarious moment for China to be dragged into a trade war.

Subscribe now

China’s Debt Addiction

This trade war comes at a time when China’s dependence on debt fueled growth is intensifying. Credit, channeled through state-owned banks, local government finance vehicles (LGFV), and non-bank financial firms, remains the Chinese Communist Party’s (CCP) primary lever of economic control. But this reliance on policy driven lending has produced an economy that’s dangerously over-leveraged. Corporate and LGFV debt now exceeds 170% of GDP, more than double the U.S., while total debt hit 303% of GDP in 2024.

China’s economic model effectively uses debt as a substitute for cash flow. This feature of China’s growth model has resulted in a continuous decline in the productivity of credit as a greater portion of loans are allocated to cover existing interest payments and uneconomic investment projects. Consequently, China requires an ever increasing amount of credit to generate a unit of GDP growth. Using China’s broader measure of credit, total social financing (TSF), the ratio of credit required to generate one unit of GDP has increased from 2.66 in 2017 to 5.52 in 2024.

Compounding China’s economic imbalances is its growing reliance on trade. In 2024, the country posted a global trade surplus of $992 billion, equivalent to 5.6% of GDP, with net exports accounting for 30% of GDP growth according to the Mercator Institute for Chinese Studies. A disruption to global trade flows could impact China exports beyond its bilateral relationship with the U.S., putting additional pressure on this critical source of domestic liquidity. In such a scenario, Beijing would likely be forced to lean even more heavily on credit to sustain growth.

At the same time China’s once booming property sector, another former key source of GDP growth, remains mired in decline. The collapse of China's residential real estate market is a prime example highlighting the risks and fragility of its Ponzi scheme like industry structures driven by debt fueled growth and speculative investment activity reliant on continuous credit expansion. With the private sector in retreat, state-owned enterprises are increasingly acting as buyers of last resort in the still struggling property market with the decline in construction activity and bad debt problems to remain an economic headwind for some time.

Political Control Meets Economic Fragility

Beyond China’s economic vulnerabilities lies another critical pressure point: the Chinese Communist Party’s dependence on information control to maintain political power. The CCP has built an increasingly Orwellian surveillance state, one of mass censorship and a vast web of surveillance systems to monitor and shape citizen behavior. While this may appear from the outside as a tightly controlled society, the reality is more fragile. Xi Jinping and China’s leadership rule not from a place of unshakable confidence, but a fear of dissent and instability.

These extensive efforts to enforce control come at a time when the regime faces mounting economic headwinds and rising youth unemployment. Left unchecked, these dynamics risk undermining the longstanding social contract that has underpinned CCP legitimacy, trading political freedom for economic prosperity, and is now under strain.

China’s history offers ample reminders of how quickly discontent can boil over from the Taiping Rebellion and the Cultural Revolution to Tiananmen Square and, more recently, the blank paper protests during the COVID lockdowns. While the immediate risk of mass unrest remains low, the state of China’s economy means a sharp external shock, like a prolonged trade war, could become a catalyst.

Such domestic instability could also create internal openings for factions within the CCP dissatisfied with Xi’s leadership. However, if Xi perceives a threat to his authority there is a risk that Beijing could escalate tensions abroad as a distraction, potentially through a naval blockade of Taiwan, as a means of deflecting from domestic economic disruption. Ultimately, China is not a winner in this trade war. At best, it faces serious economic headwinds. At worst, the wrong shock at the wrong time could trigger dangerous, unintended second order consequences that ripple well beyond economic forecasts.

The risks inherent in Coreweave's business model have been widely discussed. The value of its core asset—GPU compute capacity—is subject to rapid depreciation driven by Nvidia’s aggressive annual release cycle of increasingly advanced and powerful chips. Furthermore, CoreWeave faces extreme revenue concentration, with its top two customers accounting for 77% of its 2024 revenue—Microsoft alone making up 62%.

While valid, concerns about CoreWeave’s business model are secondary to a broader and more pressing industry risk: the quality of AI compute infrastructure demand. CoreWeave and other neoclouds —emerging AI-focused cloud providers— are aggressively building out data center capacity in anticipation of future demand, creating a speculative dynamic highly exposed to the cyclical risk of oversupply in AI compute capacity. Another red flag is the circular nature of some demand sources. For instance, Nvidia not only supplies GPUs to CoreWeave but is also both an investor in and a customer of the company, effectively generating demand for its own product, raising questions about the true sustainability of the current AI infrastructure spending growth trajectory.

This quality of demand concern is an additional cyclical risk to Nvidia, alongside the significant risk to its China revenue. This China revenue risk was outlined in my post Nvidia Navigating Elevated Expectations, and stems from the upcoming U.S. export restrictions under the Framework for the Diffusion of Advanced Artificial Intelligence Technology, with compliance required by May 15, 2025.

Subscribe now

CoreWeave and Global Crossing - Different Cycle, Similar Story

While comparisons between today’s buildout of AI compute infrastructure and the late1990s/early 2000s telecom fiber expansion are common, the company-level parallels are even more striking. The comparison between CoreWeave and Global Crossing—a telecom upstart that rode the dot-com wave and ultimately filed for bankruptcy—is particularly noteworthy.

Global Crossing and CoreWeave’s risks are quite similar. Both companies built ahead of demand, required immense upfront capital investment, and bet on being indispensable to fast-growing ecosystems—AI compute in CoreWeave’s case, internet data in Global Crossing’s. And both faced (or face) existential risks tied to timing: should demand growth slow, competition intensify, or pricing compress, the economics can unravel quickly. Global Crossing suffered as the market overbuilt in a commoditizing broadband capacity environment. CoreWeave could face similar pressure if AI compute outsourcing needs are reduced or AI demand growth fails to keep pace with current expectations.

It’s notable that both were founded and led by industry outsiders. Global Crossing’s founder, Gary Winnick, came from a Wall Street background, while CoreWeave’s CEO had previously worked as an energy sector investor and portfolio manager before entering the cryptocurrency mining business and then later pivoting to AI infrastructure services. These were financial operators bringing a capital markets mindset to highly technical markets.

Both companies strategies essentially focused less on innovation and more on capitalizing on a secular wave of demand: Global Crossing for internet bandwidth, and CoreWeave for AI compute capacity. Although each promoted itself as a critical enabler of the next era of technology, aggressively scaling capital-intensive infrastructure underpinned by optimistic market forecasts, their underlying services both had the commodity like feature of limited pricing power in the event of a market environment of surplus capacity.

One additional key parallel lies in their debt-fueled expansion strategies. Global Crossing relied heavily on debt to build out its global network, capitalizing on bullish investor sentiment and easy credit. Similarly, CoreWeave and other neoclouds rapid rise has been bankrolled by debt and structured financing—most notably, multibillion-dollar GPU-backed loans and equity commitments from firms like Blackstone. As of year-end 2024, CoreWeave held $7.9 billion in total debt and had $15 billion in lease obligations. In both cases, largely speculative capacity buildouts were driven by projected demand—an inherently risky approach for businesses with highly leveraged capital structures.

Unpacking CoreWeave’s Revenue Visibility

Although CoreWeave’s S-1 filing emphasizes value-add opportunities like optimizing Model FLOPS (Floating Point Operations Per Second) utilization, a metric for measuring how efficiently a GPU is being used, at its core the company is selling outsourced AI compute capacity. While technically complex, this is a business that exhibits commodity-like traits. Differentiation exists, but price remains a primary competitive factor.

Furthermore, the company’s revenue visibility may be overstated. CoreWeave’s S-1 highlights that “ We generate substantially all of our revenue from committed long term-contracts” which accounted for 88% and 96% of revenue in 2023 and 2024 respectively, with the balance from “on demand” or spot revenue. The company also states that “committed contracts generally have a fixed price for their duration.”

While CoreWeave cites long-term commitments from Microsoft and OpenAI as foundational to its growth outlook and revenue visibility, pointing to its $15.1 billion in remaining performance obligations as of December 2024, the precise terms of these contracts remain unclear.

These revenue streams, which comprise the majority of CoreWeave’s remaining performance obligations, may not be as secure as they appear. It’s unlikely that Microsoft would lock itself into a rigid take-or-pay structure with a relatively new AI data center operator lacking negotiating leverage. This concern has been reinforced by a report that Microsoft has backed out of parts of its agreements with CoreWeave due to delivery issues and missed deadlines.

Adding to the uncertainty, recent reports suggest Microsoft is scaling back on its capex plans, while OpenAI is aggressively building out its own AI infrastructure through its Stargate JV with SoftBank, further highlighting the risk that future demand from CoreWeave’s two largest customers could fall short of expectations.

Quality of Demand Risk

While the largest cloud providers dominate AI infrastructure spending, neocloud players now account for a meaningful share of demand for Nvidia GPUs and AI compute systems. This cohort of companies includes reinvented crypto miners pivoting into AI, smaller cloud players shifting to AI services, and early-stage startups chasing generative AI growth.

Fueled by a wave of equity and debt financing—often collateralized by chips and AI systems—many neoclouds are building out compute capacity ahead of realized demand. This speculative build-out leaves them especially vulnerable to oversupply risks.

