Big quote: Businesses are paying for AI models without consistently getting enough value from them, even as developers plan to spend more on chips, data centers and new systems. The uneven results are becoming a concern for investors trying to judge whether the industry's spending will produce the profits they expect. According to Ken Griffin, companies are still working out how to make AI useful enough to justify the expense.
"Corporate America has been very uneven in its ability to extract value from its AI models," Griffin said at the Future of Global Markets Conference in New York on Tuesday.
The billionaire, whose businesses include hedge fund Citadel and market maker Citadel Securities, said investors are "almost certainly" in an AI bubble. But he also expects the technology to create opportunities for entrepreneurs and reshape competition across industries.
Anthropic's expansion plans show how much capital AI development could require. The company is expected to publicly file paperwork for a stock market debut in the coming weeks. The offering could rank among the largest ever, raising money for the chips, data centers and talent needed to build new models.
The Wall Street Journal has reported that Anthropic could spend $750 billion on additional computing power in the coming years.
That spending would come as businesses continue to assess what they get from the models already available. Griffin said Anthropic would need to persuade investors that its technology is essential across industries.
For developers, the challenge is to keep improving their models while building demand strong enough to support those investments. For customers, it is to turn access to the technology into measurable gains at work. Griffin's comments suggest that many companies have yet to establish that connection consistently.
Investors have grown more anxious in recent weeks about the time between paying for AI infrastructure and seeing a return. The concern is that profits and productivity gains will arrive more slowly than expected, despite the substantial sums committed upfront.
Griffin's warning about a bubble came alongside a more optimistic assessment of AI's business potential. He said the technology could help launch new companies while weakening the advantages of established ones.
"You're seeing the threat to white collar jobs playing out; as a business owner, you're seeing your competitive moats be filled in by bulldozers," he said. "At the same time, you're thinking that this will be a golden era for entrepreneurs."
His view has changed over the year. Speaking in Davos early this year, Griffin called AI garbage. In May, he told an audience at Stanford Business School that he was depressed by all it could achieve. On Tuesday, he gave greater weight to its potential to spur business formation in the United States.
He argued that a speculative boom does not rule out lasting technological change.
"Any radical transformation of our economy is generally intertwined with a bubble," Griffin said. "These bubbles, if we want to use that word, often coincide with these giant technological transformations."
He pointed to the dot-com era, when what he described as fanciful business models attracted investment before the crash. Some companies nevertheless emerged with businesses that endured long after the speculation faded.
"The companies that are on the other side of that bubble are the market leaders for decades to come," he said. "I just wish I'd kept my Amazon stock in that moment."