(AppSoftwareID) – OpenAI has started leasing Google’s AI processors to support ChatGPT and their other offerings, according to an insider who spoke with AppSoftwareId on Friday.
The company behind ChatGPT is among the biggest buyers of Nvidia’s graphical processing units (GPUs). These AI-specific chips are utilized for both training their models and performing inferential computations, where an AI system applies its learned expertise to generate predictions or decisions from fresh data inputs.
Earlier this month, appsoftwareidexclusively reported that OpenAI intended to incorporate Google Cloud services to address its escalating requirements for computational power. This development signifies an unexpected partnership between two leading rivals in the artificial intelligence industry.
As part of its expansion strategy, Google is now offering broader access to its internally developed tensor processing units (TPUs) after these were previously limited to internal usage only. This move has enabled Google to attract clients such as major technology company Apple along with emerging firms like Anthropic and Safe Superintelligence—both rivals to ChatGPT that were founded by ex-leaders from OpenAI.
OpenAI’s decision to lease Google’s Tensor Processing Units (TPUs) marks their initial significant use of non-Nvidia chips. This indicates a strategic pivot for the organization under CEO Sam Altman, moving away from dependence on Microsoft’s datacenters. The Information reports this change might position TPUs as a more cost-effective option compared to Nvidia’s graphics processing units (GPUs).
According to the report, OpenAI expects that the TPUs, which they lease via Google Cloud, will assist in reducing the expense of inference.
Nevertheless, according to The Information quoting a Google Cloud staff member, Google—a competitor of OpenAI in the artificial intelligence sector—is refraining from leasing its top-tier TPUs to its adversary.
Google refused to comment, whereas OpenAI has not yet responded when reached for clarification.
Google's decision to include OpenAI as part of its client base highlights how the tech behemoth has leveraged its internally developed AI technologies—from infrastructure to applications—to spur the expansion of its cloud services division.
(Juby Babu reported from Mexico City and Kenrick Cai from San Francisco; Editing by Alan Barona)