KEY POINTS

  • The companies' discussions reportedly focused on servers that can run AI tech
  • The surge in demand for AI servers resulted in limited cloud capacity for customers
  • A global chip industry crunch has also affected the server shortage

Tech giants Microsoft and Oracle talked about possibly renting servers from each other amid a global shortage in computing power due to artificial intelligence use, a new report revealed.

The two tech companies discussed the possibility of sharing servers should any of them run out of cloud computing servers as more cloud customers use AI servers, a person with direct knowledge of the negotiations told The Information in a story published Tuesday.

The person added that Oracle chairman Larry Ellison and other senior executives at the computer software company consider a broader AI strategy, which includes utilizing AI software such as OpenAI's ChatGPT to improve Oracle's core software products.

A surging demand for servers that run AI software has resulted in a shortage that forced cloud-server providers such as Oracle and Microsoft to limit AI-running server availability for customers, The Information previously reported.

Some customers reported waiting months at times to rent hardware that can run machine-learning software, according to interviews with cloud providers and customers.

It appears that the surge in cloud server demand started as early as mid-2022.

In July 2022, a spokesperson for Microsoft told Data Center Dynamics that the company saw "unprecedented growth in the Cloud" globally. "With this surge, coupled with macro trends impacting the whole industry, we've taken steps to address customer increases in capacity while also expediting server deployment in our data centers," the spokesperson said.

The spokesperson's statement came after The Information reported that global Microsoft Azure data centers were operating with limited server capacity due to a global shortage.

Microsoft said at the time that the limited capacity of its Azure servers was temporary, as per the outlet. One Microsoft manager told the outlet that one Azure data center in central Washington and some in Europe and Asia were expected to retain limited status through 2023.

If Oracle and Microsoft come to an agreement regarding shared servers, it is expected that the two computing giants can meet surging demand for AI servers.

In February, reports emerged about Oracle's enterprise resource planning (ERP) suite NetSuite going offline for about a day. Complaints posted on Reddit revealed that the outage put businesses dependent on NetSuite's Boston-based servers to a standstill, as per The Register.

Microsoft has been working to fill the shortage gap through a new program called ND H100 v5, which is expected to improve speed for companies working with generative AI. The rent-a-server program was unveiled in mid-March after OpenAI's server was overwhelmed by surging demand for the large language model ChatGPT.

Aside from the introduction of the ND H100 v5, Microsoft is also reportedly spending billions to develop and produce its own microchips for AI servers.

Microsoft and Oracle aren't the only tech giants in the server sector that has been pressured by the capacity shortage.

"There is a huge supply shortage of Nvidia GPUs and networking equipment from Broadcom and Nvidia due to a massive spike in demand," Dylan Patel, Semianalysis analyst, told Fierce Electronics in March.

Microsoft Azure is using Nvidia's H100 Tensor Core GPUs for its server programs.

The global chip industry has been suffering from a supply crunch that has also affected the cloud computing sector. Patel estimated that Nvidia may need a year or more to catch up to the demand for H100 GPUs.

Nvidia is two to three months behind in fulfilling cloud server chip orders, Insider Intelligence reported in mid-April. While H100 chips may temporarily ease the global AI server shortage, observers argued that pricing issues may slow down the meteoric rise of the generative AI industry.

Furthermore, some industry experts also noted that Nvidia's inability to fulfill server chip demand may put rivals like AMD and Intel in the key players' game.

Last year, analysts raised concerns about the possible slowdown of the cloud market during the post-pandemic era, especially with slower growth in cloud revenues during Q2 2022. However, the rise of AI and generative artificial intelligence may shift the concern toward how server providers will manage the boom in AI server demand.

Illustration shows Microsoft Azure cloud service logo
Azure is Microsoft's cloud computing services arm. Reuters