There’s also a concern about the quality of demand, not just from the buyers of GPUs, but the demand from AI data center customers. Much of it is driven by VC-backed, high cash burn startups rather than mature, revenue-generating enterprises—raising questions about the durability of this demand. Nvidia CEO Jensen Huang acknowledged this dynamic in the company’s August 2024 earnings call, noting: “The number of generative AI startups is generating tens of billions of dollars of cloud renting opportunities for our cloud partners.”

Ripple Effects from Surplus AI Compute Capacity

Nvidia remains the valuation umbrella and sentiment anchor for AI-themed stocks. Heading into the second half of 2025, any decline in Nvidia’s revenue expectations will send ripple effects across the broader AI ecosystem. In a surplus AI compute capacity environment, neocloud players—that have built capacity ahead of demand—will be the first to feel the impact. Among hardware suppliers, Dell and Supermicro (SMCI) are particularly exposed, as CoreWeave and other neoclouds have been key growth drivers for their AI server sales. A pullback from these speculative buyers would not only pressure revenue estimates but likely trigger multiple compression.

For cloud players like Amazon, Meta, and Google, a slowdown in AI infrastructure spending could be a net positive. If reduced capex aligns with accelerating AI application adoption, improved free cash flow and diminished investor concerns over overspending could catalyze a rebound in these stocks, reversing recent underperformance.

However, a slowdown in AI compute demand would hit CoreWeave directly—pressuring pricing and utilization. Given its debt-heavy capital structure and speculative expansion, CoreWeave could increasingly resemble Global Crossing: a cautionary tale of over estimating demand and financial overreach.

The surge in expectations surrounding both public and private autonomous vehicle (AV) technology companies echoes the electric vehicle (EV) SPAC and IPO frenzy of 2020-2021, which peaked with Rivian's late 2021 IPO. At the time, bullish sentiment swept across the EV sector, fueling exaggerated expectations about the speed and scale at which new EV companies would emerge. Investors and analysts projected rapid market penetration and profitability, overlooking the immense difficulty of building an auto company from scratch and achieving the production scale necessary to compete in an industry defined by capital intensity and operational complexity.

The fallout from the EV hype cycle has been severe. Fast forward to today, most of the EV SPACs that went public during that period have either gone bankrupt or are in a distressed financial state. In the U.S., aside from Tesla, Rivian appears to be the only EV OEM with a potential path to survival — and that has come at the cost of burning through billions of dollars.

A similar pattern is now unfolding in the AV space. Analysts, investors, and the media are overemphasizing corporate press releases with limited, out-of-context performance data, often overstating the significance of new driver-assist systems and misrepresenting them as being close to fully autonomous solutions. This has fueled overly optimistic projections that numerous companies are on the brink of successfully developing fully autonomous vehicle systems.

These projections underestimate the technological challenges involved in building a driverless vehicle system. Just as scaling an EV company proved far more difficult than initial projections suggested, it’s likely that only a handful of companies in the U.S. and Europe will succeed in commercializing Level 4 systems — and eventually Level 5 systems capable of operating in all driving environments and conditions.

Subscribe now

The Technical Challenges of Autonomous Driving

In the foreseeable future, the U.S. robotaxi market is likely to consolidate around a small group of dominant technology enablers—likely three or four key players—capable of developing fully autonomous, driverless systems that can reliably operate across diverse driving environments and conditions. Waymo and Tesla are leading this race, albeit with different approaches to autonomy.

Substantial technological and capital barriers to entry in autonomous vehicle development have already forced companies like Cruise, Motional, and Argo AI to scale back operations or exit the market. This underscores the formidable challenges of establishing a commercially sustainable autonomous vehicle system and robotaxi service.

Driving is an open-domain problem with an infinite long tail of complexity and unknown scenarios. Autonomous vehicle systems must handle countless edge cases — unpredictable, low-frequency events, from changing weather conditions to unexpected obstacles to the unpredictable behavior of other drivers.

The core challenge for autonomous vehicle systems is not just recognizing objects within the driving environment but understanding the environment itself. Object detection and pattern recognition alone is not enough — systems need to infer the intent and behavior of other road users and adjust their responses accordingly. This requires more than mathematical function approximation; it demands elements of general intelligence, such as abstract reasoning, real-time adaptation, and interpreting the physical environment in a nuanced and context-aware manner to make optimal driving decisions. Bridging this gap still requires advances in decision-making algorithms, diverse data sets, and sufficient onboard sensor fusion and compute resources to support real-time inference.

A recent drive in a Tesla Model Y highlighted a key technical challenge for autonomous vehicle systems. The drive was through a familiar suburban environment, and overall, the FSD system performed well — except for two disengagements.

One disengagement occurred when the vehicle struggled to navigate out of a parking lot. The second disengagement, however, was more noteworthy and informative. While driving through a suburban village, the vehicle came to a complete stop in the middle of an intersection despite a green traffic light. The vehicle’s halt seemed to be triggered by FSD detecting a pedestrian standing at the corner, waiting to cross the street. However, the pedestrian was stationary, the traffic light was green, and the pedestrian signal showed a red "Don't Walk." Despite these cues, the vehicle stopped unnecessarily. This is a good example of an autonomous system identifying an object in its environment but failing to understand the broader context needed to make an optimal driving decision.

Tesla’s vision-only system correctly identified the pedestrian but lacked the contextual understanding to conclude that it was safe to proceed. A more robust sensor suite could have improved performance in this situation. Radar, for example, measures an object’s distance, speed, and relative motion — key data points that would have confirmed the pedestrian was stationary and not about to cross, prompting the vehicle to drive through the intersection safely.

This example illustrates a current limitation of Tesla’s vision-only approach. In contrast, Waymo’s lower disengagement rate reflects the advantages of its more comprehensive sensor suite — combining cameras, radar, and LIDAR to provide richer and more diverse environmental data. Tesla’s reliance on a vision-only approach could prove to be superior in the long run, but in the near term, it still struggles to match the decision-making consistency of a multi-sensor system.

Perception vs. Reality of Autonomous Driving Technology

The gap between advanced driver-assistance systems (ADAS) and fully autonomous technology is significant, a point often overlooked. While Level 2+/3 ADAS systems offer "eyes-on, hands-off" convenience, they fundamentally differ from a fully autonomous system. This misconception is commonly seen in the misrepresentation of new ADAS products, such as BYD’s “God’s Eye” system, as being technologically close to fully autonomous solutions. Such conflation leads to an inflated perception of how close many companies are to deploying driverless systems, while simultaneously underestimating the formidable technical hurdles that remain.

This misperception contributes to an exaggerated sense of progress, downplaying the inherent complexity of achieving true autonomy. The technological gap between Level 3 and Level 4 systems is often understated, while the transition from Level 4 to Level 5 represents an even more significant technological hurdle.

While Chinese companies are advancing in autonomous technology, some of this progress is also overstated. Many operational robotaxi networks in China are confined to fixed route services, revealing these systems limitations compared to the broad capabilities required for navigating diverse, unstructured driving environments.

Context Matters in Measuring AV Progress

Outside of China, there is a significant opportunity for third-party suppliers of autonomous vehicle systems. One such promising player is Wayve, which is positioning itself as a hardware-agnostic autonomous vehicle system supplier. Wayve, like Tesla, is pursuing an end-to-end driving system.

However, assessing autonomous technology companies, both private and public, remains challenging due to limited performance data and detailed information. This challenge is compounded by promotional company announcements that often lack the context needed to evaluate system performance or benchmark it against competitors.

For example, Wayve recently reported approaching UK performance levels in the U.S. after collecting just 500 hours of incremental U.S.-specific training data. While this sounds impressive, the significance is hard to gauge without baseline intervention rates and more quantifiable data for context.

This situation echoes the EV SPAC bubble, where investors and the media overemphasized corporate press releases, neglecting thorough analysis of a company's technology strategy and commercial viability. While Wayve may ultimately succeed, such announcements perpetuate the misconception that numerous AV players are on the verge of success, overlooking the substantial technological and operational hurdles that remain.

Mobileye’s autonomous vehicle system development strategy contrasts with Wayve’s and is more aligned with Waymo’s “compound AI system” approach — combining engineered elements with extensive use of neural networks. A compound AI system reflects a willingness to make trade-offs to achieve reduced generalization error, utilizing defined operational parameters to lower disengagement rates.

While AI technology may eventually advance to a point where a pure end-to-end neural network system demonstrates superior performance, current data supports the notion that a compound AI system still delivers better real-world performance and a lower rate of critical interventions.

Waymo Remains in the Pole Position, While Tesla is a Show Me Story

Waymo maintains its position as the autonomous technology leader, underscored by its first-mover advantage and pragmatic "compound AI system” approach that enables a more robust and commercially viable robotaxi service. This strategy, combining high-definition mapping, rigorous safety protocols, and modular AI frameworks with extensive use of neural networks, ensures greater reliability—a critical factor for commercial deployment. Waymo's approach not only reflects the most viable technology today but also provides a flexible foundation for future advancements including the eventual adoption of an end-to-end AI system as the technology matures.

