OpenAI and Anthropic are demonstrating explosive growth in revenue and valuation, but their future is increasingly tied to three nodes: access to computing power, data center financing, and regulatory favor. These are no longer just business risks, but fundamental constraints for the entire industry.
Computing as the Primary Deficit
OpenAI states outright: the company's revenue grows in sync with the available volume of computing. CFO Sarah Friar revealed that from 2023 to 2025, capacity grew from 0.2 GW to 1.9 GW, and annual recurring revenue — from $2 billion to over $20 billion. The ChatGPT developer calls computing the most scarce resource in AI and claims that expanding infrastructure would directly accelerate product monetization. Profit metrics remain undisclosed, hinting at the model's high capital intensity.
Anthropic is following the same path. In late May, the company raised $65 billion in a Series H round at a $965 billion valuation. According to their data, annual revenue based on the current sales run rate exceeded $47 billion. The funds will go towards expanding computing capacity for Claude.
The Price of Infrastructure: Trillions of Dollars and Uncertain Returns
Scaling AI models is becoming prohibitively expensive not only for developers but for the entire supply chain. Goldman Sachs Research analysts estimate that the four largest hyperscalers — Meta, Microsoft, Amazon, and Alphabet — will spend $5.3 trillion on capital expenditures from 2025 to 2030. While OpenAI and Anthropic are not included in this selection, they are the ones generating the demand.
The Bank for International Settlements warned in its 2026 annual report: the five largest cloud providers will spend over $1 trillion on AI infrastructure in 2025–2026. These commitments already outpace the companies' profits and free cash flow, forcing them to take on debt financing. If monetization expectations are not met, a sharp reduction in funding could collapse the entire investment cycle.
Regulatory Risk: When Government Becomes Part of the Infrastructure
Beyond the cost of computing, AI giants face sudden restrictions from authorities. In June, Anthropic was forced to halt access to its Fable 5 and Mythos 5 models due to a directive from the U.S. government under export controls. The company noted that the government's letter lacked specifics on safety and linked the complaints to a possible circumvention of safeguards.
In the same month, the Trump administration asked OpenAI not to release GPT-5.6 to the general public immediately. The company will first provide the model to a limited number of clients. The Department of Commerce later lifted the restrictions, but the unease remained: regulators could freeze a flagship product at any moment.
Anthropic's relationship with authorities is even more strained. In February 2026, the U.S. Army used Claude in an operation to capture the President of Venezuela, and the head of the Pentagon called the developer a "supply chain risk." CEO Dario Amodei stated that the company would prefer not to cooperate with the Pentagon rather than agree to the use of technologies that could undermine democratic values.
Going Public: A Race for Valuation Amidst Uncertainty
Despite all the risks, OpenAI and Anthropic are exploring public market listings. The former filed a confidential IPO application on June 8, the latter a week earlier. The terms of the offerings are not defined, but for investors, the key factor will be not so much the valuation as the economics of scaling: how much it costs to sustain user growth, how quickly data centers pay off, and how stable the financing channels are.
My analysis: The infrastructure dependency of OpenAI and Anthropic is a classic scaling trap. They are forced to increase capacity to justify valuations, but each new investment increases the system's fragility. If regulators or the market cut off their oxygen, the fall could be swift. Investors should look not at revenue, but at cash burn and the payback periods of data centers.