In recent weeks, tech stocks have given up some of their extraordinary gains as investors have reassessed the prospects for the artificial intelligence (AI) revolution.
Is AI a bubble? The question is on everyone’s minds. Today, nearly 40% of the S&P 500’s market cap feels the direct impact of perceptions or realities related to AI usage, investment, infrastructure construction and productivity gains. Whether to the upside or to the downside, AI will almost certainly be the most important driver of public equity market returns over the next few years. To feel comfortable allocating capital to stocks, we must feel confident that we are not about to see a bubble burst.
Market and economic bubbles follow a consistent narrative. Most bubbles start because of an investor thesis that the world is changing—undergoing a paradigm shift. Believers build capacity to meet future demand. The bubble begins to form in part because credit is widely available. Decaying underwriting standards and increasing leverage cause a disconnect between economic fundamentals and market valuations. More and more investors join the crowd—until fundamentals finally prevail and the bubble bursts.
Once we’ve established a pattern to assess irrational exuberance, we can use it to evaluate the AI trade. Here is how we think AI now stacks up relative to five key elements:
Bubbles often emerge from an idea that a new technology, demographic trend or policy shift will profoundly change the world. Notable historical examples include the railroad boom in the 1840s and the internet boom in the late 1990s. Those transformations did indeed change the world, but timing matters. From 1843 to 1853, railway miles in the United Kingdom nearly quadrupled, but railway revenue per mile was flat to down. By mid-2001, telecom companies had installed 39 million miles of fiber, but only 10% of those fibers were lit, and each lit fiber was utilizing just 10% of the wavelengths available.1
Both the railroad and internet booms featured tremendous excess capacity—capacity that was not justified by concurrent consumer demand or unit economics. Today’s AI story certainly features the rhetoric and investment you would expect to see during a paradigm shift. But we do not yet see excess capacity. Data center vacancy rates are at a record low 1.6%, and three-quarters of data center capacity under construction are pre-leased.2 Across the computing, power and data center value chain, components are scarce relative to demand. And the latest earnings season confirms that AI use is driving revenue growth for the largest companies.
That said, we think the risk that a bubble will form in the future is greater than the risk that we may be at the height of one right now.
Bubbles expand because cheap, speculative capital drives prices ever higher. In the 17th century, Amsterdam’s deep credit markets fueled tulip-mania, while the Japanese asset bubble of the 1980s relied on bank loans collateralized by artificially inflated corporate equity values. The housing bubble that preceded the global financial crisis (GFC) was inflated by subprime mortgages, securitized in an interconnected “shadow banking” sector. In the 2010s, an energy stock bubble formed as oil producers accessed inexpensive financing made possible by policy rates pinned at zero.
Oracle’s recent foray into debt markets signals that the next phase of the AI infrastructure cycle will rely more on credit. The deal was 5x oversubscribed, and we think public markets will be willing to finance the largest tech companies, which all have tighter spreads than the broad investment grade index.3 As the Federal Reserve rate-cutting cycle progresses, it seems likely that credit will finance more AI investment. This could well happen, given low leverage in the large-cap equity space and over $500 billion in private credit dry powder.4
Bubbles typically expand as financial structures magnify gains and obscure risk. The South Sea bubble5 featured debt-for equity swaps; the pre-1929 crash years roared with margin buying. More recently, SPACs expanded via redemption puts and free warrants. In the AI arena, financial innovation and engineering are accelerating.
Among recent examples: Companies such as Lambda and CoreWeave have issued debt collateralized by their high-end GPUs,6 and Alibaba recently announced a zero-coupon convertible security to fund data center investment. In terms of financial engineering, technology sector debt and data center–related asset-backed and commercial mortgage–backed security issuance have bounced back to levels last seen in 2020 and 2021.7 But these are relatively straightforward features of capital markets. If the hyperscalers decided to lever their balance sheets to 2.8x net debt to EBITDA (the median for an investment grade company), it could result in an additional $1 trillion of capital to spend.
One could also argue that the “circular” investments from the AI supply chain could be an example of financial engineering. These deals (in which key industry players buy and sell from one another using equity and computing power as currency) certainly increase risk. But they could also create a more symbiotic ecosystem with more competition for both hardware and software that could lead to a more balanced landscape.