Tesla's management team continues to pursue a more complex technological path, relying heavily on an end-to-end neural network system. This approach, as discussed in my post The Ultimate AI “Killer App”, still faces significant technical challenges in handling real-world variability and edge case scenarios, which could take considerable time to overcome.

The recent steep correction in Tesla’s stock reflects fundamental weakness in EV sales. As highlighted in my post What’s Next for Tesla, although Elon Musk’s political activities may have damaged the brand and impacted sales, the deeper issue is Tesla's stale and narrow product lineup — a fundamental headwind that predates Musk's heightened political visibility.

There are emerging expectations that the anticipated June launch of Tesla’s robotaxi service in Austin could be delayed by a couple of months. This wouldn't surprise close Tesla observers. While EV sales remain weak, the market narrative is likely to shift back toward autonomy, and could trigger a rally in its share price ahead of Tesla’s robotaxi launch. However, any more significant delays in launching Telsa’s robotaxi service or performance disappointments would likely trigger another wave of selling pressure.

Tesla's reliance on an end-to-end neural network system, and the still existing technical challenges inherent in this approach, could impede and delay a successful driverless robotaxi launch. This concern is reinforced by data from the Tesla Full Self Driving (FSD) Tracker — though an imperfect dataset — which still shows frequent FSD disengagements, indicating that the system is not ready for an unsupervised, driverless deployment.

Tesla may continue to present its system as an end-to-end neural network, however to mitigate these risks, Tesla may need to adjust its near term strategy to more closely resemble Waymo's approach. This would necessitate adopting a "compound AI system" that integrates more engineered elements with neural networks and focus the initial rollout and expansion to additional cities to fixed geographical areas. Tesla already plans to launch in a defined area in Austin, which makes sense, as unsupervised FSD is likely to perform better on familiar, frequently traveled roads.

While Tesla may continue to pursue its long-term goal of a pure end-to-end neural network system, this pragmatic shift could accelerate commercialization, improve reliability, and increase the chances of a successful deployment. Without this strategic adjustment, Tesla risks ceding further ground to market leader Waymo.

Recent developments and exposure to emerging AI end markets, such as consumer devices, strengthen the case for a narrowing of Taiwan Semiconductor Manufacturing Corp.’s (TSMC) valuation discount to its AI chip peers, positioning its stock to outperform Nvidia and Broadcom. Despite the pullback in Nvidia, Broadcom, and Marvell’s share prices, these stocks still trade at a significant premium to TSMC, with 2025 price-to-earnings multiples of 24–28x, compared to TSMC at 18x.

Subscribe now

Strength in Numbers - A Competitive Edge

The latest results from TSMC's key AI chip customers - Nvidia, Marvell, and Broadcom - reinforce the argument that TSMC's valuation discount is unjustified. The mixed reactions to these companies' earnings reflect their individual fundamentals and perceived risk, not TSMC’s fundamental strength.

Regardless of whether demand for Nvidia’s GPUs or custom AI chips grows at a faster pace —or whether Broadcom and Marvell or Taiwanese competitors like Alchip and Mediatek are gaining ground in the custom AI chip market—TSMC, with its near monopoly in advance foundry capacity, remains the ultimate winner. The contrasting investor reactions to Nvidia and Broadcom’s latest earnings underscore this point.

Risks to Nvidia’s revenue trajectory include the growing adoption of custom AI chips and the threat to China-linked revenue. Nvidia revenue is projected at approximately $200 billion this fiscal year, with a substantial $140 billion in gross profit. This profit pool represents a significant cost for customers, creating a strong incentive for cloud players to explore custom AI chip alternatives from companies like Broadcom and Marvell to lower AI compute costs and increase competitive pressure on Nvidia.

Nvidia’s more immediate and greater challenge is the growing risk to its China-sourced revenue, both direct and indirect (via Singapore), which accounted for 31% of Nvidia’s total revenue in FY25. Sales to Singapore surged 3.5x year-over-year—outpacing overall revenue growth—making it Nvidia’s second-largest market at 18% of total revenue. This revenue surge is widely attributed to the use of Singapore as a transshipment hub to circumvent U.S. export restrictions, putting this revenue stream at increasing risk as the May 15, 2025 compliance deadline for the Framework for the Diffusion of Advanced Artificial Intelligence Technology export restrictions approaches.

In contrast, Broadcom recent commentary signals faster than expected growth in its AI chip business, driving increased demand for TSMC’s foundry service. On its earnings call, Broadcom’s management disclosed that its custom AI chip customer base has expanded from two to four incremental engagements, totaling seven existing and new engagements.

As the dominant supplier of advanced foundry capacity, TSMC benefits from a diversified customer base and broad end-market exposure, reducing its reliance on any single customer and lowering business risk. If Nvidia’s China revenue comes under pressure, the impact on TSMC will be far less severe and could be offset by increased demand from other customers, such as Broadcom. This stronger position in the AI supply chain and greater revenue visibility are variables that strengthen the case that the stock warrants a comparable valuation to its customers.

Catalyst on the Horizon

While TSMC’s economic moat and competitive position alone justify a narrowing of its valuation discount to its peers, the more tangible near term catalyst could be the shift in the AI investment theme from infrastructure buildout to applications. As highlighted in my post Surf's up- Catch the AI Smartphone Wave, technological breakthroughs in AI model efficiency are set to make larger AI models more viable for on-device processing. This will enable more advanced on-device AI features and applications, driving growth in AI smartphones and consumer devices.

With 35% of TSMC’s revenue coming from smartphones, this shift will expand TSMC’s AI addressable market beyond AI infrastructure, positioning it as a key beneficiary of the next wave of AI adoption. TSMC’s exposure to consumer devices not only provides a buffer against the risk of a slowdown in AI compute infrastructure spending, but also offers potential upside to revenue expectations as more advanced on-device AI features and applications drive stronger-than-expected smartphone sales in the second half of 2025 and into 2026.

Framing the China Risk

TSMC’s valuation discount has long reflected the risk of Chinese military aggression against Taiwan, yet this same risk is not typically reflected in its customers’ stock valuations despite their direct and immediate exposure to this risk. As highlighted in my post Taiwan is Arrakis, this risk has increased in recent years. However, the near term probability of conflict remains relatively low since almost any scenario involving a military confrontation would be a losing proposition for all parties—the U.S., China, Taiwan, and the global economy.

The contradiction is that while an invasion would be profoundly negative for TSMC, it would also severely impact Nvidia, Broadcom, and other key customers—implying that the valuation premium of TSMC’s customers understates this shared geopolitical risk relative to TSMC.

TSMC recently announced an additional $100 billion investment in the U.S., raising its total planned U.S. investment to $165 billion. While this diversifies TSMC’s manufacturing footprint away from Taiwan, this additional $100 billion is equivalent to only about three years of TSMC’s total capex. TSMC will continue to invest in Taiwan, consequently Taiwan’s geopolitical significance and TSMC’s unique strategic value will remain unchanged in the foreseeable future.

However, this commitment should strengthen U.S.-Taiwan relations and provides President Trump with a political "win." Combined with Taiwan’s President Lai Ching-te's pledge to increase defense spending to over 3% of GDP, these moves could bolster U.S.-Taiwan ties and signal improved relations. Greater alignment of U.S.-Taiwan interests would likely prompt increased caution from Xi Jinping and the Chinese leadership when considering aggression toward Taiwan, and thereby reduce TSMC’s near-term geopolitical risk.

As a general rule, most internet product verticals are defined by a winner takes most market structure, where the leading company captures the vast majority of industry profits.

While each internet segment has unique variables, the networking and scale advantages established by an industry leader often reach an order of magnitude that compounds over time, becoming insurmountable for competitors.

Netflix’s management continues to deliver strong execution and post outstanding operating performance. The valuation gap between Netflix and legacy media peers reflects the market’s belief that streaming is a winner-takes-most market.

However, while Netflix remains the dominant force in streaming, the industry’s structure differs from many other internet verticals, where a single platform captures the majority of market share.

Netflix’s competitive advantages will sustain its superior profitability and a disproportionate share of industry earnings, yet the nature of the streaming and video entertainment market leaves room for a number of profitable players as the transition from linear TV progresses. However, for other streaming platforms to compete meaningfully with Netflix, industry consolidation will be essential to achieve the scale and engagement required to be viable long-term, profitable streaming competitors.

Subscribe now

Warner Bros. Discovery - A Streaming Led Comeback

Warner Bros. Discovery is a prime example. The company’s Max streaming platform, alongside Disney+, is arguably the most valuable streaming asset outside of industry leader Netflix. Max is well-positioned to play a central role in the industry consolidation that lies ahead. In late 2024 Warner announced a new corporate structure with two operating divisions - Global Linear Networks and Streaming & Studios - to enhance future strategic flexibility in anticipation of industry wide restructuring and consolidation.

Following a challenging initial period after its merger with Discovery, Warner now has improved financial and operational visibility, with significantly reduced—though still high—leverage after paying down $13 billion in debt since the second quarter 2022. However, Warner’s still significant exposure to a declining linear TV business continues to weigh on its valuation, which remains depressed at 7x 2025 EV/EBITDA, compared to Netflix’s valuation of 30x EV/EBITDA.

