{"slug": "in-game-theory-twist-openai-also-signs-open-source-ai-letter-leaving-anthropic", "title": "In Game Theory Twist, OpenAI Also Signs Open Source AI Letter Leaving Anthropic As Most Prominent Non-Signatory", "summary": "OpenAI signed the 'Open Weights and American AI Leadership' letter, joining NVIDIA, Microsoft, and over 20 other companies in endorsing open-weight AI models, leaving Anthropic as the most prominent US AI company not on the list. The letter argues that open-weight models strengthen US AI leadership and cybersecurity, though OpenAI's signing contradicts its commercial interest in proprietary models. Anthropic, now valued at $965 billion and leading in enterprise adoption, has the most to lose from open alternatives and thus declined to sign.", "body_md": "The entire AI sector in the US appears to have now come together and taken a stand that’s contrary to that of its most prominent company — Anthropic.\n\nHours after NVIDIA CEO Jensen Huang used his [first-ever post](https://officechai.com/ai/nvidia-ceo-jensen-huang-backs-open-models-in-first-post-on-x/) on X to share the “Open Weights and American AI Leadership” letter, OpenAI added its name to the list of signatories. Sam Altman quote-tweeted Huang directly: “i want the US to win in AI both in open source and proprietary models, and i am glad to see this.” Journalist Tae Kim, said that OpenAI had signed the letter. “OpenAI tells me they signed the Nvidia–Microsoft open-source/open-weights AI model letter today. What an amazing turn of events. Your move, Anthropic!”\n\nThe signatory list has since grown well past the original 20 names, adding OpenAI alongside Cisco, Cohere, DoorDash, Fireworks AI, GitHub, and Palo Alto Networks. That’s now nearly every major American AI lab and infrastructure company with one glaring exception. Anthropic, whose entire public identity has been built on caution around exactly the kind of unrestricted model access this letter defends, is now the most prominent US AI company not on the list.\n\n## Why Everyone Else Signed Anyway\n\nThe letter itself argues that open-weight models — ones anyone can download, modify, and run on their own hardware — strengthen American AI leadership, sharpen cybersecurity defenses, and keep the benefits of AI from concentrating in a handful of closed providers. On paper, that argument runs directly against the commercial interest of a company like OpenAI. It has spent billions of dollars training frontier models, prices access to them at a premium, and has actively [warned the US government](https://officechai.com/ai/us-threatens-sanctions-against-chinese-models-that-are-allegedly-distilling-us-models/) about foreign labs distilling its outputs into cheaper open alternatives. Meta, similarly, has spent the past year quietly [walking back its own open-source commitments](https://officechai.com/ai/meta-appears-to-change-open-source-policy-in-just-a-year-says-itll-now-be-careful-in-what-to-open-source/) after Llama 4 underperformed and Chinese open models took the lead in the category Meta itself had pioneered. Neither company’s actual product strategy over the last year has looked much like an endorsement of open weights as the future.\n\nSo why sign a letter that, if taken at face value, argues against your own monetization strategy? The answer sits less in the letter’s content and more in who currently doesn’t need it.\n\nAnthropic has spent much of 2026 pulling ahead. It’s now valued at [$965 billion](https://officechai.com/ai/anthropic-raises-65-billion-at-965-billion-valuation-is-now-worth-more-than-openai/), ahead of OpenAI on paper, and has [overtaken OpenAI in US enterprise adoption](https://officechai.com/ai/anthropic-has-passed-openai-in-us-business-adoption-for-the-first-time-says-ramp-data/) according to Ramp data, with its share of combined OpenAI-Anthropic business subscription spend climbing from roughly 10% in early 2025 to well past 60% by this year. Claude models have repeatedly traded the top spot on the Artificial Analysis Intelligence Index with GPT and Gemini, and on real-world agentic work specifically, Claude has been the one pulling ahead of both. Anthropic’s whole valuation case rests on staying the default choice for enterprises willing to pay a premium for the best available model. A world flooded with capable open alternatives is a direct threat to that premium, so it has every reason to sit this letter out and let the current closed, premium-priced arrangement continue exactly as it is.