OpenAI's $1.1B Statsig Acquisition: Deal, Team & Aftermath

What Statsig actually does

Statsig, founded in 2021 by Vijaye Raji, builds infrastructure for feature flagging, A/B testing, and product experimentation: the tooling that lets an engineering team ship a change to a slice of users, measure what happened, and decide whether to roll it out further. It raised a $10.4 million Series A in 2021 and a $43 million Series B in 2022, both led by Sequoia Capital, before closing a $100 million Series C in May 2025. [5]

That Series C, led by ICONIQ Growth with Sequoia and Madrona Venture Group returning, valued Statsig at $1.1 billion. Four months later, OpenAI agreed to buy the whole company for the same figure: $1.1 billion. A funding round and an acquisition landing on an identical number four months apart proves nothing by itself, but it means OpenAI paid what the last private round said Statsig was worth, not a premium for the company and not a discount for a founder eager to exit. [5][1]

Atlas interpretation: Statsig's customers are software teams that ship product changes constantly and want data on whether each change actually helped, which is the same problem OpenAI has at a much larger scale: rolling a new model, a new chat feature, or a pricing change out to hundreds of millions of users without knowing in advance what it will do to engagement or revenue. Feature flagging and experimentation infrastructure is unglamorous next to model research, and it is exactly the kind of plumbing a company built by researchers is not naturally staffed to build well. [5]

The deal and the new role

OpenAI announced the acquisition on September 2, 2025, structured entirely in stock rather than cash, sized against OpenAI's own valuation, which TechCrunch reported at roughly $300 billion at signing. Statsig was to keep operating out of its Seattle office and keep serving its existing customers, OpenAI said, and Statsig's own announcement promised the same to its customer base directly, rather than the company being folded into ChatGPT's engineering organization on day one. [1][2][4]

The acquisition's real function was visible in the personnel announcement bundled with it. Raji, Statsig's founder and chief executive, became OpenAI's first CTO of Applications, a newly created title. He arrived at Statsig after more than a decade at Meta, where he led Facebook Gaming, Marketplace, and Entertainment and ran Facebook's Seattle office, growing it past 5,000 employees. The new role put him in charge of product engineering for ChatGPT and Codex, plus OpenAI's infrastructure and Integrity teams, reporting to Fidji Simo, the former Instacart chief executive OpenAI had installed as CEO of Applications four months earlier. [3][1]

A title built to sit next to research, not inside it

Atlas interpretation: The "Applications" half of Raji's title is doing real work. It places him under Simo's product and business organization rather than under the researchers who train OpenAI's models, a division OpenAI had been building out since Simo's own hire in May 2025. Raji's job is shipping and scaling the things people actually use, ChatGPT, Codex, and the rest of the application layer, not deciding what the next model can do. Whether that boundary holds cleanly in a company still organized loosely around a handful of very senior people is a separate question, but the title was drawn to make the split explicit. [2]

The acquisition arrived bundled with a reorganization elsewhere in Applications. Kevin Weil, until then chief product officer, moved sideways to lead a new group called OpenAI for Science. Srinivas Narayanan, previously head of engineering, became CTO of B2B Applications, working with chief operating officer Brad Lightcap on enterprise accounts. Raji's hire, in other words, did not just add a title to the org chart; it cleared the product-engineering seat Weil had held by moving Weil somewhere else. [2]

Part of a pattern

Atlas interpretation: CNBC described the Statsig purchase as one of the largest acquisitions OpenAI had made, a list that already included its purchase of io Products, Jony Ive's hardware startup, three months earlier at a considerably larger headline price. Both deals were paid entirely in OpenAI stock. Both were built around acquiring a specific person and the team loyal to that person, Ive and his design studio, Raji and Statsig's engineers, more than around acquiring a product OpenAI planned to run unchanged. Both principals received newly minted, individually tailored executive titles rather than being folded into an existing team. [1]

Atlas interpretation: The price tags make more sense against the backdrop of 2025's hiring market than against Statsig's revenue. Meta spent that same year offering, by Sam Altman's own account, signing packages reaching into nine figures to try to pull researchers out of OpenAI; Altman said the offers mostly failed but did not dispute that Meta was making them. Buying a company outright, in stock that costs nothing to issue until it is issued, was one way for OpenAI to secure a specific executive without entering that bidding war on Meta's terms, while making the number read as an acquisition rather than a compensation figure. [6]

How it aged

By May 2026, the arrangement had split in a way the original announcement gave no hint of. Analytics company Amplitude took over Statsig's brand, its customer base, and the ongoing development of its platform. Per Amplitude's own announcement, the founding Statsig team kept working at OpenAI during the handoff. OpenAI had bought a company; eight months later, the product and the customers it served belonged to someone else, and only the people stayed where the acquisition had put them. [7]

Atlas interpretation: That outcome is close to a clean test of what kind of deal this actually was. If OpenAI had wanted Statsig's experimentation platform as a going product business, it would not have handed the platform, brand, and customer relationships to a competitor eight months in. It kept Raji and the engineers who report to him and let Amplitude take the software and the accounts. The CTO of Applications title, and the reasoning behind it, held up better than the acquired company did. [7]

The broader reorganization the deal was folded into did not hold up especially well either. Kevin Weil, who had moved from chief product officer to lead OpenAI for Science when Raji's hire cleared his old seat, left OpenAI in April 2026 as that group was dissolved and folded back into other research teams. Srinivas Narayanan, who had become CTO of B2B Applications in the same reshuffle, departed the same week, citing time with family. Two of the three executives whose roles were rearranged to make room for Raji were gone within eight months of the announcement that rearranged them. [8]