OpenAI's Identity Crisis, Datacenter Wars, Market Up on Iran News, Mamdani's First Tax, Swalwell Out

OpenAI and Anthropic are locked in a fierce rivalry that is reshaping the landscape of artificial intelligence, with their divergent growth trajectories, business models, and funding strategies setting the stage for a consequential industry “flippening.” A deep dive into their revenue growth, market focus, valuations, and infrastructural advantages provides insight into who may emerge as the market leader.

Anthropic's 10x Revenue Growth Outpaces Openai's 3-4x, Hinting At Market Leadership Despite Lower Revenue

While both OpenAI and Anthropic reported roughly $30 billion in revenue at the beginning of Q2, the underlying growth rates tell a different story. Anthropic has posted an extraordinary 10x annual revenue growth, skyrocketing from $1 billion to $10 billion last year and reaching $30 billion by Q1, with projections of $80–100 billion by year’s end if current trends continue. In contrast, OpenAI’s annualized growth hovers at 3-4x, requiring two years to decuple revenue, compared to Anthropic’s single year. This logarithmic growth rate, unprecedented even outside of tech, puts Anthropic in a position to rapidly overtake its rival.

Anthropic's Enterprise Coding and Metered Token Pricing Offers Scalable Revenue, Unlike Openai's Consumer Subscription Model With Lower Per-user Revenue

Anthropic’s explosive growth is tied directly to a clear strategic focus on enterprise use cases, particularly coding. Their metered token pricing, akin to an electricity model, scales as businesses increase their usage, resulting in robust, scalable revenue streams. In comparison, OpenAI prioritized the consumer segment, where willingness to pay is lower—only about 3–4% of consumers are inclined to buy premium, typically at a flat $20/month “all you can eat” model, which caps per-user revenue and hinders overall scalability.

Openai's $850b Valuation Overvalued Vs Anthropic; Secondary Market Indicates Higher Investor Confidence in Anthropic

Investors are calling OpenAI’s $850 billion valuation into question, noting that secondary markets have recently priced Anthropic higher for the first time. Analysts argue that, for OpenAI's last funding round to make sense, the company now needs to IPO at $1.2 trillion—yet there are no buyers at the current $850 billion level. This reversal of investor confidence signals a shift towards Anthropic as the likely leader.

Revenue Composition: Anthropic's $30b Is Organic, Openai's $30b Includes ai Model Revenue Sharing

Anthropic’s $30 billion run rate is primarily organic, driven by direct product use. OpenAI’s reported $30 billion, however, is “cap inflated” by $8 billion due to revenue sharing and accounting with AI model providers and channel partners. On an apples-to-apples basis, Anthropic’s adjusted revenue is within 20% of OpenAI’s but represents real, usage-based enterprise income, not just platform licensing or shared model revenue.

Openai Criticized For Lacking Focus, Pursuing Consumer and Enterprise Markets Over High-Value Coding

OpenAI faces rising internal and investor frustration over its lack of clear strategic focus. The company has been criticized for spreading efforts thinly across both consumer and enterprise markets instead of zeroing in on high-value coding applications, where the biggest growth and margins reside.

Investors Urge Openai to Prioritize Chatgpt Growth Over Enterprise and Code Applications

Anonymous OpenAI investors voiced frustration, asking why the company isn’t fully capitalizing on its billion-user ChatGPT business, which still grows 50–100% yearly, instead of diluting focus with enterprise and coding initiatives. There is a call for OpenAI to devote resources to cementing ChatGPT’s leadership, rather than splitting attention.

Internal Memo From Openai's Chief Revenue Officer Critiques Anthropic, Outlines Pivot Toward Business Customers, and Acknowledges Past Misdirection

A leaked memo from OpenAI’s Chief Revenue Officer acknowledged Anthropic’s rapid rise and critiqued their business model but also revealed OpenAI’s strategic pivot. The memo admits OpenAI previously lacked focus, but now plans to pursue business customers and win the agent platform layer, marking a shift to deeper enterprise engagement.

Openai's Bid For Openclaw's Peter Steinberg Suggests an Effort to Stifle Competition and Direct Future Innovations Into Openai Products

OpenAI’s hire of Peter Steinberger, the architect of the OpenClaw open-source project, is seen as a bid to capture talent and ensure future breakthroughs are integrated within OpenAI’s product ecosystem, potentially stifling external competition and open-source advancements.

