“Everybody’s asking me about it, ‘Is there a bubble here?’ Absolutely there’s a bubble. It’s huge,” he tells Fortune in an exclusive interview at C3.ai’s New York offices in a Midtown WeWork. 

Over the last two years, analysts have pondered whether AI companies, both public and private, could possibly live up to their lofty valuations. Tom Siebel, who built his career in Silicon Valley as a sales executive at Oracle before leaving to start his own company that he eventually sold back to his former employer for $5.8 billion, the current state of AI reminded him of the dot-com bubble. Even then a great and wondrous technology—the internet—couldn’t save a host of companies from coming crashing down.  

“So we have this similar thing going on with generative AI that we’ve seen with previous technologies,” Siebel said. “The market is way, way overvaluing.”

Tech analysts that Fortune spoke to broadly agreed with Siebel’s point that valuations across the industry were inflated. “For now, virtually every notable AI company enjoys a fair degree of investor hype,” said Sandeep Rao, senior researcher at Leverage Shares, a provider of ETPs. 

C3.ai specializes in enterprise AI applications that help companies with various business functions like optimizing their supply chain, predictive maintenance, and tracking their sales process. It also has a spate of lucrative government contracts with the likes of the U.S. Department of Defense and the U.S. Air Force. Among its biggest private sector customers are oil and gas giant Shell and energy company Baker Hughes (whose contract is up for renewal soon). 

Earlier this week C3.ai added another blue chip partner to its ranks when it announced a partnership with Microsoft. Fortune’s interview with Siebel was conducted before the partnership was publicly revealed.

In particular, Siebel took aim at OpenAI, the startup with close ties to Microsoft and that is perhaps most closely associated with the AI revolution. OpenAI currently has a $157 billion valuation after an October funding round in which it raised $6 billion. Siebel wasn’t impressed by that valuation. 

“Nobody would be surprised if that company disappeared next Monday,” he said. 

When Fortune ventured that industry observers would be surprised, Siebel responded that it had “disappeared” over Thanksgiving, a reference to the brief ouster of OpenAI CEO Sam Altman in 2023. 

“If it disappeared, it wouldn’t make any difference in the world,” Siebel said of OpenAI. “Nothing would change. I mean, nobody’s life would change. No company would change. Microsoft would find something else to power Copilot. There’s like 10 other products available that would do it equally as good.”

The OpenAI brand name has cachet because it was first to market, but that alone can’t guarantee its market position in perpetuity, said Paul Marino, chief revenue officer at Themes ETF, an exchange-traded fund based out of Greenwich, Conn. “Just because you’re very well known doesn’t mean that you can’t be copied, replicated, and maybe even surpassed,” Marino said. 

In Rao’s view, there are differences between large language models, but they are difficult to understand. “LLMs are highly proprietary and nailing down definitive distinctions isn’t easy,” he said.  

Their success, he added, is often due just as much to their business relationships as it is to their underlying tech. “An LLM’s advantage isn’t necessarily dictated by quality but could be dictated by low cost barriers and ease of use with existing tech instead,” Rao said. 

In this regard, OpenAI certainly fits the bill, having established deep ties with Microsoft. 

OpenAI did not respond to a request for comment regarding Siebel’s comments. 

Siebel sees overvaluations throughout early stage AI startups as well. 

“There’s a long list of AI startups out of Illinois, Wisconsin, Stanford, that are being financed today on Sand Hill Road, where very little ideas by people who are highly inexperienced, who are going to build generative AI applications for dentist office, veterinarians, or divorce lawyers and these ideas are being financed at multi-billion dollar valuations,” Siebel said. They’re “just five people who don’t know anything [with] four pages of a business plan. This is crazy.”

Over the last few years, a spate of AI startups with highly specific use cases has cropped up, some of which have indeed sold or raised money at large valuations. Their track records have been a mixed bag. In August 2023, Casetext, which specializes in AI for legal work, sold to Thomson Reuters for $650 million. JasperAI, a startup focused on AI for marketing departments, raised $125 million at a $1.5 billion valuation in a Series A in June 2023 only to cut its internal valuation three months later, according to The Information. 

Exempt from Siebel’s criticism are the major tech giants developing their own suite of AI products. Microsoft and Amazon, he said, are “great companies” that aren’t overvalued. Neither are chipmakers Nvidia and TSMC. “If TSMC went out of business it would be the end of the world,” he said. 

When asked where C3.ai belongs, Siebel naturally has no doubts. “C3.ai is a bargain, okay? I mean, it’s a value stock buddy,” he said. 

Disclosure: Alan Murray, the former CEO of Fortune Media is on the board of C3.ai.