This depressed valuation presents an intriguing opportunity. With continued deleveraging, streaming EBITDA growth, and an attractive valuation, Warner Bros. Discovery offers a compelling risk-reward setup, particularly in a media landscape that is poised for further consolidation that Warner is expected to be an active participate.

Warner has two idiosyncratic factors that will drive its stock performance. First, the company is set to continue reducing its debt load, leveraging $4-5 billion in annual free cash flow to further pay down its $40 billion in gross debt ($35 billion in net debt) over the next couple of years. This deleveraging process is significant because debt currently comprises a majority of the company's enterprise value. As Warner further shifts its enterprise value from debt to equity, its $27 billion market capitalization stands to benefit. Assuming a static valuation multiple and stable EBITDA, this process alone could generate double-digit percentage stock returns.

Additionally, Warner is undergoing a favorable shift in its revenue and EBITDA mix. The increasing contribution of Max to total EBITDA is gradually reducing the company’s reliance on declining linear TV, with a growing share coming from its streaming platform and highly productive studio business.

Max has reached an inflection point, delivering stronger subscriber growth and improved profitability. In 2025, the Direct-to-Consumer (DTC) segment is projected to generate $1.3 billion in adjusted EBITDA, marking an improvement of nearly $3 billion over the past two years. Subscriber momentum is also accelerating, with Warner targeting 150 million subscribers by the end of 2026, up from 116.9 million at the end of 2024.

This transition to greater contributions from a growth business warrants a higher EV/EBITDA multiple. Given Warner’s capital structure, even a one-turn increase in its valuation multiple would have an outsized impact on the company’s equity value and stock price.

Netflix’s Secret Sauce

Although Netflix has benefited—and continues to benefit—from the advantages of superior scale, much of its success has been driven by a strategy focused on maximizing engagement per subscriber. Netflix early on recognized that in a streaming-based market structure, it was crucial to strike a balance between quality and quantity, catering to a wide range of audiences is just as crucial as producing blockbuster hits.

This approach has allowed the company to build a large, global subscriber base and maintain an industry-leading low churn rate. For competing streaming platforms, driving higher engagement and reducing subscriber churn will be critical to achieving greater scale, enhancing profitability, and securing a meaningful share of the industry’s profit pool.

Before streaming, the constraints of linear TV meant limited time slots for content, making premium, tentpole programming the key to success. On-demand and streaming services upended this model, creating an industry with virtually infinite content supply. Success in streaming is not only about a handful of marquee shows but about offering a breadth and depth of content that engages as many household members and demographics as possible.

While major, must-watch series remain important, Netflix has prioritized a large volume of “good enough” programming and more niche content that may not be a cultural phenomenon but keeps a wider range of subscribers engaged.

Ironically, initially many in Hollywood mocked Netflix’s aggressive programming strategy, which diverged sharply from the traditional model. Several years ago an infamous Saturday Night Live skit parodied Netflix’s programming strategy, mocking the company's relentless drive to produce a vast volume of content catering to diverse audiences. The sketch exaggerated Netflix's approach, portraying its management team as greenlighting virtually any show idea in pursuit of engagement and subscriber retention.

Netflix has also pursued global scale by aggressively investing in local content, recognizing that international markets were key to sustaining long-term growth. The company demonstrated that in a streaming world, cross-border content has far greater reach than it did in the cable TV era, where distribution was more limited. Netflix has also excelled at reviving old series, either by producing new seasons or securing licensing deals. Its vast audience and global reach allow it to introduce these shows to new viewers, extending their lifespan and maximizing their economic value.

Streaming’s Competitive Future: More Than One Player Will Thrive

Netflix’s competitive position is driven by its scale, content strategy, and disruptive model. Its larger revenue base, higher profitability, and strong financial position fuel a positive flywheel, enabling greater investment in talent, content expansion, and engagement—reinforcing its pricing power. Additionally, Netflix’s financial strength enables greater investment in technology infrastructure, user interface, and data analytics, enhancing its recommendation engine, and overall user experience.

However, despite these advantages, Netflix holds only an estimated 8-9% share of U.S. TV viewership. This highlights the fragmented nature of the video and streaming landscape, where diverse content supply and varying viewer preferences across demographics and cultures make it challenging for any single platform to capture a dominate market share. As a result, there remains significant opportunity for other players to thrive in the global streaming market.

Internet platforms like Google Search and Uber in the U.S. ride-share market exemplify the market share concentration typical of winner-takes-most internet verticals. However, the factors driving such dominance—network effects, high switching costs, and two-sided marketplace advantages—are less applicable to streaming.

Unlike platforms like Facebook, where the value of the service increases for existing users as more users join, streaming services do not benefit from direct network effects. Netflix’s moat is the breadth and depth of its content, not user connectivity. Nor does Netflix have high switching costs or lock-in mechanisms like Amazon Prime or the Apple iOS ecosystem.

Ultimately Netflix must differentiate on content not just scale. Consequently, the company is not standing still, seeking to expand its competitive advantage—expanding into adjacent markets, including live sports, to drive even higher engagement and reinforce its pricing power.

Consolidation – A path forward

Creating viable competitors to Netflix requires adopting key lessons from its content strategy—offering enough variety to drive engagement across demographics, improve retention, win back former subscribers, and convert users from rival services. A high-engagement service at scale is critical to achieve a stickier subscriber base to compete effectively and create pricing power.

Legacy media companies generally match or exceed Netflix’s total content spend. However, their challenge isn’t increasing spending but optimizing how and where they allocate resources. Their budgets are spread across linear TV, big-budget theatrical films, expensive sports rights, and streaming platforms. Achieving “scale” isn’t about a single expenditure figure; it’s about concentrating spending where it maximizes the volume and variety of content available to streaming audiences, balancing content quantity and quality—a necessary shift from traditional media’s historical mindset.

A sustainable, multi-winner streaming industry will require restructuring among legacy players, including disentangling from declining linear TV assets and preparing for much needed streaming platform consolidation. While some mergers may face antitrust scrutiny, the current U.S. administration is expected to be more receptive to corporate M&A activity.

Strategic moves are already underway to reshape the legacy media landscape to support the separation and consolidation of traditional TV assets. Comcast is spinning off its cable TV assets, while Warner Bros. Discovery is separating its business into two divisions: linear networks and studio & streaming.

At the same time, there is a growing trend toward bundling competing streaming services. Examples include the Disney+, Hulu, and Max bundle and cable operators bundling streaming services such as Comcast has with Peacock, Apple TV+ and Netflix. While not as transformative as full mergers, these partnerships drive higher engagement, new subscriber growth, and lower churn. In the last two quarters of 2024, the Disney and Warner bundle has reportedly driven approximately 20% of total Max subscriber growth and featured strong customer retention.

Although these new streaming bundles represent progress, full mergers between competing platforms likely will be necessary to achieve the scale and engagement needed for long-term competitiveness. Consolidation will unlock both revenue and cost synergies, creating businesses that are more valuable together than as standalone entities.

The pending Skydance Media and Paramount transaction could serve as a catalyst for further industry consolidation. A merger between Warner’s Max streaming service and Paramount+ or Peacock would create a streaming platform with the scale and content diversity needed to emerge as one of the dominant streaming "super apps."

Such a combination would bring together a compelling mix of general entertainment, live sports, and premium content, enhancing the ability to attract advertisers and capture ad dollars shifting from linear TV to streaming. A broader mix of content would also accelerate subscriber growth, strengthen engagement, and improve long-term sustainability in the increasingly competitive streaming industry.

Despite strong growth and continued execution of its product roadmap, Nvidia's stock has lost momentum after a spectacular two-year run, underperforming the S&P 500 by approximately 15% since mid-June 2024.

While Nvidia remains the dominant AI player with a formidable economic moat, the lack of meaningful upside in the company’s April quarter revenue guidance challenges its beat-and-raise narrative. This lack of upside is notable given the expectation of potential pull-forward demand from China ahead of the Framework for the Diffusion of Advanced Artificial Intelligence Technology export restrictions, which took effect on January 13, 2025, with most provisions requiring compliance by May 15, 2025.

Following the recent sell off in the stock, I expect a short-term rebound leading up to Nvidia’s GTC event, which begins on March 17th with CEO Jensen Huang’s keynote address on March 18th, where the company will share its outlook and what’s next in AI.

However, looking beyond the near term, what lies ahead for Nvidia's stock? Risks to both direct and indirect (via Singapore) China revenue raise concerns about the sustainability of Nvidia’s growth trajectory in the second half of 2025, especially given elevated investor expectations. While the ramp of Nvidia’s new Blackwell platform reinforces near-term growth, limited upside to revenue and earnings forecasts coupled with risks to China-sourced revenue point to downside risk in the stock heading into and following the company’s next earnings report in May.

Subscribe now

Below are five key points highlighting why Nvidia’s valuation is under pressure, the evolving AI investment theme, and the growing risk of AI compute capacity outpacing demand—heightening the likelihood of a cyclical slowdown in AI infrastructure spending and downside risk to Nvidia’s second half 2025 and 2026 revenue forecasts.

1.) The market is implicitly assigning a lower multiple on China earnings, making this the dominate factor behind the stock’s valuation compression.