Frontier Ai Labs' Race Hinges on Profitable Growth and Maintaining Compute Advantage Via Infrastructure Control

The next phase of the AI race is defined by two things: who can achieve profitable, sustainable ...

The explosive demand for artificial intelligence (AI) compute power is colliding with growing infrastructure and regulatory constraints, heightened public opposition, and strategic moves by leading labs and hyperscalers. America is witnessing a mounting crisis around AI data centers, with regulatory headwinds and social backlash threatening both short-term progress and long-term competitiveness.

Data Center Projects Face 40% Cancellation Amid Political and Regulatory Challenges, Public Opposition Threatens Ai Infrastructure

Chamath Palihapitiya reports that out of every hundred data center projects contested, about 40 are canceled—a cancellation rate that is more than double last year's. The economic value at stake in these projects approaches $162 billion. This escalation is being fueled by shifting public sentiment: Americans are growing more negative toward AI, due to fears including job loss, wealth concentration, and the belief that technological progress benefits only the elite. David Friedberg asserts that data centers now represent a visible symbol of this perceived wealth and elite-focused progress, standing as “the temple of the wealthy” and a mechanism for tech billionaires to get ahead while the general population feels left behind. There is little broader consumer benefit felt from AI, with most people only seeing marginal improvements, such as medical advice from chatbots.

Public Sentiment Increasingly Views Data Centers As Symbols of Wealth Concentration and Elite-Focused Technological Progress

Many communities now view data centers as representing the tech elite’s interests, not public good. The backlash is particularly intense in Democratic cities, with local NIMBY (Not In My Backyard) groups blocking projects through regulatory capture and elections. Palihapitiya cites examples like a city board approving a $6 billion data center, only for half of its members to be ousted in order to reverse the decision. Entire states such as Maine have even passed laws banning new data center construction. Relocating these projects to states like Texas is not a simple fix, due to the lack of grid capacity and supporting infrastructure.

Democratic Cities and Nimby Groups Block Data Centers Through Denial and Regulatory Capture

Local and state opposition has become a formidable barrier, often fueled by coordinated campaigns. David Sacks notes that about 30 states may end up banning data centers outright as residents fear that energy-hungry projects will increase power costs without offering local benefits.

Tech Billionaire-Funded Advocacy Groups Use Climate and Water Arguments to Oppose Competitor Data Center Construction and Slow Ai Development

Regulatory opposition is amplified by billionaire-funded advocacy groups that use environmental and resource arguments to galvanize support against new sites. Climate and water impacts are cited frequently, sometimes with limited factual basis. Sacks and Palihapitiya discuss how "doomer NIMBYism" is sometimes astroturfed, with well-funded tech advocacy groups using these arguments to slow the competition's AI progress while serving their own strategic interests.

Frontier Labs Face a Choice: Accept Hyperscaler Dependency or Invest Billions in Infrastructure, With Competitive Implications

Major AI labs such as OpenAI and Anthropic confront a pivotal decision: remain dependent on hyperscalers (Amazon, GCP, Azure) for compute, or invest billions to build and control their own infrastructure. Travis Kalanick refers to the reliance on hyperscalers as a troubling dependency. For labs now operating at scale, Palihapitiya calls the lack of proprietary compute a “five alarm fire,” as inability to secure direct access to infrastructure could throttle growth and revenue, not just product quality. These labs must secure land, power, and construction, which is “turning out to be impossible” due to regulatory and Nimby obstacles.

Anthropic and Openai Must Shift From Renting Compute To Controlling Their Own Data Centers to Avoid Revenue Bottlenecks

David Sacks notes that for years, Anthropic strategically supported anti-data center sentiment to slow competitors, relying instead on renting hyperscaler compute. Now, growth has pushed them to the limits of third-party capacity, forcing Anthropic to seek proprietary infrastructure. This shift means past opposition strategies may backfire as labs compete for scarce approvals and power.

