Anthropic stands on the verge of the largest public offering in history. Bankers have pitched investors on a valuation north of $2 trillion. The numbers in its leaked IPO filing tell a story of breathtaking revenue acceleration. They also reveal eye-watering losses and fixed commitments that stretch into the hundreds of billions.

Founded just five years ago by former OpenAI executives including Dario Amodei, the company behind the Claude models has ridden the AI wave with remarkable speed. Revenue jumped twelvefold in 2025 to nearly $4.6 billion, according to documents reported by Reuters. First-quarter 2026 revenue alone reached $4.73 billion. Preliminary second-quarter figures topped $11.5 billion. By late July 2026 the annualized run rate exceeded $65 billion. Some reports now place it above $100 billion.

Yet the bottom line tells a different tale. Operating losses widened to $8.06 billion in 2025 from $2.98 billion the prior year. The headline net loss approached $42 billion, though roughly $34 billion of that stemmed from accounting charges tied to financing instruments rather than cash operating expenses. Compute and infrastructure spending hit $7.33 billion last year. Cash stood at $20.28 billion at year-end.

The real weight lies in future obligations. Leaked figures show roughly $518 billion in infrastructure commitments over the next decade, spread across six partners. About 80% appear noncancelable. Broadcom-related equipment leases account for $161.2 billion. Google comes in at $111.1 billion. Amazon adds $110 billion. These sums dwarf the revenue figures. They turn the business into something closer to heavy infrastructure than pure software.

Customer concentration adds another layer of risk. Two unnamed customers supplied nearly a quarter of 2025 revenue. Many large enterprise clients can reduce spending quickly without triggering corresponding cost cuts for Anthropic. Sales routed through Amazon and Google cloud marketplaces made up 47% of revenue. The company pays distribution fees on those transactions.

The Bull Case for $2 Trillion

Even at that headline valuation, some analysts see fair value. A Seeking Alpha analysis argues the price accounts for projected revenue scaling into the hundreds of billions by 2028. At the July 2026 run rate, $2 trillion equates to roughly 31 times sales. Compare that with 435 times 2025’s full-year revenue. The multiple compresses dramatically as growth compounds.

Anthropic projects $190 billion to $200 billion in revenue for 2028, sources told Reuters. Enterprise adoption of Claude tools for coding, analysis, and agentic workflows continues to expand. Business subscriptions have grown rapidly. Demand appears insatiable among large organizations seeking productivity gains. If the company can convert even a fraction of that demand into sustainable, high-margin revenue, the valuation begins to look less outrageous.

Investors have voted with capital. The company raised $65 billion in its Series H round in May 2026 at a $965 billion post-money valuation. That topped OpenAI at the time. Earlier rounds pushed the valuation from $61.5 billion in March 2025 to $183 billion in September 2025, then $380 billion in February 2026. Each step brought in sovereign wealth funds, big tech partners, and traditional asset managers. Nvidia has discussed a potential $10 billion anchor investment in the IPO itself.

But. The fixed costs create a different profile from typical software businesses. Revenue remains largely consumption-based. Customers can dial usage up or down. Commitments to chipmakers and cloud providers do not flex downward easily. Gross margins will face pressure until the company scales far beyond current levels.

PitchBook analysts, writing in a report summarized by Morningstar, caution that the $2 trillion price requires clear evidence of gross margin trajectory, payment schedules on those commitments, and a credible path to free cash flow. They see support for a valuation well above $1 trillion but note the leap to $2 trillion rests on optimistic assumptions about how much of the run-rate revenue turns into cash.

Recent market signals show continued momentum. As of early October 2026, discussions continue around timing the IPO after November midterm elections to include third-quarter results. The company has committed $100 million to train 10,000 engineers in deploying Claude inside organizations through its new Frontier Academy program. Partnerships with consulting giants like McKinsey, Deloitte, and Accenture aim to embed AI deeper into corporate operations.

Still the risks remain front and center. Regulatory scrutiny, potential government restrictions on AI supply chains, and competition from OpenAI, Google, and others could slow progress. The leaked prospectus reportedly highlights catastrophic risks from advanced AI systems, a theme Anthropic has emphasized since its founding.

So what happens when the music stops? If revenue growth sustains at these rates, Anthropic could justify its ambitious price tag within a few years. If usage plateaus or customers balk at rising costs, those $518 billion in commitments could weigh heavily on the balance sheet and stock performance.

The IPO will test public market appetite for frontier AI economics. It offers a rare window into the true unit economics of training and serving ever-larger models. Wall Street has cheered the growth story for years. Now it must decide whether the infrastructure bill makes the math work at a $2 trillion starting point.

One thing seems clear. The era of easy private capital for AI labs is ending. Public investors will soon own a piece of the most capital-intensive technology race in history. How they price the combination of explosive demand and crushing fixed costs will set the benchmark for the industry for years to come.