OpenAI CEO Sam Altman for a few years has been agitating for more global compute capacity to meet the massive demand created by generative AI and large-language models (LLMs), such as his company’s GPT-4.
He already works closely with Nvidia, which builds the GPUs that AI systems rely on, reportedly has evaluated buying a chip company, and has spoken with members of Congress about ways to increase the global supply of advanced AI chips.
Altman also reportedly has laid out ambitious plans for OpenAI to not only design its own chips but also to build a series of factories to produce them. It would be a departure from what other companies like Google, Amazon and Microsoft are doing, which is designing their own AI chips but letting someone else build them, a much less expensive proposition than building fabs, which can cost billions of dollars for one. Reports surfaced late last year that the CEO was talking to such companies as SoftBank and G42, an AI development company based in Abu Dhabi, about raising money to fund the fabs.
Now the industry is getting an idea of the eye-watering amount of money Altman is looking for to expand the global supply of AI chips and other infrastructure. This month, The Wall Street Journal reported that Altman is meeting with a number of investors – including the government of the United Arab Emirates – and telling them the initiative will need $5 trillion to 7 trillion.
Increasing the ability for the chip industry to supply the much-needed AI processors and accelerate the development and adoption of AI technologies is a key issue for Altman. In a post on X (formerly Twitter) last week, the CEO wrote that OpenAI believes “the world needs more ai infrastructure – fab capacity, energy, datacenters, etc. – than people are currently planning to build. Building massive-scale ai infrastructure and a resilient supply chain, is crucial to economic competitiveness. openai will try to help!”
Compute Capacity Now and in the Future
Altman has said that such a massive expansion of compute capacity is not only needed now – OpenAI kicked off the generative AI land rush in November 2022 when it released the widely popular ChatGPT chatbot – but also for down the road, as AI technologies grow their capabilities.
“The fundraising plans, which face significant obstacles, are aimed at solving constraints to OpenAI’s growth, including the scarcity of the pricey AI chips required to train large language models behind AI systems such as ChatGPT,” the WSJ reported. “Altman has often complained that there aren’t enough of these kinds of chips [GPUs] to power OpenAI’s quest for artificial general intelligence, which it defines as systems that are broadly smarter than humans.”
The CEO in November 2023 told The Financial Times that he hoped to raise “a lot more money” from Microsoft, which already is investing more than $10 billion into the eight-year-old company.
“The vision is to make artificial general intelligence [AGI], figure out how to make it safe and figure out the benefits,” Altman said at the time. “There’s a long way to go and a lot of compute build out between here and AGI.” He added that the cost of training LLMs “are just huge.”
The company relies heavily on GPUs to run its software portfolio and run its offerings – including ChatGPT, GPT-4, Codex for code generation, and Dall-E for image creation – and operations, such as servers, storage, networking and databases, on an AI supercomputer in Azure. Microsoft also uses OpenAI technologies throughout its product portfolio.
That’s a Lot of Money
The trillions of dollars that Altman is seeking for his AI initiative would dwarf the money spent right now on chips. According to the Semiconductor Industry Association (SIA), global sales of semiconductors last year was $526.8 billion, an 8.2% decrease from 2022, though sales accelerated in the second half. The group is predicting double-digit year-over-year sales growth in 2024.
“With chips playing a larger and more important role in countless products the world depends on, the long-term outlook for the semiconductor market is extremely strong,” SIA President and CEO John Neuffer said in a statement last week. “Advancing government policies that invest in R&D, strengthen the semiconductor workforce, and reduce barriers to trade will help the industry continue to grow and innovate for many years to come.”
In a statement to the WSJ, an OpenAI spokesperson said the company “has had productive discussions about increasing global infrastructure and supply chains for chips, energy and data centers – which are crucial for AI and other industries that rely on them. We will continue to keep the U.S. government informed given the importance to national priorities, and look forward to sharing more details at a later date.”