Anthropic's Decision to Hold Back Mythos Likely Reflected Compute Constraints, Reserving Resources For Opus 4.7 Launch and Marketing Impact Through Restraint

Anthropic’s release strategy offers a prime example: Sacks and Palihapitiya suggest the lab delayed commercial launch of its Mythos model largely due to insufficient compute, instead concentrating resources on Opus 4.7. This generated scarcity-driven marketing buzz and let government buyers believe the holdback resulted from responsible restraint, but actual resource constraints played a substantial, perhaps decisive, role.

Building Proprietary Infrastructure Offers a Competitive Edge With Priority Compute Access but Demands Substantial Capital and Faces Nimby Opposition and Regulatory Uncertainty

Building owned data center infrastructure offers competitive advantage—priority access to compute in an ever-tightening market. However, the capital outlay is immense and the pathway fraught with obstacles: capital requirements, slow permitting, ongoing Nimby resistance, and shifting regulatory landscapes. Palihapitiya warns that since hyperscalers still control 60% of total compute today, labs without proprietary access risk stalling at pivotal moments.

Alternative Energy Partnerships Address Data Center Constraints and Community Grid Concerns

In response, new models are emerging. Crusoe and CoreWeave are pioneering energy-independent data centers, bringing their own power—be it natural gas, diesel, or solar—on site. This “bring your own energy” (BYOE) model circumvents years-long grid interconnection waits and lessens community concerns about grid stress.

Crusoe and Core Weave Pioneer Energy-Independent Data Centers

These companies show the feasibility of rapid infrastructure expansion outside of legacy grid dependencies. By installing their own onsite power sources, they accelerate start times and provide buffer capacity for the surrounding region.

Bloom Energy & Peers See Rapid Valuation Growth With Onsite Energy Solutions, Cutting Permitting Timelines From Years to M ...

The current stock market exhibits extreme valuations, largely driven by the explosive narrative around artificial intelligence (AI), yet significant questions remain about whether these highs are sustainable given unresolved questions about AI’s return on investment (ROI) and hard physical growth constraints.

Equity Valuations Hit Extremes; Debate Over Ai Productivity Justifying High Prices Persists

David Sacks draws comparisons to the late 1990s dot-com bubble, when companies could inflate their valuations simply by associating with new technology, often without solid business fundamentals. He observes that, much like that era, today’s market awards “crazy valuations” to AI and technology firms, regardless of real gross margins or capital demands. The incremental cost for true software is near zero, but physical-world companies are also being priced as if they share the same economics, ignoring crucial structural differences.

Chamath Palihapitiya points out that traditional indicators of value, such as the Shiller Price/Earnings (PE) ratio and the Buffett index (the total value of all US equities relative to GDP), are both peaking at or near all-time highs. However, this does not reflect broad market strength: only a handful—about eight or nine—mega-cap companies are driving the S&P 500 toward all-time highs, creating a scenario of acute “dispersion” where the wider market stagnates. This highly skewed structure makes the current rally fragile.

S&p 500 Nears All-time Highs; 8-9 Mega-Cap Companies Drive Gains, Broader Market Stagnates, Creating Fragile Valuation Structure

Despite the S&P 500 nearing all-time highs and showing strong performance on the surface, Palihapitiya underscores that the index’s gains are mostly the product of a select few giants. Most other companies are not seeing comparable growth, which complicates the market picture and raises the risk that the current valuation structure is fragile and unsustainable.

Stock Market Trends Under Trump Link to Policy and War Resolution Prospects, Suggesting Investors Focus On Geopolitical Tensions Like Iran Instead of Ai Productivity Gains

Travis Kalanick argues that much of the market’s movement is dictated by policy responses and geopolitical signals, particularly during the Trump era. He describes Trump as using the stock market as a “weathervane,” maneuvering policy to buoy the S&P 500, particularly in the aftermath of events like war-related selloffs. Investors are now hyper-focused on presidential statements about conflict resolution, such as those concerning Iran, rather than on the direct productivity gains from AI. As Trump moves between stoking market anxiety and then reassuring investors through de-escalation and practical measures, seasoned traders have adapted to these cycles of volatility and recovery.