While concerns over a cyclical peak in AI capex spending have weighed on Nvidia’s stock, a more tangible and quantifiable risk is its direct and indirect (via Singapore) sales exposure to China. Singapore and China sourced revenue collectively accounted for 31% of Nvidia’s total revenue in the fiscal year ending January 2025. This helps explain why despite stronger-than-expected 2025 capex guidance from leading cloud players directly countering fears of a cyclical peak in AI compute investment, Nvidia’s stock has remained under pressure.

In FY25, sales to Singapore surged 3.5x year-over-year—outpacing Nvidia’s overall revenue growth, which more than doubled. In FY25 Singapore became Nvidia’s second largest geographical source of revenue, accounting for 18.1% of total revenue up from 11.2% in the prior year. This surge in growth is widely attributed to the use of Singapore as a transshipment hub, enabling China to indirectly acquire advanced GPUs despite U.S. export restrictions.

The U.S. government's Framework for the Diffusion of Advanced Artificial Intelligence Technology aims to restrict China's access to advanced GPUs and close existing loopholes, posing a significant risk to this revenue stream and could create an immediate headwind for Nvidia’s growth outlook. While the Trump administration may introduce new rules that supersede this framework, it is unlikely to fully reverse restrictions on China's access to more advanced GPUs.

2.) Similar to the telecom bubble of 2000, portions of AI compute infrastructure spending face quality-of-demand concerns.

While the largest cloud service providers dominate AI infrastructure spending, neocloud players represent a meaningful share of demand for Nvidia GPUs and AI compute systems. This cohort of companies comprises a mix of reinvented cryptocurrency miners that have opportunistically pivoted to exploit AI data center growth, existing smaller cloud players shifting focus to AI compute services, and more typical startups.

Neocloud companies have raised significant capital through equity and debt financing, often using chips and AI hardware systems as collateral. This funding has fueled investments in GPUs and AI systems, and the build-out of new AI compute capacity.

Much of this AI compute capacity is built in anticipation of future demand, making these companies speculative expenditures particularly vulnerable to any oversupply in AI compute capacity.

Additionally, there are concerns over the quality of AI compute demand from portions of cloud players customer base, as much of it is fueled by venture capital-backed high-cash burn startups rather than established, revenue-generating enterprises. This raises questions about the sustainability of the current AI compute spending growth trajectory. Nvidia CEO Jensen Huang highlighted this source of demand for cloud players in Nvidia’s August 2024 earnings call saying - “The number of generative AI startups is generating tens of billions of dollars of cloud renting opportunities for our cloud partners.”

3.) Measuring AI compute capacity relative to demand is challenging, as capacity expands not only through new data center builds but also via efficiency gains.

Efficiency gains in model design, system optimizations, and de-bottlenecking efforts that boost GPU utilization, effectively expand existing AI compute capacity. Algorithmic advancements reduce the compute requirements for training AI models, with the potential for continuous improvements in AI algorithm efficiency.

This is not inherently negative for Nvidia’s future demand, but is a difficult to quantify risk. In fact, algorithmic and model efficiency improvements lower AI compute costs, potentially accelerating AI application adoption.

However, the shift from the AI training infrastructure buildout to inference-driven demand may not be seamless, and could result in a period where AI compute supply outstrips demand.

4.) As the AI investment theme shifts from infrastructure spending to applications, Taiwan Semiconductor Manufacturing Corp (TSMC) presents a more compelling opportunity than Nvidia, which trades at nearly a 40% price-earnings multiple premium to TSMC.

As highlighted in my post Surf's up- Catch the AI Smartphone Wave, technological breakthroughs in AI model efficiency are poised to make larger AI models more viable for on-device processing. This will enable more advanced on-device AI features and applications, igniting growth in AI smartphones and consumer devices. With 35% of TSMC’s revenue coming from smartphones, this trend will expand TSMC’s AI addressable market beyond AI infrastructure, positioning it as a key beneficiary of the next wave of AI adoption.

Additionally, regardless of whether demand for merchant GPUs or custom AI chips grows at a faster pace —or whether Broadcom and Marvell or Taiwanese competitors like Alchip and Mediatek are gaining ground in the custom AI chip market—TSMC remains the ultimate winner as the near-monopoly supplier of advanced foundry manufacturing.

5.) Nvidia has both a track record of huge success, but also one of significant cyclical corrections in revenues, with no real signal from management of this risk in advance of the last two downturns.

In the past six years, Nvidia’s revenue has fallen short of original expectations about one-third of the time, with seven quarters during this period experiencing year-over-year revenue declines. This includes a 24% YoY revenue drop in the January 2019 quarter, followed by three successive quarters of declines. Similarly, after a disappointing July 2022 quarter, revenue declined by 17%, 21%, and 13% YoY in the following three quarters.

Different variables led to each of these cyclical downturns, the point being demand for Nvidia is dictated by external factors and a positive outlook from the company should not be simply taken at a face value. Nvidia's ultimate revenue end point still illustrates an extremely strong long term growth trajectory, however one that still can manifest significant and difficult to predicate cyclical volatility.

The negotiations to end the Russia-Ukraine war have prompted a range of reactions regarding the broader direction of the Trump administration’s foreign policy, with particular attention on China and Taiwan. Many view this as part of a broader realignment of foreign policy, defined not by shared values but by the principle of stronger nations asserting dominance over weaker ones.

At the center of these concerns are President Trump’s framing of the Russia-Ukraine conflict, along with preemptive concessions to Russia in negotiating a settlement.

Critics argue that Trump’s approach signals a broader shift toward appeasement, with some warning that Trump is undermining democratic values and aligning with autocratic regimes to reshape the global order into spheres of influence. However, while his attention to Greenland and the Panama Canal suggests an interest in great-power spheres of influence, this in no way indicates a disregard for U.S. strategic and economic interests beyond the Western Hemisphere.

Trump’s political opponents often point to his stance on Russia as evidence that he will appease China, prioritizing a trade deal that grants him a political "win" while sacrificing long-term strategic interests, ultimately paving the way for Taiwan to fall under Beijing’s control. The argument is that Trump’s relationship with Xi Jinping will follow the same pattern as his dealings with Vladimir Putin—a transactional approach that, in their view, risks compromising U.S. strategic interests in favor of personal diplomacy and short term trade concessions.

Such a conclusion on the direction of U.S.- China policy is flawed and contradicts Trump's core motivations, the actions of his administration, the hawkish national security team he has assembled, and—specific to Taiwan—ignores the immense economic and strategic consequences the U.S. would face if China gained control of the island and the world’s most valuable technological asset, Taiwan Semiconductor Manufacturing Corp (TSMC).

Subscribe now

Transactional Diplomacy: Trump’s Global Playbook

While Trump’s worldview may emphasize "might makes right," reducing his foreign policy to this alone is overly simplistic. A broader understanding of Trump's approach to policy—both foreign and domestic— comes from examining his core motivations. Two consistent driving forces shape his stance on various issues.

First, personal standing and public perception play a crucial role. Trump views his success through measurable outcomes—elections, polls, economic indicators, and stock market performance—assessing how events enhance his image as a leader, businessman, and historic figure.

Second, Trump sees the world through a transactional lens. Whether dealing with allies, adversaries, or economic policies, his perspective is shaped by whether he believes he—or the U.S.—is getting a good deal or being taken advantage of. This deal making mentality influences everything from trade negotiations to military alliances, often reducing complex relationships to a balance sheet of perceived wins and losses.

While this approach risks prioritizing short-term transactional gains over long-term strategic considerations, there are indications of broader connectivity in Trump’s policies. His foreign policy reflects a deliberate effort to recalibrate U.S. global positioning based on economic leverage and bargaining power. Ceding Taiwan and TSMC to China would directly contradict this strategy, representing an extraordinary strategic loss that would severely weaken U.S. economic power while damaging Trump’s historical legacy and personal standing.

Rather than pursuing a purely isolationist or authoritarian-aligned foreign policy, Trump’s approach appears to center on redefining alliances and adversarial relationships through the lens of economic and geopolitical advantage. His rhetoric and tactics often create uncertainty, but they also serve as leverage in negotiations.

It is noteworthy that Trump’s aggressive rhetoric is not uniformly directed at all of the U.S.’s long-term and closest allies. While his administration has frequently expressed dissatisfaction with Canada and European nations, Trump has simultaneously fostered and maintained good relationships with two of the U.S.’s most critical allies: Japan and Israel.

This contradicts the narrative that Trump is systematically turning against all traditional allies in favor of autocratic regimes. Instead, his approach suggests a recalibration of alliances, where he applies pressure on allies he perceives as benefiting disproportionately from U.S. trade or security guarantees.

However, even Trump’s staunch support for Israel underscores his transactional mindset. His recent claim that the U.S. should own Gaza is a prime example of this perspective. In Trump’s deal-making worldview, decades of U.S. military and political support for Israel equate to a form of investment—one for which he likely believes the U.S. is entitled to compensation. In this context, suggesting that Gaza should effectively become a U.S. asset follows the same logic as striking a mineral rights deal with Ukraine. This reflects Trump's mentality, where foreign policy is often framed in terms of ownership, transactions, and leverage.

