In January, Sam Altman told employees OpenAI would "dramatically slow hiring." In March, he told the Financial Times he plans to nearly double the company's headcount from 4,500 to 8,000 by December. The reversal — roughly 12 new hires per day for the rest of the year — tells you everything about the velocity of the AI arms race and the particular terror of watching your competitor eat your lunch while you're still setting the table. [1]

The competitor is Anthropic. While OpenAI spent late 2025 issuing what it called a company-wide "code red" over ChatGPT's consumer growth, Anthropic's Claude model was quietly threading itself into enterprise operations at scale. Wall Street banks, consulting firms, and government agencies that once defaulted to OpenAI began kicking the tires on Claude, and many stayed. Anthropic's enterprise revenue, by some estimates, has grown fivefold since mid-2025. [1]

OpenAI's expansion targets the gap. The company is leasing additional office space in San Francisco and hiring aggressively across product development, engineering, research, and sales. It is also building what it calls a "technical ambassadorship" program — essentially consultants embedded inside enterprise clients to help them integrate AI tools into daily operations. The model borrows from Palantir's playbook, where forward-deployed engineers became the sticky layer that made switching costs prohibitive. [1]

The strategic partnerships are equally telling. OpenAI is courting private equity firms, pitching them on introducing ChatGPT's enterprise products to their portfolio companies — a distribution channel that bypasses the slow grind of individual corporate sales cycles. If a PE firm with forty portfolio companies says yes, that is forty enterprise accounts acquired through a single relationship. [1]

Both companies now have IPOs on the horizon. OpenAI completed a $40 billion funding round in late 2025 at a $300 billion valuation. Anthropic, backed heavily by Amazon and Google, is reportedly exploring a public offering by 2027. The headcount war is partly about building the revenue machinery that public market investors demand. You cannot IPO on research papers. You IPO on recurring enterprise contracts with logos that make analysts nod. [1]

The irony has not been lost on observers. OpenAI is the company that popularized the idea that artificial intelligence would automate a significant share of human work. It is now hiring at a pace that suggests the AI revolution still requires an enormous amount of human labor — at least for the companies building it.

-- DAVID CHEN, Beijing