‘At Some Point, People Stop Trusting’: Billionaire Mark Cuban Says OpenAI CEO Sam Altman Is ‘All Over The Map’ And Warns ‘That Will Backfire on Him’

Mark Cuban built his reputation by challenging incumbents, talking big, moving early, and refusing to wait for the establishment to validate his ideas.

So when Alex Kantrowitz asked Cuban on the Big Technology Podcast whether he sees some version of himself in today’s AI leaders (specifically OpenAI’s Sam Altman and Anthropic’s Dario Amodei) the question seemed natural.

Cuban’s answer was revealing.

“Maybe a little bit in Dario,” Cuban said. “But not in Sam.”

That split says a lot about how Cuban views the personalities now leading the artificial intelligence boom. He does not see all AI confidence as the same. He appears to draw a line between aggressive conviction and chaotic overreach.

On Amodei, Cuban’s view was mixed but notably more sympathetic.

“Dario now is trying to scare the shit out of everybody now,” Cuban said. “But that’s part of raising money.”

That may sound like a shot, but Cuban immediately connected it to his own career. When he was building Broadcast.com in the 1990s, Cuban said, he also made sweeping claims about where the internet was headed.

“When we had Broadcast.com back in the day, I would say, ‘We’re just going to replace cable and satellite and all that,’” Cuban said. “And I kept saying, ‘It’s going to be in a couple years.’ And here we are 30 years later, and we’ve kind of gotten there.”

That is the paradox of big technology predictions. They can be wrong on timing and still right on direction.

Cuban’s old streaming prediction did not come true in “a couple years.” It took decades of broadband expansion, smartphones, connected TVs, cloud infrastructure, rights negotiations, subscription platforms, and consumer habit changes. But the broad idea, that internet distribution would eventually reshape television, radio, sports, and entertainment, now looks obvious.

That history helps explain why Cuban may be willing to give Amodei some room.

Amodei has become one of the most prominent voices warning that advanced AI could create serious risks. Anthropic has built much of its identity around AI safety, constitutional AI, enterprise adoption, and Claude’s reputation among developers. To Cuban, the warning-heavy posture may be partly strategic. Scaring people about the high stakes of AI can draw attention, attract capital, influence regulation, and position a company as the responsible alternative in a chaotic market.

But Cuban also suggested that Anthropic has something more concrete than fear.

“I think Dario is pushing forward with Claude,” Cuban said. “And I think, to your point, their niche with programming and agents, I think is going to hold for them.”

That is a significant compliment. Cuban is not simply saying Anthropic is good at messaging. He is saying Claude may have a durable lane.

Programming and agents have become one of the most important battlegrounds in AI. Coding assistants are not just flashy demos. They are among the clearest areas where AI can save time, produce measurable productivity gains, and become embedded into professional workflows. If a model becomes the preferred tool for developers, it can earn real enterprise revenue and gain deep daily usage.

Cuban appears to think Anthropic understands that.

He also said Anthropic may need to move toward a broader “world view” environment, suggesting that Claude cannot remain only a narrow tool. To compete long term, it may need to become more context-aware, more integrated and more capable of operating across a user’s broader digital environment.

Still, his assessment of Amodei was far warmer than his assessment of Altman.

“Sam is all over the map,” Cuban said. “And I think that will backfire on him.”

That is the most explosive line in Cuban’s answer.

Altman has become the defining public figure of the AI boom. Under his leadership, OpenAI turned ChatGPT into one of the fastest-growing consumer technologies ever, pushed frontier AI into the mainstream, struck massive partnerships, and became the company most associated with the race toward artificial general intelligence.

But Cuban’s criticism is not that Altman lacks ambition. It is that ambition can become a liability if partners, suppliers and customers begin to question whether a company will follow through.

Cuban pointed to reports and industry chatter around OpenAI’s huge infrastructure and chip-buying ambitions.