Ukraine-Russia Calculus: Strategy, Leverage, and Deals

Trump's comments blaming Ukraine for the war and accusing President Zelensky of being a dictator surely to some extent stem from his personal history with Zelensky—specifically, the infamous phone call that led to his first impeachment—as well as his transactional mindset, which questions what the U.S. is getting in return for its financial and military support to Ukraine. This also explains his pursuit of a mineral rights deal with Ukraine, he simply sees U.S. aid as a direct monetary cost that should yield a tangible return.

Arguably, this mineral rights deal could enhance Ukraine’s security by giving the U.S. a vested economic interest in the country's stability. By having "skin in the game," the U.S. would be more incentivized to ensure Ukraine's long-term security and sovereignty.

The broader concept of standing up to Russian aggression and defending international law is too abstract for Trump. His focus is not on upholding global norms, but rather on assessing whether supporting Ukraine advances U.S. national interests and what the U.S. is getting in return—particularly when weighed against the risks of escalating tensions with a nuclear power.

Trump likely views Russia as a “no-cost” relationship, where Putin flatters his self-image without demanding much directly from Trump in return. Unlike China, Russia is a diminished power that does not pose a comparable strategic threat to U.S. interests. Its conventional military strength has been severely weakened—to the point of needing North Korean soldiers to fight in Ukraine—highlighting limitations to its ability to project power well beyond its borders.

Trump’s approach to negotiating an end to the war—bolstering Putin’s standing and treating Russia as an equal—could be seen as a strategy to offer Putin a face-saving exit. However, this approach sends the wrong message, signaling tolerance for unprovoked military aggression. While personal diplomacy may aid negotiations, it is a mistake to imply that military aggression against a sovereign state can be justified or will be tolerated. It remains to be seen whether Trump’s approach to Putin serves a purpose beyond negotiating an end to the Russia-Ukraine war, and is part of a broader strategy to weaken the bond between Russia and China.

Misplaced Concerns About China Appeasement

President Trump’s stance on the Russia-Ukraine war does not indicate a similar strategy toward China. The early actions of the new Trump administration clearly indicate a hawkish stance toward China, not a policy of appeasement. Trump has already imposed an initial 10% tariff on all Chinese imports followed by an additional 10% tariff effective March 4, taken steps to eliminate the de minimis trade exemption that benefits discount Chinese retailers like Temu and Shein, and recently threatened additional restrictions on semiconductor sales to China.

Additionally, the Trump administration recently issued fact sheet on foreign investment and a separate memo America First Trade Policy not only reinforce a hardline stance on China but also highlight the administration’s focus on strengthening U.S. industry and securing its supply chains to mitigate national security risks posed by overreliance on China.

Some point to Trump’s comments on TikTok and his efforts to broker a sale as evidence that he may be soft on China. While I believe downplaying the platform’s use as a tool in China’s digital propaganda war against the U.S. is a grave mistake, this stance to a large extent reflects Trump’s belief that TikTok’s was helpful in his reelection efforts rather than a broader representation of his China policy intensions.

For years, the U.S. has tolerated a lopsided relationship in which China has restricted access to its domestic market, stolen intellectual property and trade secrets, subsidize Chinese companies that undercut on price U.S. competitors, and engaged in an ongoing digital propaganda war against the U.S. At the same, China has also aggressively expanded its regional ambitions unchecked, including constructing militarized artificial islands in the South China Sea, threatening key international shipping lanes.

Yet, these ongoing realities—representing actual Chinese aggression—have been tolerated by many who now voice alarm over Trump’s foreign policy. These concerns have led to oversimplified conclusions, such as drawing a straight line from Trump’s stance on Russia to an increased risk of conceding Taiwan to China. Such assumptions contradict his administration’s historically aggressive stance on China and overlook, as I detailed in my post Taiwan is Arrakis, the critical strategic value of Taiwan and TSMC to both the U.S. and the global economy.

Trump’s comments threating tariffs on semiconductors imports from Taiwan are not a sign of reduced support for Taiwan, but rather consistent with Trump’s transactional mindset. To the contrary his administration is already sending symbolic, but important signals that Chinese military aggression against Taiwan will not be tolerated. For example, a joint statement following a recent meeting between President Trump and Japan’s Prime Minister Ishiba Shigeru emphasized the importance of maintaining peace and stability across the Taiwan Strait, and clearly stated strong opposition to any unilateral attempts to change the status quo by force or coercion.

Additionally, the U.S. State Department recently removed the phrase “we do not support Taiwan independence” from its fact sheet on U.S.-Taiwan relations—a significant policy shift that prompted a predictable response from China.

Strong support for Taiwan aligns with current U.S. interests, given the strategic importance of TSMC and is consistent with Trump’s foreign policy approach where U.S. economic leverage is central to advancing U.S. national security interests. As discussed in my post China-Taiwan Unfinished Business, understanding Trump’s transactional mindset and negotiation tactics provides better insight into what he is likely attempting to achieve.

As with many of Trump’s statements, it is often more revealing to observe his actions rather than his words. The National Security team assembled includes prominent China hawks in key positions, which provides a more reliable indication of the administration’s stance on Taiwan and China. Furthermore, it is highly unlikely that Trump would pursue policies that increase the risk of a Chinese invasion of Taiwan—an event that would bring significant damage to the U.S. economy and weaken American influence in the Asia-Pacific region.

Concerns over semiconductor tariffs are also likely overstated. U.S. companies dominate semiconductor imports, and such remarks likely aim to promote domestic advanced semiconductor manufacturing. While this is a sound strategic goal, it will take years to achieve. In the meantime, U.S. reliance on TSMC for advanced semiconductor production will remain unchanged in the medium term.

However, President Trump’s comments have had some tangible impact, particularly his call for Taiwan to increase its defense spending and prioritize boosting its military capabilities. Taiwan’s President Lai Ching-te has responded by committing to greater investments in the U.S. and proposing a “special budget” to raise Taiwan’s defense spending to over 3% of GDP.

An X-factor that could impact U.S.-China policy and has raised concerns is the influence of Elon Musk. Given Tesla’s business interests in China, Musk has consistently avoided criticizing China’s authoritarian regime, repeating China’s view that Taiwan is a fundamental part of China. As outlined in my post China-Taiwan Unfinished Business, this characterization is historically inaccurate. Musk’s comments suggest he views Chinese control of Taiwan as inevitable, carefully avoiding any direct challenge to the China’s stance.

However, Musk’s silence on China is not unique. Apple CEO Tim Cook, whose company is even more reliant on China than Tesla, has similarly refrained from public criticism. The key unknown is what Musk says in private. While his broader business interests align with the U.S., his business interest in China raises concerns about whether his advice to Trump in some circumstances would prioritize his own financial considerations over long term U.S. strategic interests.

That said, Musk is an important but singular voice in Trump’s orbit. There are numerous influential figures in both Musk’s and Trump’s circles who recognize the strategic threat posed by China and are committed to policies that counter and reverse past economic and geopolitical imbalances. While Musk’s influence is notable, it is unlikely to singularly dictate Trump’s China policy.

Conclusion

Trump’s foreign policy often appears chaotic, with less than clear strategic goals, largely because it is driven more by transactional deal-making than by ideology. His approach to Russia and Ukraine, though controversial, reflects a broader pattern of assessing relationships based on perceived costs and benefits rather than traditional geopolitical strategy.

Despite concerns about his stance on Russia, there is little evidence to suggest Trump will adopt a conciliatory approach to China. In fact, certain aspects of his efforts to resolve the Russia-Ukraine conflict could expand U.S. leverage and signal a tougher stance. Improving ties with Putin could serve as a strategic move to drive a wedge between Russia and China. Additionally, securing a mineral rights deal with Ukraine would also be a means to reduce U.S. reliance on China for some critical minerals such as rare earth metals. Ultimately, while Trump’s foreign policy may often seem highly transactional, it does not indicate a shift toward appeasing China.

Amid the ongoing debate over Tesla's Full Self-Driving (FSD) technological capabilities, a critical topic remains largely overlooked: Tesla's go-to-market strategy for robotaxis. While Tesla has a definitive technology roadmap for autonomous vehicles, its operational and commercial strategy for this service appears to still be evolving.

Tesla has announced plans to launch an unsupervised robotaxi service in Austin in June 2025, aiming for expansion into multiple cities by year-end. However, this rollout will initially rely solely on an internal fleet—not individual Tesla owners’ vehicles, a distinction from Musk’s vision of a shared fleet model.

The initial reliance on an internal fleet underscores the complexities of transitioning from supervised to unsupervised FSD. While Tesla aims for a direct shift to a driverless robotaxi service, delays are likely, and expansion to additional cities may take longer than expected. Beyond technological hurdles, Tesla must also build a rideshare platform while simultaneously developing fleet management and logistics infrastructure on a city-by-city basis.

Rather than building the operational capabilities and infrastructure necessary for a nationwide rollout, Tesla could acquire them. Acquiring Lyft would provide immediate access to a proven marketplace, fleet management infrastructure, local market regulatory expertise, and an established rider base that could accelerate the rollout and mitigate initial risks in launching and scaling its robotaxi service.

