- Nvidia’s chips have been hot property in Silicon Valley this year thanks to the AI boom.
- But a shortage in the chips could ease as soon as next year, Sam Altman said.
- The OpenAI boss told the Financial Times that rivals to Nvidia would soon enter the market.
Sam Altman has reportedly predicted that a “brutal crunch” triggered by a shortage in expensive chips needed to power tools like ChatGPT will ease next year.
The OpenAI boss told The Financial Times that he expected the chip shortage to ease as companies enter the market with rival offerings to the hottest property in AI: Nvidia’s popular $40,000 H100 processors.
Businesses like OpenAI, which have been at the forefront of the generative AI boom this year, have been dependent on Nvidia for cutting-edge processors, known as GPUs, which offer the computing power needed to make apps like ChatGPT work efficiently.
OpenAI has had help securing the expensive GPUs with the financial backing of Microsoft, which made a multibillion-dollar bet on the ChatGPT maker earlier this year.
Nvidia’s success in selling processors to AI companies has been so great this year that the company has become a member of a select few in Silicon Valley with a market capitalization that tops $1 trillion.
The chip giant reportedly set a production target of 500,000 GPUs in 2023, but is seeking to triple production to at least 1.5 million units in 2024.
In the interview, Altman said, “The magic of capitalism is doing its thing,” and “a lot of people would like to be Nvidia now.” Companies looking to compete with Nvidia include heavyweights such as Google and Microsoft, as well as chip specialist AMD.
Though Nvidia has played a huge role in accelerating this year’s AI boom, CEO Jensen Huang has remained cautious.
At an event hosted by The Harvard Business Review last week, which Fortune reported, Huang said, “There are no companies that are assured survival.”
OpenAI did not immediately respond to Insider’s request for comment, made outside of normal working hours.
Read the full article here