Microsoft's Maia chip finally lands a major customer? Anthropic's entry could reshape the cloud computing competitive landscape

```

Anthropic is currently in talks with Microsoft to rent servers powered by Microsoft’s self-developed Maia AI chips. If successful, this move would mark a significant breakthrough in Microsoft’s in-house chip strategy and further reshape the competitive landscape of the cloud computing market.

According to the latest report from The Information, insiders revealed that Anthropic is seeking to rent Maia chip servers to obtain more computing power in order to meet the growing demand for its Claude model. If the negotiations succeed, Microsoft would win a major external customer, giving a strong boost to its efforts to catch up with Google and Amazon in the field of self-developed chips.

For the market, this development means that the cloud computing partnership between Microsoft and Anthropic is deepening, and also reflects the industry-wide trend among leading AI companies to accelerate reducing their reliance on Nvidia chips.

Microsoft Maia Chip: The Rescuer’s Path

The strategic logic behind Microsoft’s development of the Maia chip is the same as Google’s TPU and Amazon’s Trainium—by creating alternatives to Nvidia hardware, they can reduce hardware costs of cloud services and protect or enhance profit margins. Cloud computing executives generally believe that, with high Nvidia hardware costs, developing in-house chips has become a necessary means to maintain profitability.

However, Microsoft’s journey in in-house chips has not been smooth. The Maia chip project experienced delays last year and remains in an earlier stage compared to Google and Amazon. In January this year, Microsoft announced that the latest version, Maia 200, had been deployed in Azure data centers. Azure chief Scott Guthrie stated that Maia 200 has been used to reduce the operational costs for the Copilot AI tool, which is powered by OpenAI and Anthropic’s models.

From a positioning perspective, the Maia chip is designed to run existing models faster than Nvidia chips, but is not intended for training new models—making it more suitable for large-scale inference deployment, rather than cutting-edge model development.

Anthropic’s Multi-Chip Strategy: Not Putting All Eggs in One Basket

In terms of chip procurement, Anthropic has long deliberately differentiated itself from competitors. Unlike OpenAI and xAI, which heavily rely on Nvidia hardware, Anthropic has built a diversified computing infrastructure including chips from Amazon, Google, and Nvidia, and is reportedly negotiating with a UK chip startup for cooperation.

The negotiations with Microsoft over chip use are a natural extension of this multi-chip strategy. If a deal is reached, Anthropic would gain another option for running its Claude model, and could have the opportunity to participate in influencing the design direction of future generations of the Maia chip, making it better suited to their needs.

From a financial perspective, Anthropic has sufficient motivation to expand its cloud spending. According to The Information, Anthropic’s revenue this quarter is expected to approach $1.1 billion, more than doubling from the previous quarter, with projected quarterly profits of around $560 million. Meanwhile, the company has already committed at least $33 billion in cloud spend with the three major US cloud providers. Analysts point out that if cloud vendors are willing to subsidize their custom chips, using custom chips could theoretically offer Anthropic a better cost-performance ratio.

Microsoft and Anthropic’s Growing Partnership

Over the past year, Microsoft’s relationship with Anthropic has grown notably closer, coinciding with OpenAI increasing cooperation with Microsoft’s competitors.

On the commercial side, Microsoft is currently one of Anthropic’s most important customers and is expected to purchase at least $500 million worth of Claude model access to power Copilot products. Earlier this year, Microsoft began reselling Anthropic’s models on the Azure cloud platform alongside OpenAI’s models. At the end of last year, Microsoft announced up to a $5 billion investment in Anthropic, after which Anthropic committed to spending $3 billion on Azure cloud computing.

In terms of compute resource allocation, a direct source said that since last November, Anthropic’s usage on Azure has continued to rise. Recently, Microsoft not only allocated more of its existing Nvidia server resources to Anthropic, it also began building dedicated new server clusters for the company, reducing resources available to some mid- and small-scale clients to a certain extent.

These chip-level negotiations are seen as a signal of further expansion of the companies’ current cloud agreements. In comparison, the scale of Microsoft’s cloud agreement with OpenAI is still much larger than with Anthropic.

Multi-Chip Strategies Becoming the New Consensus

Anthropic’s multi-chip strategy is inspiring industry imitation. Reports say that Meta and OpenAI are both trying to copy this approach, and have struck deals with multiple chip suppliers. At the same time, both are independently developing their own chips. Microsoft has set aside some data center space for OpenAI’s upcoming in-house chips, but the two sides have yet to reach a formal procurement agreement.

These developments indicate that competition in the AI computing power market is shifting from “Nvidia’s dominance” to a more diversified supply chain. For cloud providers, the ability to sell self-developed chips to external customers will be a key measure of their chip strategy’s success; for AI developers, building diversified sources of computing power is both a practical decision for cost reduction and efficiency and a strategic move to mitigate supply chain risks.

Risk Disclosure and DisclaimerThe market carries risks, and investment requires caution. This article does not constitute personal investment advice, nor does it take into account the individual investment goals, financial situation, or needs of any particular user. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment made on this basis is at their own risk. ```