Even at a premium to its current share price, Lyft's market capitalization would be less than 1% of Tesla's, making an all-stock acquisition easily digestible. Lyft would be a more valuable asset owned by Tesla than on a standalone basis and help Lyft overcome its subscale vulnerabilities.

Strategically owning this platform would enable Tesla to expand its robotaxi fleet while offering a hybrid service during the multi-year transition from driver-based to driverless ridesharing. As the market evolves toward fully autonomous transportation networks, Tesla could deliver an integrated platform combining driver-based and driverless services—creating a potential point of differentiation that could challenge Uber in markets where it does not have an active robotaxi partnership.

Subscribe now

There has been speculation that Amazon could be interested in acquiring Lyft to accelerate the rollout of its Zoox robotaxi service. Any interest from Amazon would reinforce the idea that Lyft's platform could be a valuable asset for the successful launching and scaling of a robotaxi service.

While Lyft could accelerate Tesla's robotaxi rollout, the rise of robotaxis also signals a structural shift that challenges marketplaces like Lyft and Uber. This risk may create an opportunity for Tesla to acquire Lyft at a favorable price, given the potential challenges facing Lyft as a subscale player in a rapidly evolving market.

Traditional rideshare marketplaces thrive on a market structure characterized by highly fragmented driver supply, which enhances the value and economic leverage of a two-sided marketplace. However, in a robotaxi market with a more concentrated supply of services, the utility of third-party platforms matching supply and demand diminishes, with more of the value extracted from this service accruing to the technology enablers. Early robotaxi deployments are also likely to be supply constrained, which will reduce marketplace reliance—much like Waymo in San Francisco, which bypasses third-party platforms in favor of a direct-to-consumer model.

Tesla’s Ambitious Robotaxi Launch Timeline

As highlighted in my note The Ultimate “Killer App,” while Tesla’s pursuit of a pure end-to-end autonomous vehicle system holds significant promise, this approach still faces technical challenges in handling real-world variability and edge-case scenarios, and overcoming these hurdles may take considerable time.

The company’s decision to initially launch in a fixed geographic area, similar to Waymo, makes sense, as Tesla’s unsupervised FSD will likely perform better on familiar, frequently traveled roads. However, a direct transition from supervised FSD to an unsupervised, driverless robotaxi service without a lengthy interim testing phase seems ambitious and unlikely. For context, Waymo spent considerable time testing its unsupervised system with safety drivers before transitioning to a fully driverless service. Given this precedent, Tesla’s rollout in Austin and subsequent expansion to other cities may be more gradual than currently expected.

Faced with competitive pressure from Waymo’s rapid expansion, Tesla could opt for a "compound AI" system approach in its initial robotaxi deployment—incorporating more engineered elements alongside extensive use of neural networks. While still pursuing its long-term goal of a pure end-to-end neural network system, this pragmatic shift could accelerate commercialization, improve reliability, and pose a more immediate challenge to Waymo’s first-mover advantage.

The Advantages of a Hybrid Approach

A Lyft acquisition would provide Tesla with a hybrid rideshare-robotaxi platform that ensures consistent vehicle supply while transitioning to full autonomy. Additionally, Lyft's platform could support Tesla's expansion in regions where driverless services are not yet permitted or feasible. The infrastructure and regulatory experience Lyft brings—particularly in working with city governments and transportation authorities—would also help Tesla navigate the complex requirements for deploying robotaxi services at scale.

Lyft’s nearly 25 million active riders and more than 40 million annual riders would give Tesla instant scale, reduce customer acquisition costs, and provide valuable consumer data. Lyft’s existing rideshare marketplace would enable Tesla to increase supply density, reduce wait times, and improve customer adoption as it builds out its robotaxi service. It would also allow Tesla to support rideshare use cases that Tesla vehicles cannot currently meet, such as those requiring larger passenger capacity.

Lyft's partnership with May Mobility and its plans to ally with Mobileye—targeting fleet operators deploying purposed built vehicles equipped Mobileye’s AV systems on Lyft’s platform, as evidenced by its recent three-way partnership with Mobileye and Marubeni—should not deter a Tesla acquisition. Instead, these collaborations highlight the value of Lyft's marketplace as the rideshare industry shifts toward driverless robotaxis.

Lyft – Ready-Made Logistics Infrastructure

Tesla has provided limited information about its plans for the operational infrastructure needed to support its robotaxi service. While Tesla has posted a job opening for teleoperations engineers, this alone provides limited insight into the current state of its robotaxi operational support infrastructure development.

In contrast, Lyft could offer Tesla an advantage in this area. Lyft’s Flexdrive subsidiary, which is already converting its car rental locations into robotaxi depots, could provide Tesla with a ready-made platform for fleet operations, including dispatch, charging, maintenance, and customer support.

A Third-Party Supply Strategy – Will Consumers Bite?

Tesla’s long-term strategy of relying heavily on individual owner vehicle supply for its robotaxi network is questionable. While Tesla-owned Cybercabs may serve as a supply buffer, the overarching plan appears to be an asset-light model, with most vehicles coming from fleet operators and individual owners.

While Elon Musk and Tesla's management have a proven track record of innovation and turning what others deem impossible into reality, there have been some recent missteps in gauging consumer tastes and behavior—as I discuss in my post What’s Next for Tesla -

Elon Musk and Tesla's management team have recently faced challenges in their EV business, largely stemming from misjudging consumer tastes. As recently as Fall 2023, Tesla’s management continued to reference its 50% EV volume CAGR target originally outlined in early 2021. However, EV volume growth projections have since become more cautious. Against this backdrop, Tesla overestimated the sales potential of the Model 3 and Model Y by underestimating consumers' demand for variety and choice. More recently the radical design of the Cybertruck has failed to resonate with most consumers, leading to sales significantly below expectations.

Musk and Tesla's management have also misjudged consumer reactions to the company’s vehicle price cuts over the past two years, which were implemented to support EV revenues. This strategy has likely eroded Tesla’s brand equity, particularly by contributing to a sharp decline in vehicle residual values, undermining customer confidence in the long-term value of their purchases.

Mr. Musk may be misjudging consumers again, describing Tesla's future robotaxi service as an “Airbnb on wheels.” Tesla’s vision for a third-party vehicle supply model, in which Tesla owners contribute their vehicles to the robotaxi fleet, may overestimate consumer willingness to participate. Potential concerns include privacy and liability risks, coordination of personal and shared vehicle use, potential damage caused by riders, insurance costs, and uncertainty over availability during emergencies. These challenges could limit adoption and create supply constraints that hinder Tesla’s ability to meet demand.

Over-reliance on third-party supply could lead to unpredictable vehicle density, resulting in longer wait times and lower service reliability. While Tesla has not shared estimates of how many owners would opt into a robotaxi network, insufficient supply density would weaken service viability in many areas. The novelty of taking a robotaxi service is diminished if a rider cannot get to their destination at the desired time. A better strategy would likely involve a mix of Tesla-owned vehicles and fleet operators rather than excessively relying on individual owners.

Conclusion

While Tesla continues to make advancements in FSD, its go-to-market strategy for robotaxis continues to evolve. Acquiring Lyft would instantly provide Tesla with the infrastructure, fleet management capabilities, and customer base to scale robotaxi services more effectively over time. Tesla’s acquisition of Lyft could accelerate deployment, improve consumer adoption, and potentially capture market share at Uber’s expense.

Financially, Lyft is current trading at an undemanding valuation of approximately 7x 2026 EBITDA. Lyft remains an inexpensive asset, yet is still challenged as a weaker, sub scale No. 2 player. Under Tesla’s leadership, operational efficiencies and cost synergies could drive improved profitability while potentially accelerating Tesla’s robotaxi rollout. Given Tesla’s market capitalization, an all-stock acquisition could be structured in a way that is non-dilutive to shareholders, making it a financially viable and value enhancing transaction.

While the recent cyclical downturn experienced by semiconductor companies like Texas Instruments, ADI, Microchip, ON Semiconductor, and NXP exhibits characteristics of a typical cyclical slowdown, it also signals a deeper shift in the competitive landscape. Although there is ample attention to the growing long term competitive threat from China, the full magnitude of this challenge remains underestimated.

Backed by government subsidies and China’s debt-fueled industrial policy, Chinese semiconductor firms are aggressively expanding capacity, escalating competition while playing by a different set of rules. This subsidized competition distorts market dynamics, undercuts foreign firms through non-market pricing, and threatens U.S. and European companies.

The U.S. can no longer remain passive as Chinese firms gain ground through unfair competitive advantages. The Trump administration must continue advancing reciprocal policies to safeguard U.S. economic and corporate interests, including stronger trade enforcement and targeted investment incentives to counter China’s state-backed expansion.

Subscribe now

Management teams that underestimate this threat do so at their peril. Across industries, Chinese firms have consistently exceeded expectations, leveraging aggressive pricing and steady product improvement to move up the value chain. In industrial, automotive, and consumer electronics, Chinese semiconductor suppliers don’t need components that outperform foreign competitors—being "good enough" is sufficient to gain market share as Chinese buyers align with national semiconductor industry goals.