\n\nOpenAI, Meta, and everyone else signing this letter are in the opposite position, and this is where the game theory gets interesting. None of them currently hold the clear lead. Signing a letter that champions open weights doesn’t cost you as much if you’re not the market leader — but it costs the market leader a whole lot more. For OpenAI, Meta, Microsoft, and the rest, backing open models is a way of making the ground shift under the company currently ahead of them, even while none of these companies particularly want their own frontier models commoditized into open weights either. It’s a coalition of the second-place and lower, banding together against the effects of someone else’s lead, using an argument about American competitiveness that happens to also serve their own competitive interest.\n\n## The Prisoner’s Dilemma Underneath It\n\nThis is a fairly textbook example of a scenario where the dominant strategy for everyone except the leader is to cooperate on undermining the leader’s advantage, even at some cost to their own individual position. Every signatory that isn’t Anthropic has to weigh the same trade-off. Public support for open weights makes it slightly harder to justify big margins on their own frontier models down the line. But it’s a cost worth paying if it also slows down or narrows Anthropic’s current lead, because a world where models broadly commoditize is a world where nobody’s edge, including Anthropic’s, survives on pricing power alone.\n\nMicrosoft’s incentive here is especially direct, and predates this letter by months. Microsoft AI CEO Mustafa Suleyman has said [openly](https://officechai.com/ai/anthropic-is-extremely-expensive-many-are-urgently-looking-for-alternatives-microsoft-ai-ceo-mustafa-suleyman/) that Microsoft wants to reduce and eventually eliminate what it pays Anthropic, despite being one of Anthropic’s largest enterprise customers. CEO Satya Nadella has separately argued that [foundation models are getting commoditized](https://officechai.com/ai/foundational-ai-models-are-getting-commoditized-microsoft-ceo-satya-nadella/) as a category, describing even OpenAI, in which Microsoft owns a 27% stake, as “not a model company” but a product company. A letter that accelerates model commoditization is, for Microsoft, simply an extension of a position it had already staked out.\n\nFor OpenAI, the calculation is murkier but still coherent. It doesn’t currently have Anthropic’s enterprise lead, and Altman has previously admitted the company was on the “wrong side of history” for not open-sourcing more of its work, before eventually releasing its gpt-oss models. Signing this letter costs OpenAI little it hasn’t already conceded rhetorically, and it lets Altman frame the move as being about American competitiveness against China rather than about undercutting a specific rival. That framing matters, because it’s the same framing Anthropic itself has used when arguing for chip export controls and tighter frontier model oversight. Watching OpenAI now use an open-weights argument to make Anthropic look like the outlier on American AI leadership, using language Anthropic has itself relied on, is the sharper part of the twist Tae Kim was pointing to.\n\nNone of this means the letter’s substantive arguments about open weights, competition, and security are made in bad faith by any single signatory. But the shape of who signed and who didn’t tells you plenty about where each company currently sits in the race, and how little it costs a company in second place to argue for a more level playing field.", "url": "https://wpnews.pro/news/in-game-theory-twist-openai-also-signs-open-source-ai-letter-leaving-anthropic", "canonical_source": "https://officechai.com/ai/in-game-theory-twist-openai-also-signs-open-source-ai-letter-leaving-anthropic-as-most-prominent-non-signatory/", "published_at": "2026-07-25 06:21:20+00:00", "updated_at": "2026-07-25 06:28:13.277976+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-policy", "ai-startups", "ai-products", "ai-ethics"], "entities": ["OpenAI", "Anthropic", "NVIDIA", "Microsoft", "Sam Altman", "Jensen Huang", "Meta", "Cohere"], "alternates": {"html": "https://wpnews.pro/news/in-game-theory-twist-openai-also-signs-open-source-ai-letter-leaving-anthropic", "markdown": "https://wpnews.pro/news/in-game-theory-twist-openai-also-signs-open-source-ai-letter-leaving-anthropic.md", "text": "https://wpnews.pro/news/in-game-theory-twist-openai-also-signs-open-source-ai-letter-leaving-anthropic.txt", "jsonld": "https://wpnews.pro/news/in-game-theory-twist-openai-also-signs-open-source-ai-letter-leaving-anthropic.jsonld"}}