Building massive infrastructure without proportional revenue growth
In the unfolding contest over artificial intelligence's future, two companies have arrived at a philosophical fork in the road: one believes discipline and trust will outlast ambition, the other that scale itself is the strategy. Anthropic is building toward profitability by 2028 through focused corporate relationships and efficient infrastructure, while OpenAI is committing to losses that could reach $74 billion annually by 2030, wagering that dominance requires consuming capital at a pace few industries have ever witnessed. The divergence is not merely financial — it is a disagreement about what the market will ultimately value: reliability and alignment, or reach and raw capability.
- OpenAI's projected $74 billion in annual losses by 2030 — against $13 billion in revenue — signals a bet so large it borders on existential, with $1.4 trillion in infrastructure commitments stretching across the next eight years.
- Anthropic is quietly outpacing OpenAI in the corporate market, projecting $3.8 billion in API revenues for 2025 — nearly double OpenAI's estimated $1.8 billion — suggesting enterprise clients are already choosing safety and cost predictability over scale.
- The tension between the two models is sharpening: Anthropic's burn rate is projected to fall to just 9 percent of revenue by 2027, while OpenAI's remains flat, raising urgent questions about whether aggressive infrastructure spending can be sustained without proportional growth.
- Both companies are courting the same pool of major investors and cloud partners — Google, Amazon, Microsoft — in what amounts to a parallel race for capital legitimacy ahead of what could be among the largest technology IPOs in history.
- The outcome hinges on a single unresolved question: whether the AI market will reward the company that moves fastest and broadest, or the one that earns the deepest trust from the clients who matter most.
Two companies are staking the future of artificial intelligence on opposite convictions. Anthropic, the maker of Claude, is building toward profitability by 2028 through disciplined spending and a deliberate focus on corporate clients. OpenAI, creator of ChatGPT, is projecting annual losses of $74 billion by 2030 even as it generates $13 billion in revenue — a scale of capital consumption that reflects not recklessness, but a fundamentally different theory of how markets are won.
The numbers are stark. Anthropic expects its cash burn to fall from roughly 70 percent of revenue this year to just 9 percent by 2027. OpenAI's burn rate stays flat across the same period, and its leadership has committed to infrastructure deals — chips, data centers, computing capacity — that could obligate the company to spend up to $1.4 trillion over eight years. Sam Altman's vision is to establish OpenAI as a multi-trillion-dollar technology giant that sets the pace for the entire sector.
Anthropics strategy runs in the opposite direction. The company has avoided computationally expensive ventures like image and video generation, built hybrid models that scale processing intensity to task complexity, and diversified its hardware suppliers across Google, Amazon, and Nvidia to reduce costs and vendor dependency. Corporate clients now represent roughly 80 percent of its revenue, and its API projections for 2025 — $3.8 billion, nearly double OpenAI's estimated $1.8 billion — suggest those clients are already choosing alignment and predictability over raw capability.
The investment community is watching both companies position for what could be historic public offerings. Google is in talks to value Anthropic above $350 billion. Microsoft anchors OpenAI's cloud infrastructure. Nearly every major Silicon Valley investor holds a stake in one or the other. Yet the deeper question remains open: whether either model is truly sustainable, and which vision of the market will prove correct — the one that prizes efficiency and trust, or the one that bets everything on being first, largest, and everywhere at once.
Two companies are betting the future of artificial intelligence on opposite visions of how to win. Anthropic, the maker of Claude, is building toward profitability by 2028 with disciplined spending and a focus on corporate clients. OpenAI, which created ChatGPT, is burning through capital at a scale that would make most companies flinch—projecting annual losses of $74 billion by 2030, even as it generates $13 billion in revenue that same year. The divergence is not a matter of different tactics within a shared strategy. These are fundamentally different bets about what the AI market will reward.
Internal financial documents reviewed by The Wall Street Journal reveal the starkness of the choice. Anthropic expects to consume roughly $3 billion this year against $4.2 billion in revenue—a burn rate of about 70 percent. By 2027, the company projects that figure will drop to just 9 percent of revenue. OpenAI, by contrast, will spend $9 billion against $13 billion in sales this year, and the company's leadership, led by Sam Altman, has committed to consuming roughly 14 times more cash than Anthropic before reaching profitability in 2030. Those commitments are staggering in scale: Altman has disclosed that OpenAI's recent deals with chip and computing providers could obligate the company to spend up to $1.4 trillion over the next eight years.
