Microsoft Bets Big on Anthropic as It Loosens Its Reliance on OpenAI
Microsoft has redrawn the frontier AI map again, this time by striking major new partnerships with Nvidia and Anthropic. Announced on Tuesday, the deals mark one of the clearest signals yet that Microsoft is preparing for a more plural, less OpenAI-centric future in artificial intelligence.
Under the agreements, Microsoft will invest up to $5 billion into Anthropic, while Nvidia will contribute up to $10 billion. Those commitments have propelled Anthropic’s valuation to roughly $350 billion, according to a person familiar with the deal. That is almost double the USD 183 billion valuation the company held in September and places the Claude developer among the most highly valued private AI companies in history.
Anthropic locks in enormous compute contracts
Alongside the investments, Anthropic has agreed to buy $30 billion of Azure compute capacity and has contracted for additional capacity up to one gigawatt. It will also purchase up to one gigawatt of compute built on Nvidia’s Grace Blackwell and Vera Rubin systems, binding the startup tightly to both companies’ infrastructure over the coming years.
The scale is staggering. One gigawatt of compute capacity places Anthropic in the same league as the largest AI labs globally, and the volume of Azure spend gives Microsoft a long-term customer whose training needs are still expanding at breakneck speed.
For Nvidia, the deal represents a new era of cooperation. “This is a dream come true for us,” Jensen Huang said in a video on Tuesday. The company noted that this is the first time it has entered into a deep co-development partnership with Anthropic. Nvidia engineers will work alongside Anthropic to optimise the Claude family of models for both performance and efficiency, while Anthropic will shape aspects of future Nvidia architectures to suit its workloads.
A strategic shift for Microsoft
The move lands amid growing industry scrutiny of Microsoft’s reliance on OpenAI. The companies’ partnership remains tight. Microsoft holds a stake valued at roughly $135 billion, equal to about 27 per cent of OpenAI’s for-profit business on a diluted basis. But OpenAI’s internal turbulence over the past two years, plus its explosive growth as a commercial enterprise, has pushed Microsoft to broaden its alliances.
Anthropic is a logical partner. Founded in 2021 by former OpenAI executives, including CEO Dario Amodei, it has positioned itself as both a technical leader and a governance-oriented counterweight in frontier AI. Its Claude models have become known for their reasoning performance and their attempts at more predictable behaviour. Amazon, an early investor, made Amazon Web Services Anthropic’s primary cloud provider in 2023 and its lead training partner in 2024. Anthropic confirmed Tuesday that AWS will continue in both roles even as it deepens ties with Azure.
Microsoft now finds itself financing two direct rivals. Far from shying away from the contradiction, chief executive Satya Nadella framed it as a strength. “As an industry, we really need to move beyond any type of zero-sum narrative or winner-take-all hype,” he said. The future, he argued, depends on broad, durable capabilities built across companies rather than a single centralised AI provider.
A widening competitive field
Anthropic’s new valuation underscores a dramatically shifting competitive landscape. Five years ago, OpenAI was the clear centre of gravity in frontier model research. Today, the map is more crowded, with companies like Anthropic, Google DeepMind and a growing roster of well-funded independents all vying to define the next scale frontier.
Anthropic’s expanded compute access could alter the pace of that competition. Training runs at the gigawatt scale open the door to significantly larger models, deeper multimodal architectures and more computationally intensive safety techniques. They also cement Anthropic as a core customer for both Azure and Nvidia, offering revenue stability that few other AI labs can match.
Nvidia, for its part, is set to gain yet another anchor partner ahead of its quarterly earnings report on Wednesday. Its strategy of embedding itself deeply in the development pipelines of the most ambitious AI labs has paid off handsomely in recent years, with demand for Blackwell-class systems outstripping supply.
The big picture
Taken together, the agreements reflect a maturing of the AI sector. Microsoft is hedging against concentration risk by backing multiple frontier labs. Nvidia is reinforcing its position as the indispensable supplier of training infrastructure. Anthropic is securing the capital and compute needed to challenge OpenAI head-on.
The real story is less about rivalry and more about scale. As frontier model development grows ever more expensive, partnerships of this magnitude are becoming prerequisites rather than luxuries. No single company can fund, build and operate the next generation of AI systems alone.
Microsoft’s strategy now hinges on that reality. Its future in AI will be shaped not just by its stake in OpenAI, but by an increasingly broad network of alliances designed to ensure that whichever lab builds the next breakthrough, it will be running on Microsoft’s cloud, accelerated by Nvidia’s silicon and seeded by investments from Redmond.
