AI Bubble: Experts Reveal Which Companies Are at Risk to Money.it

The AI bubble is now discussed on a daily basis, and even more so in recent days, as Wall Street-listed tech stocks continue to skyrocket.

Rather than mere words, specific figures tell the story of what is happening and explain why it has become inevitable to question the sustainability of the current market.

One need only look at the trend of the Philadelphia Stock Exchange Semiconductor Index, which tracks the shares of chipmakers: the index is on track to close the best quarter in its history, having surged 69%.

Many experts consider these gains blatantly excessive. This lends credibility to the fears and predictions of those who argue that the AI bubble is already here and about to burst.

Alongside the alarm bells comes another burning question: if the bubble bursts, who will be the casualties left to foot the bill?

Money.it gathered insights from several leading experts.

AI Bubble: Anthropic, OpenAI, and xAI at Greatest Risk

Igor Pejic, high-tech market strategist, banker, and author of Tech Money, Blockchain Babel, and Big Tech in Finance, weighed in on the risks:

“The highest risk companies are the ones going public in the next months: Anthropic, OpenAI, and xAI (as part of SpaceX). Not only are they pure AI players, but they focus on a particular type of AI, namely frontier models. They are pursuing a super-AI that is on par with human reasoning. This is a high-stakes bet that will require further breakthroughs to work out. If they fail to achieve it, their entire business might collapse. Moreover, they are dependent on outside capital and thus on the macro-economic climate. On the other hand, players like Google, Amazon, and Microsoft have stable, well-diversified business with massive free cash flows. Plus, they benefit from scaling AI, whoever comes up with the best model”.

On whether investors are underestimating the risks, the guru - frequently quoted by The New York Times, American Banker, Forbes, and Bloomberg, and author of the Substack newsletter The New Frontier - told Money.it that “there are many parallels to other productive tech bubbles like the dot-com era”, adding that “ AI is at the peak of its hype cycle . Adoption has exploded. Corporate valuations are at a similar level. We are starting to see some circular financing”.

But according to Pejic, there is one crucial difference: “ The tech economy today is much more advanced. The ratio of corporate investment to revenue is better and it is dropping despite an increase in funding. Investments are still done mainly from free cash flow. And I don’t see a likely scenario in which Big Tech is going out of business, nor even yielding its top spot in the market cap list ”.

This means that “compared to previous bubbles, investors have become much more sophisticated and technologically savvy” and that “though many of the investments might be driven by FOMO (fear of missing out), most investors are mentally putting their AI stocks in the ’high-risk bucket’”.

However, he pointed out a hidden danger, warning that “ what investors are underestimating is the impact an AI collapse would have on the overall economy. Especially in the US, tech has reached an unseen proportion of the financial markets. Even if you own index stocks, an AI crash would be devastating”.

Some See Even the AI Titans - Amazon, Google, Meta, Microsoft - at Risk

David Viney, IT sector professional and AI consultant, also spoke with Money.it.

When asked whether AI stocks are overvalued, he stressed that “ everything is priced to perfection, which means it only takes one of many possible things to go wrong for something to go wrong”.

He noted that “ Goldman Sachs projects $1.15 trillion in AI capex between 2025 and 2027”.

This massive spending is “ not irrational if enterprise demand materializes at the scale the infrastructure thesis requires”, Viney observed, “but my sense is it won’t, not yet”.

His skepticism runs so deep that, according to him, the companies most exposed to an AI bubble burst are the “Mag4” themselves: Amazon, Alphabet-Google, Meta, and Microsoft.

These infrastructure builders and hyperscalers are “financing their own demand” and have “ guided $635-665 billion in 2026 capex alone ”.

Viney made an interesting comparison between what’s happening and the Woodstock festival: “ To borrow from the Woodstock analogy, they are organising the festival and booking the bands. The crowd -enterprise customers with genuine AI ROI - hasn’t bought its tickets yet ”.

