NEW YORK, October 25, 2026 — Anthropic has delayed its initial public offering, a move that exposes a widening chasm between artificial intelligence infrastructure costs and projected returns. The postponement occurs despite the company reporting third-quarter 2026 revenue of $16.8 billion, a 45% increase quarter on quarter.
The bottom line is stark: the market is pricing in growth that the underlying economics may not support. By the numbers, the gap between what investors are paying and what the technology currently generates is reaching historic proportions.
The $3.6 Trillion Deficit
Gartner data reveals the scale of the disconnect. The research firm forecasts at most $3.6 trillion in AI data center revenue through 2030. However, covering planned investment, financing costs, and a 12% return requires $8.5 trillion. That leaves a $3.6 trillion hole in the projected returns.
This deficit is not an isolated metric. It reflects a broader structural issue in the sector. Companies are spending billions on compute and infrastructure before seeing commensurate revenue. The figure that matters here is not just the revenue growth, but the cost of maintaining that growth.
Anthropic’s Cost Structure
Anthropic’s leaked prospectus provides a detailed look at these costs. In 2025, the company spent $7.33 billion on compute and infrastructure. This figure exceeded its total revenue of $4.6 billion for that year. The company reported a net loss of $42 billion in 2025.
Investors will note that $34 billion of that loss was a noncash accounting charge related to financing instruments. However, the cash burn is significant. Anthropic expects to spend at least $518 billion on AI infrastructure over the next decade. According to the confidential IPO prospectus reviewed by Reuters, about 80% of this total is noncancelable or payable regardless of usage.
The company is still targeting a $2 trillion valuation for its November listing. This represents a multiple of 435 times its 2025 revenue. A quarter of Anthropic’s 2025 revenue came from just two unnamed customers, a concentration of risk that complicates the growth narrative.
OpenAI and the Safety Pause
The delay is not unique to Anthropic. OpenAI has stated it will not go public in 2026. CEO Sam Altman called an IPO at this moment “ill-advised” given AI safety concerns. While this positioning may serve a strategic purpose, it signals caution among the sector’s leaders.
New York City Comptroller Mark Levine also urged Anthropic to pause its listing. Levine, who oversees pension funds holding roughly $7 million in Anthropic exposure, cited risks outlined in Anthropic’s own prospectus. The document warns of “existential risks to humanity” from increasingly autonomous AI systems. These concerns have added weight to the decision to delay.
Open-Weight Models and Debt
The economic pressure is intensifying due to open-weight AI models. Enterprise adoption of these models rose from 23% to 67% in 2026. Companies report cost savings of 70% to 90% by switching away from closed models.
DeepSeek R1 offers frontier-level performance at 7 cents per million input tokens. Independent analysis suggests Anthropic’s comparable pricing is roughly 18 to 23 times higher. This price gap undermines the premium that closed-model companies rely on to justify their valuations.
The debt burden is rising in tandem. Net debt plus future obligations for six major tech firms surged from $576.66 billion to $3.61 trillion in two years. Despite this leverage, only 13% of CIOs report significant value from their AI investments. The combination of high costs, low reported utility, and falling pricing power creates a fragile foundation for the current market enthusiasm.
Sources
- The AI IPO Wave Is The Dot-Com Bubble's Evil Twin. Anthropic’s Delay Just Proved It.
- The AI Bubble Has A $3.6 Trillion Problem, And It's Getting Worse (NYSEARCA:SPY)
Photo: TechCrunch (CC BY 2.0)