“I just think they were buying up 40% of this company’s memory chips and they just backed out of that,” Cuban said. “You can’t do that.”

His point was not merely about memory chips. It was about trust.

“At some point, people stop trusting and stop wanting to do business,” Cuban said.

That warning goes directly to the economics of the AI buildout. The biggest AI companies are no longer just software startups. They are becoming infrastructure giants by necessity. To train and run frontier models, they need long-term commitments involving chips, data centers, energy, cloud platforms, memory suppliers and financing partners.

Those relationships depend on credibility. If an AI company announces enormous plans, pressures suppliers to reserve capacity, helps distort market expectations, and then changes direction, the damage can go beyond one canceled deal. It can make the next supplier more cautious, the next partner more skeptical, and the next investor more focused on whether the company’s strategy is stable.

Corporate Trust Is a Compounding Asset 

Cuban’s argument is that trust compounds just like technology does.

That is especially important in a market where OpenAI, Anthropic, Google (GOOGL), xAI (SPCX), Meta (META) and others are all racing for talent, compute, enterprise customers, and developer loyalty. The companies are not only competing on model performance. They are competing on reliability.

A business choosing an AI partner wants to believe that company will exist, support its products, control costs, honor commitments and move in a coherent direction. A chip supplier or data center partner wants to believe massive orders are real. A developer wants to believe the platform they build on will not suddenly pivot. An enterprise customer wants to believe its AI provider will not become unstable, unaffordable or distracted.

That is where Cuban sees risk for Altman.

OpenAI’s strategy has often seemed expansive: consumer AI, enterprise AI, coding tools, agents, video generation, search-like products, infrastructure, chips, robotics-adjacent ambitions, partnerships with global corporations, and long-term talk of artificial general intelligence. To supporters, that breadth reflects the scale of the opportunity. To Cuban, it may look like a company trying to do too much at once.

That is why his contrast with Amodei matters.

Cuban seems to view Anthropic as having a clearer wedge: Claude, programming, agents and enterprise usefulness, wrapped in a safety-first message. He may not agree with every warning Amodei gives, but he appears to see a more coherent business identity.

Altman, by contrast, represents the risk of the AI era’s biggest ambitions becoming too sprawling to manage.

The irony is that Cuban himself has always been a showman. He knows the value of bold predictions. He knows that founders sometimes have to sell a future before the market fully believes it. Broadcast.com was built on a vision of internet media that sounded absurd to many people at the time. Cuban’s whole career is a case study in how early conviction can look ridiculous before it looks obvious.

But Cuban is also saying there is a difference between selling a vision and eroding trust.

His Broadcast.com example is telling. Cuban overestimated the speed of the transition from cable and satellite to internet distribution, but the direction was consistent. He was pushing one big idea: media would move online. That idea became truer over time.

His criticism of Altman is that OpenAI’s direction may be harder for outsiders to pin down. If a company keeps making huge commitments, shifting priorities and expanding its ambitions, the market may eventually ask whether the strategy is visionary or scattered.

That is a dangerous question for any company. It is especially dangerous for one trying to raise and spend at historic scale.

Cuban’s comments also underscore a broader shift in the AI conversation. The first phase of the ChatGPT era was about astonishment: what can this technology do? The current phase is about trust: who can build it, fund it, deploy it and maintain confidence while the costs explode?

That is where founders like Altman and Amodei matter as much as the models themselves. Their public statements can move markets, shape regulation, attract capital and influence whether companies feel safe betting their workflows on their tools.

Cuban’s verdict is not that Amodei is perfect or Altman is doomed. It is more specific. He sees Amodei’s fear-based messaging as part of the fundraising and positioning game, but backed by a Claude strategy that may hold in programming and agents. He sees Altman as brilliant but scattered, and he thinks that scattered approach could eventually cost OpenAI something more valuable than money: trust.

In the AI race, the companies are spending billions to buy compute.

Cuban’s warning is that credibility may be harder to replace.