One in Four Businesses Now Pay for Anthropic. A Year Ago It Was One in 25.

OpenAI plans to nearly double its workforce from roughly 4,500 to 8,000 employees by the end of 2026, according to Financial Times, which first reported the plans on March 21 citing two people familiar with the matter. The company did not respond to requests for comment.

The hire-up is concentrated where the company needs it most: product development, engineering, research, sales, and what the company is internally calling "technical ambassadorship" — a category that reads less like a job title and more like a diagnosis. Enterprise customer success, dressed up in startup language.

Read the headcount plan alongside the other OpenAI news from the past two weeks and a clearer picture emerges: this is not a growth story. It is a defense.

The ground has shifted

Ramp, which processes payments for tens of thousands of businesses, publishes a monthly AI adoption index that has become one of the most reliable real-time signals in enterprise AI. Its March 2026 update, written by Ramp economist Ara Kharazian and published March 11, contained a number that should be alarming to anyone at OpenAI.

Anthropic's adoption grew 4.9% month over month in February — its largest single-month gain since Ramp started tracking. Nearly one in four businesses on Ramp now pays for Anthropic. A year ago, it was one in 25.

OpenAI's adoption fell 1.5% in the same month. Kharazian described it as "the largest decline in any single month for any AI model company since we started tracking business AI adoption."

The timing matters. February was the month the OpenAI-Department of Defense contract became public knowledge and drew significant backlash from researchers and enterprise customers alike. Ramp's data suggests some of that backlash translated into actual spend decisions. Anthropic, which had previously withdrawn from DoD work, was the direct beneficiary.

The split goes deeper than brand sentiment. Among companies buying AI tools for the first time — the cohort that determines long-term market structure — Anthropic now wins approximately 70% of head-to-head matchups against OpenAI. That is a reversal of the trends Ramp observed throughout 2025, when OpenAI's consumer momentum was still converting into enterprise wins.

Kharazian's explanation for Anthropic's moat is worth quoting directly: "Maybe it's culture. Maybe it's that Anthropic has become — for lack of a better word — cool." He stops short of calling it decisive but notes that the choice between OpenAI and Anthropic may be "becoming less like an enterprise procurement decision and more like the green bubble/blue bubble distinction in iMessage." Whether or not that framing holds, the data behind it is real.

The financial architecture

OpenAI's enterprise revenue is approximately $10 billion annualized, according to Reuters, which reported last week that the company is in advanced talks to form a joint venture with private equity firms TPG, Bain Capital, Advent International, and Brookfield Asset Management. The pre-money valuation of the proposed venture: $10 billion.

The structure is revealing. PE firms would commit about $4 billion and receive equity stakes, board seats, and — most importantly — influence over how OpenAI's technology is deployed across their portfolio companies. This is distribution-as-investment: OpenAI gets a faster route into corporate adoption; the PE firms get an early position in an AI vendor before IPO and a hedge against AI disruption hitting their existing holdings.

OpenAI's total revenue is reportedly around $25 billion annualized. Enterprise at $10 billion means roughly 40% of the business. Anthropic, which has a significantly smaller total revenue base, appears to be winning a larger share of new enterprise spending. That asymmetry explains why a company that just raised at an $840 billion valuation is simultaneously assembling a PE consortium to distribute its products.

The $110 billion funding round closed February 27. This hiring surge — announced roughly three weeks later — is the first major allocation of that capital. The money makes the headcount possible; the Ramp data explains why they think they need it.

What doubling headcount actually buys

There is an honest counterargument here. Hiring 3,500 people does not fix a brand perception problem. OpenAI's gap with Anthropic in first-time enterprise buyer matchups is not primarily a sales coverage problem — it's a trust problem, and Anthropic's advantage is partly supply-constrained compute, partly safety positioning, partly the kind of intangible preference that no amount of "technical ambassadors" can manufacture.

The headcount could still help in areas where it's clearly additive: deeper engineering capacity for enterprise integrations, more research bandwidth for model improvement, broader sales coverage in international markets where the DoD association carries less weight. But if the Ramp data is right that Anthropic wins because it's culturally preferred among early adopters — the engineers and "AI guys" who influence what their companies buy — then headcount is a slow instrument for a fast problem.

OpenAI remains the AI company used by the most businesses overall. It is not in crisis. But the February numbers are a warning sign that the consumer-to-enterprise conversion that was its structural advantage is starting to reverse, and the response so far has been to spend very large amounts of money very quickly. Whether that is the right move is a question the company's IPO will eventually have to answer.