OpenAI cuts Stargate pledge to $600B

1/ OpenAI’s Stargate story now has three different numbers attached to it, and that’s the first thing to get straight. In January 2025, OpenAI and SoftBank announced Stargate as a new company that “intends to invest $500 billion over the next four years” in U.S. AI infrastructure, with $100 billion to begin immediately. (openai.com) 2/ The newer figure is about OpenAI’s own compute spending outlook. CNBC reported on February 20, 2026 that OpenAI told investors it was targeting roughly $600 billion in total compute spend by 2030, down from an earlier $1.4 trillion figure that Sam Altman had touted in infrastructure commitments. (cnbc.com) 3/ So the clean way to read this is: - $500 billion = the original publicly announced Stargate project framework in January 2025. - $1.4 trillion = the larger ambition Altman discussed around infrastructure commitments. - $600 billion by 2030 = the more recent investor-facing compute target. (openai.com) 4/ The second important change is structural, not just numerical. OpenAI’s own April 29, 2026 infrastructure post describes Stargate as a long-term effort built “with partners,” and says “no single company can build the infrastructure for the Intelligence Age alone.” It repeatedly emphasizes partner-led buildout, financing flexibility and optionality. (openai.com) 5/ That matters because the original pitch carried a more self-directed buildout image: Stargate would build new AI infrastructure for OpenAI in the United States, with SoftBank, OpenAI, Oracle and MGX as initial equity funders, and Oracle, Nvidia and OpenAI collaborating to build and operate the system. (openai.com) 6/ By spring 2026, the language had changed. OpenAI said the “financing models and partnership structures may evolve,” and framed the goal less as owning everything itself and more as getting capacity online “at scale, on time,” while preserving flexibility as technology changes. (openai.com) 7/ In plain terms, that is the shift people are talking about when they say OpenAI is renting compute rather than building all of it itself. The company is still pursuing enormous capacity, but the emphasis has moved toward securing access through partners across cloud, chips, energy, construction and finance. (openai.com) 8/ OpenAI’s April update also gives a concrete operating datapoint: it said that when Stargate was announced in January 2025, it had committed to securing 10 gigawatts of U.S. AI infrastructure by 2029, and that “just over a year later” it had already surpassed that milestone, with more than 3GW added in the prior 90 days. (openai.com) 9/ That suggests the company wants investors and customers focused on capacity secured, not on whether every megawatt sits inside a wholly owned OpenAI balance-sheet build. That is an inference from the company’s language and timing, but it fits both the April post and the February investor reset. (openai.com) 10/ The financial backdrop helps explain the reset. CNBC reported that OpenAI generated $13.1 billion in revenue in 2025, burned through $8 billion, and was projecting more than $280 billion in revenue by 2030. The outlet said the lower spending plan was meant to tie more directly to expected revenue growth after concerns that earlier expansion ambitions were too large for the revenue likely to follow. (cnbc.com) 11/ That makes this less a story about OpenAI abandoning infrastructure than about matching ambition to financing reality. Even at $600 billion, the target remains one of the largest compute spending plans in the industry. CNBC also reported OpenAI was finalizing a funding round of more than $100 billion, with strategic investors including SoftBank and Amazon, while Nvidia was discussing an investment of up to $30 billion. (cnbc.com) 12/ The cloud angle is the real competitive consequence. OpenAI’s January 2025 Stargate announcement already said the company would continue increasing its consumption of Microsoft Azure, even as Stargate added capacity. By April 2026, OpenAI was explicitly describing a partner-centric model spanning cloud providers and other infrastructure players. (openai.com) 13/ That means hyperscalers and infrastructure partners do not just finance the AI boom; they increasingly intermediate it. If frontier model labs need faster deployment, lower execution risk and financing flexibility, then cloud providers, chipmakers, data-center operators, utilities and capital partners become more central to who captures value. That framing is supported by OpenAI’s partner-centric description; the value-chain conclusion is an inference. (openai.com) 14/ It also means “build vs. rent” is probably the wrong binary. OpenAI’s own language points to a hybrid model: secure massive capacity, bring more compute online quickly, and use evolving financing and partnership structures to do it. The company says its strategy “has not changed” on the need for more compute, even as the structures around that buildout evolve. (openai.com) 15/ The bottom factual takeaway: OpenAI has not walked away from Stargate. It has moved from a looser, headline-grabbing infrastructure ambition toward a more defined 2030 spending target and a more explicit reliance on partners to supply and finance the compute it needs. (openai.com)