Report: OpenAI Taps Google’s Cloud Service for Extra Computing Capacity

OpenAI and Google reportedly finalized a deal in which OpenAI will tap Google’s cloud service for additional computing capacity to train and run its artificial intelligence (AI) models.

The companies had been discussing a deal for months but had been unable to sign it until an agreement that made Microsoft OpenAI’s exclusive data center infrastructure provider ended in January, Reuters reported Tuesday (June 10), citing unnamed sources.

OpenAI and Google finalized their deal in May, according to the report.

Reached by PYMNTS, Google declined to comment on the report.

OpenAI did not immediately reply to PYMNTS’ request for comment.

The deal marks a collaboration between competitors, as OpenAI and Google are both prominent in AI and in search, where OpenAI’s ChatGPT may challenge Google’s search business, according to the report.

It also signals a diversification of compute sources by OpenAI, which partnered earlier this year with SoftBank and Oracle on the Stargate infrastructure program and with CoreWeave for more compute, the report said.

Microsoft announced in a Jan. 21 blog post that it would no longer be OpenAI’s exclusive cloud provider for its AI models. Instead, it will have the right of first refusal to host OpenAI’s workloads in Azure.

The company had become OpenAI’s exclusive cloud provider in a 2019 agreement after investing $1 billion in AI startup. Since then, Microsoft reportedly has invested a total of nearly $14 billion in the AI startup.

For Google, the reported deal with OpenAI is the latest in a series of deals that have been aided by the company’s expanding availability of its in-house chips known as tensor processing units. In other recent deals, Google’s cloud unit has also gained the business of Apple, Anthropic and Safe Superintelligence, per the Reuters report.

AI data centers are needed because traditional data centers and power grids are struggling to accommodate the intense computational power, data storage and energy required by AI, PYMNTS reported in January.

“AI data centers are fundamentally different because they require specialized hardware and infrastructure to handle the massive parallel processing needed for AI workloads,” Deborah Perry Piscione, co-founder and CEO of Work3 Institute, an AI and Web3 advisory firm, told PYMNTS at the time.

The post Report: OpenAI Taps Google’s Cloud Service for Extra Computing Capacity appeared first on PYMNTS.com.