Is This an AI Bubble? Or Just the Cost of Building the Future?

If AI is the future, it’s starting to look a lot like 1999. Venture capital is pouring in. Big Tech is spending at record levels. And valuations for AI companies have soared to eye-popping heights.

Michael Albert, a professor in the Data Analytics and Decision Sciences area at the University of Virginia Darden School of Business, isn’t surprised by the frenzy — but he urges cooler heads.

“There’s an increasing perception that there is an AI bubble,” Albert told Darden’s Brett Twitty in the latest episode of Office Hours, a faculty spotlight series. “And frankly, maybe there is an AI bubble. Even if this technology is hugely transformative — but it’s transformative in 50 years’ time — then what we’re seeing now is probably an AI bubble.”

But while Albert sees echoes of past tech booms, he doesn’t think the story ends there. Some companies, he says, are playing a long game.

“My own read of their strategy is that they're viewing this in a startup-focused manner. What they are trying to do is gain technical expertise and market share — those two things are hugely valuable,” he says. “And so for those companies, I think their massive-level investment makes sense.”

The imbalance between cost and return is striking. OpenAI, for example, reported around $4.3 billion in revenue in the first half of 2025 — while spending $2.5 billion on research, development and compute.

“It’s really asymmetric between revenue and investments at this point in time,” Albert observes.

Even so, he doesn’t see disaster ahead. “I don’t think that this infrastructure spending boom is in the same level of bubble as we saw in the dot-com era or the housing crash in 2008 where we had a lot of subprime lending,” he says.

Moreover, some risk is being priced into the market. “People aren’t just throwing money at AI companies,” he says. “And I think that pushes a little bit against the AI bubble narrative.”

Why Most Firms Still Get AI Wrong

From a business standpoint, Albert says the key challenge for leaders is understanding what AI can — and can’t — do effectively.

“You’re starting to see a lot of articles that are along the lines of, ‘company tried AI, didn’t find any value in it.’ Ninety percent of companies have found no return on their AI investments,” Albert says. “By and large, that’s because a lot of people are using AI in the wrong way. They don’t really understand what it is, and they’re using it like a replacement for Google search.”

That, Albert argues, will never justify the massive compute budgets and staffing costs required to run the largest models. “We’re replacing one thing that was already pretty good with something that’s slightly better,” he says. “You’re not going to see big productivity gains.”

Instead, he sees opportunity in well-scoped, measurable applications. “Business leaders need to think about AI as custom solutions for specific, relatively well-scoped tasks that their company is dealing with and that they can scale,” he says.

The key, he adds, is matching the tool to the right kind of problem — and understanding what success looks like.

Albert believes the next wave of innovation could emerge not from smarter algorithms alone, but from better human understanding of how to use them.

“What I think is more promising — but I think is being sold poorly — is this move to what we call agentic AI, where we have these AI agents that are supposed to be doing tasks on our behalf,” he says. “For a lot of people, they don’t have a good, deep understanding of the limitations of what these models can and can’t do. So they pick automation tasks that actually are poorly suited.”

Teaching the Next Generation of Data-Savvy Leaders

Albert’s research focuses on combining machine learning and algorithmic techniques to automate the design of markets.

In addition to his research, Albert teaches across Darden’s Master of Science in Business Analytics (MSBA) program — a joint offering with UVA’s McIntire School of Commerce — and in the Darden MBA elective “Machine Learning and AI for Business.”

His MSBA students learn to translate data into better business decisions. “The real emphasis is on practical, business-focused analytics,” Albert says.

His MBA elective is both popular and challenging. “It's a hard elective,” says Albert. “One of the pitches that I have for that class, and I think for a graduate degree in general, and for Darden in particular, is that it's a wonderful time to stretch yourself.”

The class is set up for people with no technical background. “[Students are] enthusiastic about trying to understand the fundamentals of machine learning and AI and, more importantly, understand how those things can actually drive value,” he says. “The basic thesis of the class is that understanding the technical details, at least at a high level, helps you understand the value proposition much better.

The goal, he says, is less about producing coders and more about creating translators — leaders who can spot opportunities and evaluate claims critically. “One of the things we’re seeing right now in the AI industry is that there’s a lot of good marketing around AI, and maybe less strong substance behind that marketing,” he says. “The goal of the class is to help students have a much better foundation for trying to parse: what is the actual technical contribution? Is it likely to actually drive value?”

Keeping Perspective in the Age of AI Hype

For Albert, the long game is clear: Business leaders who combine technical literacy with strategic judgment will be ready for whatever comes next.

“It would be unjust of us as a business school not to spend a lot of time preparing students for wrestling with these questions,” he says. “Long term, I think we’re dealing with some profound, fundamental shifts in our global economy.”

To listen to the full conversation, visit  “Office Hours” with Professor Michael Albert, Presented by Darden Ideas to Action.