OpenAI is looking for a replacement for Nvidia chips. What happened?
The ChatGPT developer is unhappy with some of Nvidia's latest chips and has been actively looking for alternative suppliers since last year.
The ChatGPT developer is unhappy with some of Nvidia's latest chips and has been actively looking for alternative suppliers since last year.
Tensions have arisen between the two main drivers of the artificial intelligence boom, OpenAI and Nvidia, over the Taiwanese company's new chips and amid protracted negotiations over a multibillion-dollar investment, Reuters reports , citing eight sources familiar with the matter.
OpenAI's shift in strategy is due to its focus on inference (the process of generating answers from an AI model) rather than just training. According to sources, OpenAI is unhappy with the speed at which Nvidia's hardware handles specific requests, such as writing code or AI interactions with other software.
Sam Altman's company is looking for new hardware that would cover about 10% of their computing needs for these narrow-profile tasks in the future.
Because of these problems, OpenAI has already signed agreements with Nvidia's competitor, AMD, and has also been in talks with startups Cerebras and Groq, which create chips for rapid inference.
The situation is complicated by the fact that the companies are in the process of closing a large-scale investment deal. Back in September, Nvidia announced its intention to invest up to $100 billion in OpenAI . The deal was expected to close in a few weeks, but negotiations have dragged on for months. One reason for the delay, the source said, is a change in OpenAI's technical requirements for the hardware.
Despite insider information about the search for alternatives, publicly the company leaders are trying to smooth over the differences. Jensen Huang called the rumors of tension “nonsense” and confirmed plans for huge investments. And Sam Altman wrote on the social network X after the Reuters publication that Nvidia makes “the best chips in the world” and expressed the hope of remaining their “giant customer for a very long time.”
Preply is ready to spend millions of dollars on AI infrastructure, currently has 6-7 engineering teams in this direction and large contracts with OpenAI. However, there are problems with hiring AI specialists
Preply is ready to spend millions of dollars on AI infrastructure, currently has 6-7 engineering teams in this direction and large contracts with OpenAI. However, there are problems with hiring AI specialists
Sam Altman said OpenAI plans to "slow down" hiring pace
Sam Altman said OpenAI plans to "slow down" hiring pace
“You don’t need a PhD in computer science to do this.” Nvidia CEO reveals how AI boom will make plumbers and electricians six figures
“You don’t need a PhD in computer science to do this.” Nvidia CEO reveals how AI boom will make plumbers and electricians six figures
Read the country's main IT news in our Telegram
Read the country's main IT news in our Telegram
