Who will survive the AI bubble?

Sky-high valuations. Miles of data center build-outs. Circular funding deals. Soaring debt. Hundreds of billions of dollars being committed to — and by — loss-making companies.

What began as whispers of an AI bubble have crescendoed to a fever pitch, as some of the most powerful names in tech and finance warn that the multi-trillion dollar industry could be headed toward a cliff-edge.

“Are we in an AI bubble? Of course we are,” former Intel CEO Pete Gelsinger said in October. “We’re hyped. We’re accelerating. We’re putting enormous leverage into the system.”

Amazon founder Jeff Bezos had a similar warning. “When people get very excited … for example, every experiment gets funded,” he said. “Every company gets funded — the good ideas and the bad ideas — and investors have a hard time … distinguishing between the good ideas and bad ideas.”

Chipmaking giant Nvidia’s latest earnings smashed expectations, with record revenue on the back of surging demand for AI chips – news that industry proponents took as vindication that AI infrastructure growth isn’t just speculation.

But many analysts and economists still warn of a high-volatility boom built on unproven, long-term bets.

The amount of money being poured into AI, and the speed at which data center construction is accelerating, is dizzying. Total AI spending globally is on course to reach $375 billion this year, and is predicted to hit $500 billion in 2026, according to a forecast by UBS. OpenAI is eyeing an eventual IPO with a valuation as high as $1 trillion, and says it will have $1.4 trillion in compute costs over the next eight years — all while registering annual losses.

The bet is that future demand and productivity returns — and the payoff from being at the front of the pack in the AI race — will be high enough to more than justify the costs. But the nascent tech’s promised economic and productivity gains remain yet to be seen, and may take far longer to materialize than many investors anticipate.

And the wider economy is now tied to the fate of these AI firms, as some of the world’s largest companies and stock market heavyweights rely on OpenAI and its competitors to fulfil enormous contracts and cultivate demand.

Many market watchers have drawn parallels to the Dotcom bubble of the early 2000s. But the reality is more complicated, and doesn’t fit into a black-and-white narrative.

AI bubble vs. Dotcom bubble

For starters, the major tech companies currently pouring money into AI investments have cash flow, while most who dove into the late 1990s internet boom did not. Blue chips like Oracle, Meta, Microsoft, Amazon and Google have war chests of resources to invest, backed by proven business models.

Still, many believe the share of the economy devoted to AI investment is far too high — “nearly a third greater than the share of the economy devoted to internet-related investments back during the dotcom bubble,” warned Jared Bernstein, who served as chair of the U.S. Council of Economic Advisers under the Biden administration.

And capital expenditure by S&P 500 companies, as a proportion of GDP, is higher than it was during the dotcom bubble.

But those capex figures are still about 40% of operating cash flow, compared to more than 70% during the dotcom rush, according to HSBC Global Investment Research.

For many in the industry, the question of the bubble bursting is not “if” but “when”.
Former Intel chief Gelsinger doesn’t see it happening for “several years”, adding that while there’s been an "industry shift to AI," businesses have "yet to really start materially benefiting from it."

And that brings us to what is arguably the single most important variable in our bubble conundrum: when will businesses truly start seeing ROI from AI, and to what degree?

Currently, it’s not looking great. A report from McKinsey earlier this year found that roughly 80% of companies deploying AI in their businesses saw no impact on their bottom line. And while seemingly everyone is trying to name-drop AI use in their earnings calls, they struggle to show value gained — 42% of companies abandoned most AI projects in 2025 due to cost and unclear value.

Still, 97% of business leaders plan to increase generative AI investments, according to McKinsey’s research — meaning a tremendous market opportunity for AI tools that are actually fit for strategic, enterprise-wide deployment.

This gap between investment and value is where the next generation of AI platforms will be determined.

Opus, which merges advanced AI with expert human review, is one of those tools. The first AI-native platform for supervised automation delivering reliable execution for complex business workflows, Opus is designed for enterprises in highly-regulated industries, saving companies countless costs and man-hours. Streamlining administrative tasks, eliminating risk and ensuring accuracy and regulatory compliance, Opus enables businesses to drive real productivity at scale, delivering those coveted returns on investment.

Who will survive?

Like in the late 90s, some companies will disappear and ultimately become worthless, while others will survive and shape the future of our societies and economies, in the same way that the internet did.

Tech investors typically bet that a small handful of their portfolio companies will become big enough winners to offset the losses. In venture capital, roughly 6% of investments account for about 60% of returns, according to an a16z blog from 2015 citing the previous three decades of data. Many VCs believe that when it comes to AI, an even smaller fraction of investments will generate far more than 60% of the returns.

Which companies will survive, and which will become casualties, is yet to be known — as is the scale or timeline of any anticipated bubble bursting.

But when it comes to B2B AI platforms, watch for the companies whose tools laser-focus on scalability, integration capabilities, deep technical due diligence architecture, and security frameworks. And the ability to solve business’s specific, real-world problems.

It’s worth remembering that in 1998, the Nasdaq dropped nearly 20% — but 18 months later, its market cap tripled.

Fast forward to today: how much has the tech-heavy index grown from its dotcom trough?

A casual 1,430%.