5 Stocks That Could Go to Zero if the AI Bubble Bursts

Listen to the audio version of this article (generated by AI).

With the massive amount being invested in artificial intelligence (“AI”), the scorching run-up in AI-related stocks, and the terrible economics at OpenAI and Anthropic, there are growing fears that a massive AI bubble has formed.

And if it pops, many AI stocks can expect to see a sharp decline in value. But some may go to $0 or so close that it doesn’t even matter.

The idea that AI is a bubble is so prevalent that even institutions are now calling it out, and it’s no longer just cranky short sellers shouting into the void.

Recently, the Bank of International Settlements noted:

In the near term, the ongoing AI investment boom raises questions about the sustainability of the current economic expansion. The five largest hyperscalers are set to spend over a trillion US dollars on AI-related capital expenditure from 2025 through 2026…. Disappointment in returns could trigger a sudden pullback in financing and turn the capex boom into a protracted investment bust, with potential knock-on effects on financial conditions….

An AI bubble burst could even lead to widespread declines in stocks unrelated to AI. That’s just the nature of market-wide declines.

But even within the sector, the impact will hit differentially. AI is the dominant income stream for some companies, while it’s only one of many business units for others, even if it’s still an important revenue source for these firms.

So, companies with heavy AI exposure, such as Microsoft (MSFT) or Amazon (AMZN), but with broad and profitable business lines, have more stability than a pure-play AI firm heavily reliant on one or two contracts. Yet even these stalwarts must be careful not to take on too much debt, as Alphabet (GOOGL) showed when it pivoted to raising $80 billion in stock rather than issuing new debt.

But many AI firms depend so much on capital spending from hyperscalers, such as Microsoft and Amazon, that they won’t survive if AI investment falls off a cliff. Moreover, the economics of AI modelers, such as OpenAI, are so poor – the firm lost $20.9 billion in 2025 – that it needs constant infusions of cash to stay alive and meet its own $1.4 trillion in spending commitments.

Given the precarious economics of AI, the following five stocks could be among the biggest losers when the AI bubble pops.

1. Smartbird (BIRD)

You likely know Smartbird (BIRD) by its former name, Allbirds, back when it was a shoe company, of all things. After that failure, the company pivoted under new management to AI. It’s hard to be too cynical when a firm with a $36 million market cap shifts to the world’s hottest investing trend.

In a press release announcing the shift, Smartbird said that it “will initially seek to acquire high-performance, low-latency AI compute hardware and provide access under long-term lease arrangements” as it focuses on customers that “hyperscalers are unable to reliably service.”

The shift is exactly the kind of move that companies make when they want to piggyback off a trend. Investors saw the same playbook back in 1999-2000, when any company with even the flimsiest business model appended “.com” to its name to raise millions or billions.

Here’s more about why investors should be skeptical about this company’s AI pivot.

2. Coreweave (CRWV)

Coreweave (CRWV) is another company spending heavily to create growth that, in some cases, may never materialize. Earlier this year, the company announced that it planned to spend $32.5 billion in 2026, while expecting sales of $12.5 billion. If the company’s backlog of $100 billion in revenue doesn’t arrive, it will likely be tough times for the company.

But even revenue growth is not translating into higher profitability, as margins deteriorate. Operating margin slipped from around negative 3% to negative 7% in the first quarter year over year, despite sales growth of about 112%. At the same time, net margin fell from about negative 32% to negative 36%, with a total loss of $740 million in the first quarter of 2026. Growth seems to lead to even wider losses.

Coreweave relies heavily on liabilities to finance its assets. Shareholders’ equity amounted to just $4.8 billion in the first quarter, while assets were $55.6 billion, leaving liabilities of $50.8 billion.

Finally, insiders – the people who know the company best – are dumping the stock aggressively. In June alone, insiders such as the chief development officer, CEO, and CFO sold more than $200 million in aggregate. But it’s not just June. Insiders have been steadily selling all year long, and there have been no reported buys from them. It’s an enormous negative sign for the business.

3. Cerebras Systems (CBRS)

Let’s start with the worst news upfront: AI chipmaker Cerebras Systems (CBRS) relies heavily on OpenAI for its future revenue. It’s inked two contracts this year for $20 billion in revenue to be delivered over the next few years. For context, management expects to report sales of less than $1 billion in 2026, so Cerebras really, really needs OpenAI to continue to survive.

But the risk of customer concentration has always been part of the investment risk at Cerebras. The company has signed a multi–year deal with Amazon to offer its solutions on Amazon Web Services, with revenue beginning to come in over the next year. But Cerebras still needs to rapidly expand its customer role to reduce its over-reliance on the money-burning OpenAI.

Still, Cerebras has a lightning– quick chip that can speed up AI inference workflows, and it boasts that it’s the “maker of the world’s fastest AI infrastructure.” With estimated 2026 sales of $865 million at the high end, the stock trades at a nosebleed 57 times sales.

4. Nebius (NBIS)

Nebius is another provider of AI cloud infrastructure that’s heavily reliant on future revenue to continue its build-out of data centers. The economics don’t look so great here, either.

In the first quarter, revenues exploded 684% percent higher, to $399 million. Yet the company’s adjusted operating loss actually increased from negative $120.3 million to negative $128.0 million. Gross margin plummeted from last year’s first quarter, falling from 49% to 26%.

Nebius doesn’t rely on liabilities as much as Coreweave does to fund its assets – liabilities of $15.1 billion support assets of $22.3 billion. Still, it needs its massive revenue backlog to convert into actual revenue to keep funding its own data– center investments.

If the big firms that Nebius has major contracts with – Meta Platforms (META) and Microsoft (MSFT), for example – decide to pull back, Nebius is going to feel the brunt.

5. Oracle (ORCL)

Yes, one hyperscaler made this list, too. It’s a long-shot bet, unlike other names here. While Oracle (ORCL) has a few other strong business lines, it’s investing in AI by rapidly increasing its borrowing, creating greater risk. As a result, this mega-cap stock has been surprisingly volatile.

A huge chunk of Oracle’s future revenue depends on OpenAI, too. Oracle signed a deal with the ChatGPT maker last year, under which OpenAI will spend $300 billion over five years. But OpenAI is already experiencing some issues growing its revenue and has even considered a price war with key rival Anthropic, as it works to meet a total of $1.4 trillion in spending commitments.

Oracle is borrowing money – $43 billion in fiscal 2026 – to build out its own AI infrastructure in anticipation of that OpenAI money rolling in. The thing is, Oracle now has more than $122 billion in long-term notes payable, and funding a build–out with more debt could prove troublesome.

So, the company announced that it wouldn’t issue more debt in calendar year 2026. Instead, the company announced that it will issue $20 billion in equity in 2026, followed by a $20 billion debt raise sometime in early 2027, an admission of the growing risk. Alphabet (GOOGL) has made a similar pivot toward issuing equity, as its debt issuance swelled to an all-time high.

Oracle is not quite as bulletproof as it might seem. Operating income came in at $20.6 billion in fiscal 2026, with interest expense eating up $4.6 billion of that, or 22.3%. More borrowing in 2027, AI revenue that fails to come in the door, and a broader economic downturn or bust may make Oracle’s income-expense ratio a lot narrower than it should be for a mega-cap stock.

Oracle is also on the sell list of financial-technology pioneer Marc Chaikin, who just explained why he’s letting go of this massive long-term winner. But Chaikin is offering his replacement stock pick poised for strong upside, free for viewers of his exclusive presentation.

Regards,

James Royal, PhD