We are searching for signs that underwriting standards are deteriorating, whether for power purchase agreements or for private equity and venture investments. To date, aggregate cash flows from operations still exceed capital expenditures and dividends for the major players. Leverage will likely grow as AI investment continues, but AI spending today is fueled by cash flows.
In every bubble, valuations increase beyond what fundamentals, cash flows or use cases alone would justify. During the dot-com bubble, companies went public with no revenues. Cisco’s stock price increased by 40x from 1995 to 2000, while its earnings grew by just 8x. Today, we are seeing pockets of froth in private markets. Unicorns—private companies with greater than $1 billion in market cap—are now worth nearly 12% of the Nasdaq; that share is close to its peak in 2021.8 And the valuation growth of AI startups has consistently outpaced that of non-AI companies across every series. For example, the median Series B step-up is 2.1x for AI startups versus 1.4x for non-AI startups. AI companies command median valuations that are 56% higher at Series C and 230% higher at Series D+ than non-AI companies.9
But in the public markets, AI companies have generated their returns entirely through earnings growth. Over the last three years, the forward price-to-earnings (P/E) multiple of publicly traded AI stocks has declined, while earnings per share (EPS) estimates have more than doubled. Over the past five years, Nvidia’s stock price increased 14x, while earnings grew 20x.
Every bubble attracts new participants convinced that rising prices are a self-fulfilling prophecy. Dutch artisans bought tulip bulbs for multiples of their annual incomes, and Las Vegas bartenders flipped houses in 2005. Recent IPO performance suggests more signs of froth. Exuberance is building, but it would need to reach much higher levels before we would grow more cautious.
When we consider the evidence, it seems clear that the ingredients for a market bubble are present. That said, we think the risk that a bubble will form in the future is greater than the risk that we may be at the height of one right now.
Moving past the AI bubble debate, here’s the more important question for investors to ask: Who will ultimately capture the value from this technological transition? Unfortunately, history provides no clear pattern for which companies will ultimately capture the value of technological transitions.
In some instances, such as U.K. railways, fiber optic cables and telecoms, the first movers suffered painful drawdowns only to see new entrants capitalize when asset prices had collapsed. On the other hand, first movers in the information technology transition (e.g., IBM, Microsoft, Cisco and Amazon) were able to capture and retain market share even as other entrants capitalized on the ecosystem that developed. U.S. electric utilities maintained market share, but regulations ended up curtailing the ultimate return to investors.
No one knows who will ultimately capture the value from the AI revolution. Investors will want to avoid the risks of technological obsolescence and “irrational” exuberance surrounding AI. But the biggest risk, to us, is not having exposure to this transformational technology.
We can help you navigate a complex financial landscape. Reach out today to learn how.
Contact usInvestment Strategy
LEARN MORE About Our Firm and Investment Professionals Through FINRA BrokerCheck
To learn more about J.P. Morgan’s investment business, including our accounts, products and services, as well as our relationship with you, please review our J.P. Morgan Securities LLC Form CRS and Guide to Investment Services and Brokerage Products.
JPMorgan Chase Bank, N.A. and its affiliates (collectively "JPMCB") offer investment products, which may include bank-managed accounts and custody, as part of its trust and fiduciary services. Other investment products and services, such as brokerage and advisory accounts, are offered through J.P. Morgan Securities LLC ("JPMS"), a member of FINRA and SIPC. Insurance products are made available through Chase Insurance Agency, Inc. (CIA), a licensed insurance agency, doing business as Chase Insurance Agency Services, Inc. in Florida. JPMCB, JPMS and CIA are affiliated companies under the common control of JPMorgan Chase & Co. Products not available in all states.
Please read the Legal Disclaimer for J.P. Morgan Private Bank regional affiliates and other important information in conjunction with these pages.
Bank deposit products, such as checking, savings and bank lending and related services are offered by JPMorgan Chase Bank, N.A. Member FDIC.
Not a commitment to lend. All extensions of credit are subject to credit approval.
