OpenAI mulls setting up data centers, reducing reliance on Microsoft – report

Last week, OpenAI told some employees that the company would play a bigger role in setting up data centers and AI chips to develop technology rather than depending solely on Microsoft (NASDAQ:MSFT), The Information reported citing a person who heard the comments.

CFO Sarah Friar of the Microsoft-backed AI startup — which last week raised $6.6B, valuing the company at $157B — had previously told some stockholders that Microsoft did not move swiftly to supply OpenAI with adequate computing power, thus the requirement to pursue other data center deals, which it had begun to do, the report added.

OpenAI had recently inked a data center deal in Texas with a competitor of Microsoft. The companies had previously included a provision in their contract which allowed OpenAI to ink such deals, despite a deal which made Microsoft its exclusive cloud server provider, the report noted citing a different person familiar with the agreement.

Lately, CEO Sam Altman has been seeking to make the companies’ computing collboration more flexible. Altman was was worried about Microsoft’s inability to provide servers to OpenAI fast enough to stay ahead of competitors such as Elon Musk’s xAI, the report added citing four people working on data center projects for OpenAI. Musk had co-founded OpenAI with Altman and others.

As OpenAI goes ahead with a plan to convert to a for-profit from a nonprofit, it is discussing financial details with Microsoft, such as the size of the stake the tech giant would have. Currently, in return for Microsoft’s over $13B investment in OpenAI, Microsoft has the rights to 75% of OpenAI’s future profits until Microsoft’s principal investment is repaid, and 49% of profits after that up to a theoretical cap, the report noted.

Due to Microsoft’s outsize rights to future OpenAI profits under the current nonprofit set up, the company could land a significant part of OpenAI’s shares after the startup turns into a for-profit.

The nonprofit currently overseeing OpenAI would also receive a stake in the new for-profit company, which could be atleast 25%, the report stated, citing a person who had talked to company leaders.

Meanwhile, Microsoft had some time ago invested $650M to hire a team from AI start up Inflection in an effort to develop software which could rival those from OpenAI.

Microsoft also became more cautious about paying for ever-bigger server clusters for OpenAI as the company aims to make sure that it does not incur loss on costly data centers which may not bring in consistent revenue in the coming decades.

OpenAI is currently in discussions with Oracle to lease an entire Abilene data center site, which could increase to 2 gigawatts if Oracle is able to access more power at the site, the report noted, citing a person involved in the deal.

The site is on the path to draw just under 1 gigawatt of power by mid-2026, meaning it could house several hundred thousand Nvidia AI chips. It is not known what would Microsoft’s role be in the proposed expansion in Abilene.

OpenAI and Microsoft have previously discussed how they could work together on larger plants, including a $100B supercomputing server cluster, which Microsoft code-named Mercury and OpenAI leaders referred to as Stargate.

Last week, Friar said her team was thinking of how to use debt financing to pay for such a cluster, implying that it could happen with or without Microsoft’s hardware, similar to the Oracle agreement, the report noted.

Altman had told colleagues last week that the company developing its own AI chip could reduce the cost of operating such a cluster. The status of the chip was not known, however, OpenAI has discussed it with Broadcom (AVGO), which develops AI chips with several companies, including Google.

OpenAI currently rents most of its AI chips from Microsoft in the Phoenix area, as per the report.