Tuesday, Nov 10 | --:--
Back to home

Microsoft Is Letting Its Oracle AI Rental Lapse Because the Older Nvidia Chips No Longer Fit, Bloomberg Reports

Microsoft has been switching off AI computing capacity it rented from Oracle under a 2023 deal after declining to renew the contract, Bloomberg reported on November 6, citing people familiar with the matter; the hardware used an older generation of Nvidia chips that no longer met Microsoft's needs. Oracle told Reuters demand for its AI infrastructure, including earlier-generation GPUs, remains strong. Microsoft did not immediately comment, and neither side has disclosed the deal's size.

Times of AI Desk 3 min read Redmond / Austin View as Markdown
Two hollow industrial pipes rest disconnected on concrete saddles, with official Microsoft and Oracle logo paper clippings tucked onto the respective pipe and support edges.

How long rented AI capacity stays useful is one of the open questions behind the data-centre debt boom, and Microsoft has just given one answer: about three years, at least for this contract. Microsoft is winding down an arrangement to rent AI computing from Oracle, Bloomberg News reported on Friday, November 6, citing people familiar with the matter, as part of a wider reassessment of billions of dollars in cloud partnerships.

What was reported#

According to Bloomberg, as relayed by Reuters, Microsoft has been turning off the rented capacity in recent months after declining to renew the contract. The capacity became less valuable over time because the Oracle hardware used an older generation of Nvidia chips that no longer met Microsoft's needs, the report said. Oracle announced the multi-year deal in 2023, saying Microsoft would use Oracle Cloud Infrastructure to power the AI version of Bing; financial terms were never disclosed.

"Demand for OCI's AI infrastructure remains strong, including for earlier-generation GPUs," an Oracle spokesperson told Reuters. Microsoft did not immediately respond to Reuters.

The wider pattern#

Reuters tied the report to other signs that Microsoft is changing how it buys capacity. A day earlier, AI cloud company IREN said it had delayed some infrastructure deliveries after Microsoft asked for changes to its network design, pushing its target for bringing more than $4 billion of annualised revenue capacity online from year-end to January, possibly late February. Bloomberg also reported on Thursday that Microsoft was tightening oversight of data-centre projects run by neocloud providers including IREN, Nebius and Nscale amid delays and rising costs, according to Reuters.

For Oracle, the timing is awkward rather than decisive: the company priced a $20 billion stock and bond raise on November 2, money it has tied to its AI data-centre build, and Bing-era capacity was never the core of that pitch. Its on-record answer, that older GPUs still find buyers, is the claim to test.

Limits#

  • Every detail of the wind-down comes from Bloomberg's anonymous sources. Reuters and The Verge relay that report; neither is an independent confirmation, and we could not read Bloomberg's article.
  • The size of the Oracle contract, the amount of capacity being switched off and its end date are not disclosed.
  • Microsoft had not commented in the material we read.
  • Oracle's statement on demand is the company's own characterisation.

Sources#

Prior Coverage

Earlier Times of AI reporting on this thread.

Scroll to continue reading