On February 4, 2026, Microsoft integrated Claude — Anthropic's AI — directly into GitHub, alongside OpenAI's Codex. The same day, the Wall Street Journal reported that Copilot, the product Microsoft built on its $13 billion OpenAI investment, had lost a third of its market share in six months. The same day, sources told The Information that Amazon was discussing investing tens of billions in OpenAI — despite being Anthropic's largest backer. The same day, Anthropic announced that "Claude will remain ad-free." Four moves, one structural shift: the exclusive AI partnerships that defined the industry's corporate phase are dissolving.
The Failure
The WSJ headline was blunt: "Microsoft's Pivotal AI Product Is Running Into Big Problems." Copilot — the product that was supposed to justify Microsoft's $13 billion investment in OpenAI — saw its share of paid users' first choice fall from 18.8% to 11.5% in six months. Google's Gemini rose from 12.8% to 15.7% and overtook it.
This followed yesterday's Apptopia data showing ChatGPT itself losing 24 points of market share in a year. Microsoft bet on OpenAI exclusivity and built Copilot as the flagship product of that bet. The market responded by choosing alternatives.
The Hedge
Microsoft's response was not to double down. It was to hedge.
In January, The Information reported that Microsoft had become one of Anthropic's top clients, spending nearly $500 million per year on Anthropic's AI to power Microsoft products. Microsoft was simultaneously OpenAI's largest investor and one of Anthropic's biggest customers.
On February 4, that hedge became visible to every developer. Microsoft integrated Claude and Codex as AI coding agents directly into GitHub, GitHub Mobile, and Visual Studio Code — available to Copilot Pro Plus and Enterprise users. OpenAI's investor was now offering OpenAI's competitor's model inside the platform OpenAI helped build.
The $13 billion exclusive bet had produced a product losing market share. The $500 million hedge produced the integration. The logic was clear: if exclusivity doesn't buy market share, exclusivity isn't worth paying for.
The Mirror
Microsoft wasn't alone. Every major AI partnership was breaking in the same direction.
Amazon and OpenAI: Amazon has invested over $4 billion in Anthropic and hosts Claude on AWS Bedrock. On February 4, The Information reported that Amazon was discussing a deal with OpenAI — potentially investing tens of billions and getting OpenAI to dedicate researchers to building customized models for Amazon. Anthropic's largest backer was courting Anthropic's largest competitor.
Nvidia and OpenAI: Bloomberg reported that Nvidia was nearing a deal to invest $20 billion in OpenAI as part of its $100 billion round. Nvidia also participated in xAI's $20 billion Series E in January. The chipmaker was investing in both the company that just got acquired for $250 billion and the company suing it.
Talent: On the same day, Bloomberg reported that OpenAI had hired Dylan Scandinaro, who worked on AGI safety at Anthropic, as its new "head of preparedness" — at a base salary of up to $555,000. Even the people were becoming interchangeable between the labs.
The Counter
While the platform layer was dissolving its exclusive model partnerships, Anthropic made a different move. Its announcement, titled "Claude is a space to think," stated: "There are many good places for advertising. A conversation with Claude is not one of them."
This was published the day after OpenAI was asking advertisers for $200,000 minimums and the day after Altman's AGI claim appeared alongside ChatGPT's declining market share. The timing was not subtle. Neither was the message: when platforms treat your model as a commodity — when Microsoft puts Claude next to Codex, when Amazon courts your competitor — the way to differentiate is not better benchmarks. It's a better relationship with the user.
Anthropic's tender offer at a $350 billion-plus pre-money valuation suggested the market agreed. Ad-free was not a concession. It was the product.
What Ended
Between 2023 and 2025, AI's corporate structure was defined by exclusive allegiances. Microsoft backed OpenAI. Amazon backed Anthropic. Google had Gemini in-house. The assumption was that picking the right model early would create a durable competitive advantage — that AI would work like mobile operating systems, where early platform choices locked in developers and users for a decade.
February 4 showed that assumption was wrong. Microsoft's $13 billion didn't lock in developers — Copilot lost share. Amazon's $4 billion didn't lock in the market — it needed OpenAI too. Nvidia invested in everyone because it sold to everyone. The models were converging. The switching costs were low. The exclusivity was worth less every quarter.
The platforms learned that exclusivity doesn't buy capability. The models learned that capability doesn't buy loyalty. What's left is the question Anthropic answered: if every platform will offer every model, what makes a user choose yours?