Furthermore, investors are underestimating “physical constraints”, meaning that “there isn’t enough capital in the world to finance it all, and there isn’t enough power—in Europe at least—to run it all".

For instance, " Ireland, the Netherlands, and Denmark are all closed for business; grid connection moratoriums are stranding billions in fully permitted projects”.

The reality, he argued, is harsh: “ You cannot financial-engineer your way around the laws of thermodynamics , and that constraint is almost entirely absent from the financial models being presented to investors”.

When asked if the fever for the AI can be defined as a bubble, Viney hesitated to use the term:

“Bubbles burst and leave nothing behind. This is more like a soufflé - the rise is real, the deflation is physics, and what remains when you put your spoon in is still good. The dot-com crash gave us near-zero bandwidth costs, which enabled the cloud infrastructure we rely on today. This will follow the same pattern. The infrastructure will survive. Many investors won’t”.

A bubble like 2000? Absolutely

Mike Roberts, co-founder and CEO of City Creek Mortgage, has no qualms about calling it an AI bubble:

“AI stocks are being traded purely on speculation rather than hard, tangible financial data, much like what we saw during the formation of a classic asset bubble. The large-cap tech giants can obviously afford to spend tens of billions to build out this massive infrastructure. Conversely, numerous mid-tier software companies are seeing their valuations rise dramatically solely because they have labeled their product ’an AI’, and, as a result, investors are speculating wildly about their future potential”.

This means that, according to Roberts, worrying about an AI bubble makes perfect sense. “The way money has been flowing toward this new technology prior to the development of scalable business models is extremely similar to how money flowed into the dot-com bubble ”, he told Money.it.

The CEO of City Creek Mortgage underscored that, “as before, there is a strong assumption in the current market structure that each AI venture will be highly successful, regardless of the high computational cost of running many of these models”.

He added that “from my experience, a lot of early-stage infrastructure growth booms experience a major price correction before becoming widely accepted and common-use utilities”.

AI bubble: beware of those counting their chickens before they hatch

Brent Fisher, Co-Founder & Head of GTM at at Cognetryx, also commented on the potentially devastating consequences of an AI bubble burst.

On the most exposed segments, he noted that “the companies with real downside are within the cloud-AI middleware layer, whose revenue projections assume regulated industries will keep routing their most sensitive data into hosted models”.

Fisher recalled that “ the Fed, OCC, and FDIC issued revised model-risk guidance in April (SR 26-2), HIPAA’s Security Rule has been updated to address AI governance directly, and the EU AI Act is being enforced”, cautioning that “ healthcare, banking, legal, and government buyers, who control trillions in spend, are increasingly answering the cloud-AI pitch with ’we cannot send that data outside our network”.

So, “any company whose growth curve depends on those buyers ignoring those rules is repricing soon ”.

On whether a comparison to the dot-com bubble is justified, Fisher offered this perspective:

“There is one, but most people draw the wrong version of it. The dot-com bust didn’t kill the internet. It killed speculative consumer front-ends that lacked a revenue path. Speculation is often the key erosion factor in capital that drives investment in innovation. The infrastructure plays (networking, databases, hosting) survived and compounded, and we will see the same thing here. The labs and the companies with real enterprise contracts and application defensibility will be fine. What dies is the second tier: AI products whose business model depends on processing data the buyer is no longer allowed or willing to send out. From where I sit, that’s a meaningful share of the current market cap - you can see proof of that in the sheer amount of companies advertising to that effect”.

The Bottom Line

The debate over the AI bubble is not a question of if a market correction is coming, but where it will hit hardest.

While speculative, second-tier startups face the most immediate danger, US Big Tech titans—including the “Mag 4” (Amazon, Google, Meta, and Microsoft) are also at risk due to their colossal capital expenditures.

Ultimately, the AI “soufflé”, as Viney said, may deflate, but the underlying infrastructure is here to stay, leaving over-leveraged investors and overextended giants to foot the bill.