OpenAI Rents Google AI Chips for ChatGPT Boost

OpenAI has recently entered into an agreement to rent Google's artificial intelligence (AI) chips to enhance its computing power for ChatGPT and other products, as reported by Reuters. This collaboration marks a notable shift in OpenAI’s strategy, as it diversifies its AI infrastructure and moves away from its heavy reliance on Nvidia GPUs.
A Surprising Collaboration Between Two AI Giants
OpenAI, a leading player in the AI industry, is one of Nvidia’s largest customers, using Nvidia’s graphics processing units (GPUs) for training its AI models and inference computing, where AI models make predictions or decisions based on new data.
However, OpenAI has recently expanded its computing capacity by incorporating Google Cloud services into its infrastructure, creating a surprising partnership between OpenAI and Google, two key competitors in the rapidly growing AI space.
According to Reuters, this move is part of Google's broader strategy to increase the availability of its proprietary Tensor Processing Units (TPUs), which were previously used exclusively within the company.
Google’s decision to make these chips available externally has attracted notable clients, including major players like Apple, as well as AI startups such as Anthropic and Safe Superintelligence, companies that were founded by former OpenAI leaders.
Also read: AMD Unveils MI400 AI Chips, Teams Up with OpenAI
Shift to Non-Nvidia Chips for Inference Computing
For OpenAI, leasing Google's TPUs is a vast change from its long dependence on Nvidia's GPUs.
According to reports, OpenAI is particularly interested in using Google’s TPUs to help lower the cost of inference, a critical process in AI development. The use of Google's TPUs could offer a more cost-effective solution compared to Nvidia’s GPUs, potentially making the technology more accessible for OpenAI’s growing computing needs.
However, it's important to note that Google is not providing its most powerful TPUs to OpenAI, according to a Google Cloud employee who spoke to The Information. This decision may reflect Google’s strategic positioning as both a competitor and a partner in the AI race.