Even as U.S. export restrictions limit China’s access to advanced semiconductor production equipment, Chinese companies are investing in mature manufacturing technology nodes, targeting analog, microcontroller, and power semiconductors, that are the backbone of the auto and industrial markets.

Quantifying the scale of China’s semiconductor capacity expansion and the risk of overcapacity is challenging due to gaps in data on fab output, utilization rates, and production yields. However, semiconductor equipment sales growth to China serve as a reliable proxy, indicating a production base expanding well beyond end-market growth.

This expansion persists despite weak global demand. The analog, microcontroller, and power semiconductor markets likely already suffer from overcapacity, even accounting for suppliers currently under-shipping relative to end-market demand. Rising on-balance-sheet inventories further signal supply-demand imbalances. Texas Instruments, an industry bellwether, saw inventory grow from $2.4 billion in Q3 2022 when the company’s quarterly revenue peaked to $4.5 billion by the end of 2024—equivalent to an increase from 133 days of sales to 241 days.

Across industries, China has consistently pursued national strategic goals, resulting in overcapacity and long-term industry disruption. Even if most of China’s semiconductor capacity remains domestically focused, its impact on global competition will be significant.

With China as the world’s largest semiconductor market, even a gradual displacement of foreign suppliers will erode China-sourced revenue for Western firms, triggering a ripple effect beyond China. As companies seek to offset lost revenue, competition will intensify, further pressuring market dynamics worldwide.

China Inc.’s Debt Driven Growth Model

China’s semiconductor industry is following a familiar pattern seen across its domestic industries: a debt-fueled, state-backed growth model where profitability and cash flow take a back seat to strategic objectives. China aggressive pursuit of self-sufficiency in semiconductors, is using state subsidies and strategic credit allocation to accelerate the expansion of domestic production, rapidly scale operations, and undercut foreign competitors.

The control of credit is critical to the Chinese Communist Party's (CCP) grip on the economy, with government-controlled non-bank financial firms, local government finance vehicles (LGFVs), and state-owned banks serving as the primary channels for policy-directed lending. Corporate credit growth is essential to sustaining China's economic growth, leading to an increasingly leveraged economy. Total corporate and LGFV debt is more than 170% of GDP, more than twice that of the U.S., with total debt (total social financing) to GDP reaching 303% in 2024.

China’s economic model systematically uses debt as a substitute for cash flow. This feature of China’s growth model has resulted in a continuous decline in the productivity of credit as a greater portion of loans are allocated to cover existing interest payments and uneconomic investment projects. Consequently, China requires an increasing amount of credit to generate a unit of GDP growth. Using China’s broader measure of credit, total social financing (TSF), the ratio of credit required to generate one unit of GDP has increased from 2.66 in 2017 to 5.52 in 2024.

The collapse of China's residential real estate market is a prime example highlighting the risks and fragility of its Ponzi scheme-like industry structures driven by debt fueled growth and speculative investment activity reliant on continuous credit expansion.

China's Semiconductor Sector: A Subsidy-Fueled Push

This growth model is now being applied in China’s semiconductor sector. Competition, in principle, is a catalyst for operational efficiency and product innovation. Yet, China's semiconductor industry is not competing under traditional market-based principles. Western firms face not only competition but also state-backed industrial policy, financing structures, and subsidies that distort fair competition.

While many players in China’s semiconductor industry are private, making it difficult to get a complete financial picture, the financial performance of publicly listed Semiconductor Manufacturing International Corporation (SMIC) offers some insight.

The implicit guarantee of government support allows SMIC and its Chinese peers to operate and gain share using anticompetitive and non-market means. SMIC’s financials differ drastically from those of U.S. peers or its Taiwanese counterpart, TSMC. While TSMC and U.S. firms are highly cash generative, SMIC continues to burn cash, driven by a capital expenditure binge aimed at meeting national strategic goals.

Over the past 5 years SMIC had an estimated US$14 billion in negative free cash flow, with its capital expenditures equivalent to approximately 100% of revenue during this period. In 2024 alone, SMIC’s capital expenditures were 2.3 times its depreciation expense. These metrics reflect an unsustainable level spending without government support.

SMIC’s gross margin of about 20% is also significantly lower than TSMC’s average gross margin of 55% over the past two years. While a heavy depreciation expense burden contributes to this, it also suggests that SMIC is aggressively pricing its services, lowering the cost for its customers that compete with foreign semiconductor players.

China’s domestic demand from industries like electric vehicles (EVs) further contributes to this distorted industry structure. Although EVs remains a critical auto semiconductor growth driver, this demand is propped up by an industry reliant on unprofitable players. While China leads in EV production, this growth is built on a weak financial foundation, with the majority of China's EV OEMs operating at a loss.

Enabling China's Semiconductor Ambitions

China’s push for semiconductor self-sufficiency has been supported by U.S., Japanese, and Dutch semiconductor equipment suppliers, which have significantly increased sales to the region. Before COVID, China accounted for around 30% of total sales for many of these companies, including a substantial share to non-Chinese semiconductor firms operating within China.

In recent years, however, this exposure grew to over 40% in some quarters, with the majority of sales now directed to Chinese companies. For instance, Applied Materials’ China sales reached $10.1 billion in FY24, up from $4.2 billion in FY19. In FY24, China revenue accounted for 37% of Applied Materials total sales, with China revenues above 40% on a quarterly basis in two of the four quarters. The implementation of export restrictions has tempered these figures but still leaves China as a critical market.

The Downturn and Chinese Competition

The analog, microcontroller, and power semiconductor markets have experienced a severe downturn and are now searching for a bottom and path to recovery. Company revenues have declined significantly, with quarterly peak-to-trough declines averaging over 30%. NXP sales has been relatively resilient, declining nearly 20%, while Microchip has suffered a 50%+ drop. Corporate margin trends also remain weak, reflect rising pressure from declining utilization rates, the unwinding of prior price increases, and likely intensifying price competition in China.

Leading analog, MCU, and power semiconductor firms have China sales exposure ranging from approximately 20% to 35%, but China’s importance extends beyond revenue share. Its rapid adoption of EVs and advance driver assistance systems has outperformed other regions, making it a key market for auto semiconductor growth.

While demand in China has been more resilient, this strength is relative to weaker regions and largely driven by strong demand from electric vehicles with high semiconductor content. However, this provides false reassurance, as China’s EV market is ripe for Chinese semiconductor suppliers to gain market share.

For example, BYD, China’s EV national champion, recently developed EV electrification system illustrates the company’s ongoing efforts to reduce its reliance on foreign suppliers. As China accounts for over 50% of global EV sales, its increased sourcing of domestic chips will reshape the automotive semiconductor landscape, forcing non-Chinese suppliers to more aggressively pursue other markets to offset lost revenue.

Simultaneously, Chinese auto OEMs continue to gain domestic market share, impacting foreign semiconductor suppliers that supply Western auto OEMs in China. Currently, Chinese-produced chips account for only an estimated 15% of domestic demand, indicating that Chinese suppliers’ market share gains are still in their early stages. As China’s self-sufficiency push advances, market share erosion for Western firms will only intensify.

Stock Valuations at Risk

A competitive shift is underway in the analog, microcontroller, and power semiconductor markets, yet some stock valuations fail to reflect this changing landscape. While there are signs of improved cyclical dynamics for select companies, the prospect of a strong, broad-based recovery in the near term remains muted. Rising Chinese competition further clouds the long-term outlook, with the risk of market share losses and pricing pressure threatening longer-term earnings estimates, potentially leading to a derating of some stocks as terminal value multiples compress.

Texas Instruments stands out as particularly vulnerable, trading at 34x 2025 earnings, a premium to the broader market. With rising competitive pressures and potential pricing headwinds, this valuation is increasingly difficult to justify.

Analog Devices remains the best-in-class among this group, yet even ADI in the long term is not immune to the evolving landscape. Trading at 31x forward earnings, its valuation premium is also at risk as Chinese competitors gradually advance in analog markets. NXP Semiconductor remains at risk due to its above-average China exposure, making it more susceptible to China competition.

ON Semiconductor is less well positioned to withstand this competitive shift. However, ON and Microchip—as the most fundamentally depressed U.S. names—could see a more significant snapback in their stock prices when demand improves. However, sustained competitive pressure from China will challenge their longer-term fundamentals.

U.S. Policy Response and Strategy

China plays by a different set of rules, often disregarding traditional market-based principles. The shifting competitive landscape in China’s semiconductor market and the growing threat to U.S. and European suppliers is part of a larger challenge that must be addressed. The Trump administration must stay on course with policies that incentivize domestic manufacturing and enforce local content requirements to build a more secure and resilient supply chain across industries.

Just as China pursues policies that favor its domestic companies, the U.S. can respond in kind, not through heavy-handed intervention, but with a more market-based approach that fosters stronger economic growth and greater innovation. Beyond semiconductor policy, the U.S. and EU must pursue a broad economic and trade policy agenda that creates and enforces a policy regime to prevent China from flooding global markets with excess production at artificially low prices. Without firm trade policies to counteract industrial overcapacity and predatory pricing, China’s ability to undercut U.S. and European firms will continue to threaten U.S. industries.