Warren Buffett's $300 Billion Cash Position Indicates Value Investors' Caution About Market Sustainability Despite Retail Enthusiasm For Ai Exposure

Chamath highlights Warren Buffett’s near $300 billion cash position as evidence of value investors’ skepticism. While retail investors pile into AI-exposed equities, Buffett’s strategy signals caution; he appears to be waiting for a correction, guided by indicators showing that equities are overvalued relative to the broader economy. The management shift at Berkshire Hathaway following Buffett’s reduced day-to-day influence is noted, but the underlying posture remains a significant indicator of traditional value investing’s read on the market’s sustainability.

Frontier Ai Models Encounter Growth Limits Due to Compute Scalability and Infrastructure Constraints

David Sacks cautions that, despite the breathtaking revenue growth from leading AI firms like Anthropic, such exponential scaling cannot continue indefinitely. As AI companies reach new scales, physical limitations such as compute power, electricity, semiconductors, land, and data center real estate create significant 10x growth barriers. Sacks notes that Anthropic is already running into these constraints.

10x Growth Limits: Physical Constraints Like Electricity, Data Centers, Semiconductors, and Land

AI companies dependent on ever-expanding compute face stark realities: increasing model capability requires vast new resources, but securing enough electricity, semiconductor supply, and land for data centers is becoming an ever-greater challenge. This imposes hard ceilings on both speed and scale of development.

Claude's "Thinking Less" Signals Ai Firms Hitting Infrastructure Limits Before Opus 4.7

There is growing evidence of these limits in user experience. Sacks points to complaints that Anthropic’s Claude model began “thinking le ...

Ai Productivity Claims and Enterprise Value Diverge as Firms Struggle to Translate Ai Investments Into Bottom-Line Improvements

Large enterprises are struggling to realize profitable outcomes from their AI investments, as highlighted by David Sacks and Travis Kalanick. Sacks references a major McKinsey study, concluding that many enterprise AI transformation projects are failing, primarily because of change management challenges and organizational inertia. Kalanick elaborates that, especially in big companies, resistance comes from entrenched layers of middle managers, technocrats, bureaucrats, and complex, often undocumented, processes. He emphasizes that change management is inherently a human, not strictly technological, struggle and proves particularly tricky in companies where critical procedures haven’t been fully mapped.

In contrast, small businesses and startups are demonstrating clear returns on investment from AI in niche domains. Jason Calacanis provides examples from his portfolio, such as Micro One, which leverages AI to enable dark pool data collection for large language model companies, and TaxGPT, an AI tool now used by six to seven percent of accountants, driving significant productivity in tax services. In these focused applications, AI is able to deliver market growth and operational efficiency.

However, Chamath Palihapitiya cautions that the success of startups and small players at the edges does not prove AI’s transformative power at enterprise scale. He contends that unless AI demonstrates effectiveness in high-leverage, complex enterprise use cases, its perceived value will remain largely theoretical. Meanwhile, enterprise customers are pushing back: they reject inadequate AI-generated content, demand greater accountability for spending, and CTOs report exhausted budgets while requiring strict justification for future AI expenditures.

Tech Company and Founder-Led Productivity Gains Show Success but Remain Exceptions, Not Indicators of Broad Ai Transformation

Founder-led and tech-native firms like Meta and Uber serve as notable exceptions to the sluggish enterprise AI adoption curve. Calacanis observes that these companies are experiencing accelerated feature rollouts post-AI integration, attributing this speed to a culture that embraces, rather than resists, AI-native change.

Yet, Kalanick and Palihapitiya stress that technology is not the main constraint—enterprise culture is the dominant bottleneck. Resistance from middle managers and process-bound bureaucrats slows or even stalls AI deployments. Further compounding this, Kalanick points out, are gaps in legacy system documentation—many historical processes are only informally known, making AI deployment particularly difficult when explicit instructions and logic are absent.

Ai Agents Excel In Automating Repetitive Tasks but Struggle With Novel Problem-Solving and Independent Decision-Making Without Human Guidance

AI agents today are effective at automating clearly defined, repetitive tasks, but they fall short when confronted with novel problems or when required to make decisions independently. Kalanick underscores the lack of sophistication in current agents, noting “they’re not that smart yet” and often make basic logical errors, such as taking conflicting positions on the same asset without explicit instruction. Both he and Calacanis agree that human oversight remains essential; AI agents require humans in the loop to provide direction, validation, an ...