The nearing end of the AI bubble could signal a new start for some AI technology companies.
AI technology has attracted the attention of both consumers and investors due to its rapid innovation and revenue potential, resulting in AI tools flooding the market. However, despite the wave of innovation, few efforts have made it past the development stage. That's because they were built on fragmented tools and were too disconnected from business goals, said Juan José López Murphy, head of data science at Globant.
This situation made some investors uncomfortable.
“The frustration we experience, some people call deflation, is the fatigue of too many models and too few strategies,” Lopez-Murphy said. “Rather than investors and management teams pulling out completely, we are starting to see a shift in focus from volume to value.”
Investors are now looking for projects that deliver sustainable and measurable outcomes across teams and systems, resulting in a more disciplined market, he said.
The big difference between what's happening now and previous investment bubbles is that AI technology is real, in terms of the actual value it brings, said Matt Hasan, CEO and Ph.D., aiRESULTS Inc. Hasan said that as an AI strategist and economist, he views the market through two lenses: technology and the money that goes after it.
Unrealistic expectations lead to overinvestment in generative AI, with 95% of organizations experiencing zero returns, according to an MIT report. Currently, the amount invested exceeds the amount invested.
“The risk comes from how quickly investors are coming together,” Hasan said. “Money is flowing into data centers, chips, and startups faster than actual implementation can keep up. The hype is outpacing the actual business value.”
According to Jonathan Bittner, CEO of Dimensional Analytics, there are three key factors that distinguish this bubble from other bubbles. These methods include:
- Circular financing. Major companies in the AI industry are investing in each other, creating the illusion of economic activity. Bittner gave the example of Nvidia investing $100 billion in OpenAI, OpenAI buying Nvidia chips, and Oracle buying Nvidia chips for its OpenAI center.
- Profitability timeline. Many startups initially make losses, but have a plan to achieve profitability. OpenAI is losing $12 billion every quarter and expects to lose another $44 billion by 2029, Bittner said. “They spend $2.25 to make $1. No dot-com company survives at that burn rate.”
- National security framework. AI companies are folding themselves into defense contracts, which could lead to bailout requests, Bittner said.
Early signs of bubble bursting
Forrester predicted a correction in the AI market in 2026 as organizations grapple with the ever-widening gap between inflated vendor promises and actual value delivered. By 2026, CEOs will rely on CFOs to approve AI investments based on ROI, as fewer than one-third of decision makers can translate the value of AI to financial growth for their organizations, the report says.
“The era of AI hype will end in 2026 as pressure increases to deliver real, measurable results from secure AI initiatives,” Sharyn Lieber, Forrester's chief research officer, said in the report. “As the era of volatility continues, technology and security leaders will be required to realign their investments with greater financial oversight and governance while addressing increasingly complex geopolitical and economic risks.”
Hasan is experiencing a degree of deflation as venture funding cools, valuations flatten and companies realize that promised profits will take longer to materialize.
“It's not a collapse. It's a pause to catch your breath,” Hasan said.
Bittner said if you look deeper, you'll see additional red flags, including:
- Economic dependence. Without investment in AI, Bittner said, economists could already be heading into a recession. The manufacturing industry contracted for the seventh consecutive month.
- Infrastructure bottlenecks. The country is expected to face an electricity shortage of 35 gigawatts by 2028, Bittner said. The data center requires 57 GW, but only 21 GW is available. “Most of the new data centers in the Internet hub of Ashburn, Virginia, are powered by natural gas generators, because the power supply won't even catch up,” Bittner said. “This is not a temporary bottleneck. It's a physical issue.”
- National sentiment changes. Consumer confidence is at its lowest level since 1997, and AI is not showing the promised ROI returns. “Public perception is very important,” Hasan said. “When the excitement outweighs the results, people begin to question whether AI is really improving their jobs and lives. As that doubt creeps in, the emotional drive behind the boom begins to dissipate.”
Impact of bursts on businesses
Hasan said the bursting of the bubble will not doom AI, only the unrealistic expectations surrounding it. He believes that companies that focus on solving real problems are stronger.
Bittner shares Hasan's optimistic outlook. He said the burst itself would hit organizations in three waves:
- Wave 1. AI companies, including many AI startups, will see layoffs. “You're talking about tens of thousands of highly paid engineers suddenly coming onto the market,” Bittner said.
- Wave 2. Bittner said 56% of companies are undershooting their AI cost projections by 11 to 25%, and one in four companies are undershooting their AI cost projections by more than 50%. As CFOs demand results, entire AI departments will be cut, Bittner said.
- Wave 3. Real innovation will accelerate. True builders will survive the crash and start solving real problems with the right technology, Bittner said. He pointed to companies like Google, Amazon, and PayPal, which gained momentum after the dot-com bust because they had solid business models. He believes the same thing will happen with AI.
“The bursting of the AI bubble could mark the beginning of a healthier, more focused innovation cycle for AI, with companies starting to consolidate project budgets or potentially delaying hypothetical projects,” Lopez-Murphy said. “This period does not mean that innovation has completely stopped, but rather indicates a maturity in project prioritization.”
He believes the focus will shift to building a framework that lasts over the long term.
How CIOs can adapt and protect their organizations
With the threat of the AI bubble bursting, industry leaders will need to see through the hype and focus on AI technologies that solve concrete business problems. Hasan advises leaders to stay grounded and connect AI projects to uncover outcomes and track costs.
“If companies calm down and focus on the real impact, the adjustment won't be a disaster. The adjustment will reset and be healthy,” he said.
Bittner added his own recommendations:
- Demand measurable ROI upfront. If an AI vendor can't demonstrate time savings, error prevention, and revenue generation within 90 days, walk away.
- I prefer narrowly focused solutions to general AI. A tool that does one thing extremely well is better than a general-purpose tool that does everything poorly, he said.
- Don't think you need the most advanced model. This is where most organizations waste money, Bittner says. Many production AI applications work fine with small open source models that can run at the edge.
- Let's see who is making money. AI companies that are losing billions of dollars every quarter are not building sustainable technology. He said they should partner with companies that have real business models.
- Audit your spending on AI. Most organizations find that 80% of their AI spend is spent on experiments that never ship. Invest in the 20% that works.
For Hassan and Bittner, the bursting of the bubble will signal a new, more efficient approach to investing in AI technology.
“The bursting of the bubble won't be a disaster. It's a reset. And the hard-working, smart people of America will use that reset to revolutionize the economy with AI. That's the right way to go,” Bittner said. “It's always a people problem, not an algorithm problem.”
Julie Hanson is a freelance writer who covers local news in Massachusetts and New Hampshire.