The two companies are making opposite bets about what matters most. Anthropic is concentrating its efforts on corporate customers, who now represent roughly 80 percent of its revenue. The company has deliberately avoided the most expensive ventures—image and video generation, which demand enormous computational power—and instead built hybrid AI models that adjust their processing intensity based on the task at hand. A simple question gets a quick answer; a complex one gets deeper reasoning. This approach gives customers more control over costs and, in theory, sets a new standard for efficient AI deployment. Anthropic has also diversified its hardware suppliers, incorporating chips from Google and Amazon Web Services alongside Nvidia's expensive processors, reducing its exposure to any single vendor and lowering its overall infrastructure costs.
OpenAI is pursuing a different logic entirely. The company is reserving roughly $100 billion in data center capacity as backup infrastructure, betting that demand for AI will continue to exceed supply and that being the company with the most available computing power will prove decisive. Sam Altman's strategy is to establish OpenAI as a multi-trillion-dollar technology giant that sets the pace for the entire sector. This requires constant capital infusion and carries real risk: if the market loses interest or if revenue growth cannot keep pace with infrastructure spending, the company faces what some observers have called capital incineration. OpenAI is expanding aggressively into consumer products—ChatGPT itself, the video generation tool Sora, the Atlas platform—and is developing new hardware devices, e-commerce features, advertising capabilities, and even humanoid robots.
The financial projections reveal how different these paths truly are. In 2026, Anthropic expects its cash burn to represent about a third of its revenue, while OpenAI's will be 57 percent. By 2027, Anthropic's burn rate is projected to fall to 9 percent; OpenAI's stays flat. Anthropic's gross margins are expected to exceed OpenAI's through 2028, driven by revenue growth paired with controlled infrastructure spending. Meanwhile, Anthropic projects API revenues of $3.8 billion in 2025—nearly double OpenAI's estimated $1.8 billion—suggesting that corporate customers are already voting with their wallets for the company's emphasis on safety, alignment, and predictable costs.
The investment community is watching closely. Google is in talks to invest in Anthropic at a valuation exceeding $350 billion. Microsoft is OpenAI's primary cloud provider, while Amazon and Google serve Anthropic. Nearly every major Silicon Valley investor has a stake in one company or the other, and both are positioning themselves for what could be among the largest technology IPOs in history. Yet the question hanging over the sector is whether either model is actually sustainable. The AI industry faces a fundamental challenge: maintaining the pace of economic scaling. OpenAI could post annual losses of $8.5 billion in some scenarios, a figure that underscores the risk of building massive infrastructure without proportional revenue growth. Anthropic, for all its efficiency, still depends on continued corporate adoption and on the assumption that its premium pricing for aligned, safe models will hold.
What happens next will likely depend on which vision of the market proves correct. If corporate customers continue to prioritize safety, alignment, and predictable costs, Anthropic's disciplined approach could establish a durable competitive advantage. If the market instead rewards the company that can move fastest and offer the broadest range of capabilities—regardless of cost—then OpenAI's aggressive expansion may prove prescient. For now, both companies are executing their strategies with conviction, and the outcome will reshape not just the AI industry but the broader technology landscape.
Notable Quotes
The demand for artificial intelligence exceeds the available computing capacity today. Every dollar we invest in AI infrastructure goes toward serving hundreds of millions of consumers, companies, and developers who depend on ChatGPT to be more productive.— OpenAI spokesperson to The Wall Street Journal
The Hearth Conversation Another angle on the story
Why does Anthropic think it can reach profitability so much faster than OpenAI? Is it just being more conservative?
It's not conservatism exactly. Anthropic is making a deliberate choice about which customers to serve. Corporate clients want predictable costs and safety guarantees. That's a narrower market than OpenAI is chasing, but it's also one where you can charge premium prices and maintain margins.
And OpenAI is chasing the broader market—consumers, developers, everyone?
Yes, but at a cost. To serve hundreds of millions of consumers, you need massive infrastructure. OpenAI is betting that if you build enough capacity, demand will fill it. Anthropic is betting that you don't need to be everywhere to be profitable.
The $1.4 trillion in commitments OpenAI has made—that's not just spending, it's a lock-in. They can't walk back from that.
Exactly. It's a bet that the demand will materialize and that they'll be the company positioned to capture it. But if demand softens or if customers choose a more efficient alternative, those commitments become a liability.
So Anthropic's multi-vendor hardware strategy isn't just about cost savings. It's about optionality.
Right. If Nvidia prices spike or if there's a supply disruption, Anthropic can shift to Google or AWS chips. OpenAI is more dependent on Nvidia and on the assumption that the supply chain will hold.
Which approach is actually riskier?
They're risky in different ways. Anthropic risks being too small, too focused, missing the market shift toward consumer products. OpenAI risks burning capital faster than it can generate revenue. The winner will likely be whoever correctly predicted what the market actually wants.