AI Crash Endgame: Bubble Could Burst in Early 2028, Tech Stocks May Plunge 40%

Global markets are currently concentrating their bets on an unprecedented infrastructure build-out—the AI investment boom. Yet the underlying logic of this boom is becoming extremely fragile: current profits rely mainly on internal "capex recycling" and debt financing among tech giants, rather than real payments from external end users.

TS Lombard Chief Macro Strategist Dario Perkins said in a research note on Oct. 22 that, under the ultimate scenario, this AI capex frenzy is highly likely to end in early 2028, triggering a violent market reset.

The report states that the S&P 500 will fall 30%, tech stocks will plunge 40%, about $20 trillion in market value will evaporate, and the tech weighting in the U.S. stock market will return overnight to levels seen before ChatGPT's launch. Although tech giants can survive thanks to strong balance sheets, weaker players upstream in the AI ecosystem (such as emerging cloud providers and independent model developers) will face a wave of defaults, with estimated credit losses of $125 billion to $190 billion.

TS Lombard believes the macro economy will slip into a mild recession: the sharp drop in AI capex will reduce U.S. GDP by about 2 percentage points and eliminate roughly 800,000 data center construction-related jobs. But because private sector balance sheets are healthy and the Federal Reserve has ample room to cut rates, this will not turn into a systemic financial crisis like 2008.

AI Investment Boom: The Largest Infrastructure Gamble in History, Sustainability in Doubt

AI capex has become one of the largest infrastructure projects in human history. U.S. tech companies are spending about $1 trillion annually on AI build-out, and cumulative investment in super data centers has already surpassed the Manhattan Project, the Apollo moon landing program, and even the Marshall Plan. Hyperscaler capex as a share of U.S. GDP is near historic peaks.

However, the foundation of this boom is not solid. Current revenue growth in the AI ecosystem is still largely a product of "capex recycling"—one company's capital expenditure is another company's revenue. As long as hyperscalers keep spending massively, revenue within the ecosystem will keep growing, but that does not represent true sustainability.

The real marker of sustainability should be a significant increase in the share of revenue coming from outside the ecosystem—ordinary consumers and non-tech corporate users. But as of the report's publication, that share has barely improved. AI monetization remains the biggest open question: OpenAI and Anthropic derive about 80% of their revenue from just 1% of customers, and most of those customers are loss-making AI startups that rely on continued venture capital funding.

The report says the current AI boom is essentially a "capex recycling" loop—as long as tech giants keep injecting money into the system, revenue will grow. But as debt becomes the main source of marginal financing, this model is highly vulnerable to a fatal blow from a slowdown in capex.

Six Crash Triggers: Any One Could Set Off a Chain Reaction

The report lays out six potential risk points, arguing that an AI crash is already in an "overdetermined" state—any one of the following six triggers could be the blade that pops the bubble.

Trigger 1: Revenue Massively Misses Expectations

Multiple studies show that for current capex to earn a reasonable return, the AI industry would need annual revenue of $2.5 trillion to $3 trillion by 2030, roughly 14 times current levels. That growth rate is almost impossible to achieve in such a short time, especially as Chinese open-source (open-weight) models rapidly seize market share. Ordinary consumers are broadly resistant to paying for AI services, and AI services face commoditization risk.

Trigger 2: Credit Risk Rises Sharply

Hyperscalers previously relied mainly on their own cash flow for investment, but debt has become the marginal source of financing for AI build-out. In 2027 alone, AI companies are expected to issue about $700 billion in new bonds, of which hyperscalers account for $400 billion and the rest comes from other AI ecosystem participants—a figure that exceeds the total bond issuance of all eurozone governments.

Even more worrying is the rapid expansion of "off-balance-sheet liabilities." Take Meta's Hyperion data center in Louisiana: a $30 billion investment project in which Meta sold 80% of the equity to Brown Owl, forming a joint venture called Beignet that raised $27 billion in new debt, guaranteed by Meta (which committed to leasing the data center for 20 years). Because the lease was split into five four-year leases, only a tiny fraction of the leverage appears on Meta's balance sheet. Hidden liabilities across the AI ecosystem are estimated at $1.5 trillion to $2 trillion. Some emerging cloud providers (such as CoreWeave) are already paying financing rates as high as 15%.

Trigger 3: The "Teaser Rate" Trap and the Depreciation Cliff

The report compares the current AI financial structure to the "teaser rates" of the subprime crisis in the 2000s. Data centers do not need to book depreciation during construction, but once they come online, depreciation pressure will suddenly intensify. At the same time, multi-year data center leases signed by AI labs are not counted as costs until the data centers are operational. Once more data centers come online, the financial position of AI companies could fundamentally change within 12 months.

Trigger 4: IPO Wave Reveals the Truth

Anthropic and OpenAI both plan IPOs (the latter has been postponed to 2027). Historical experience shows that IPO waves often mark cycle tops. More importantly, IPOs will for the first time give outsiders a window into the true economics of the AI industry—currently disclosed metrics (such as "annual recurring revenue" or ARR) are clearly embellished, and there is a great deal of opacity.

Trigger 5: Semiconductor Cycle and Supply Chain Risk

NVIDIA is the "picks and shovels" of AI build-out and the barometer of the entire AI boom. NVIDIA is engaging in large-scale "vendor financing," lending to customers so they can buy more chips—what the industry calls "balance sheet as a service." According to the latest research from the Bank for International Settlements (BIS), more than 55% of investment flowing into AI companies comes from other AI companies, a clear sign of circular financing. In addition, semiconductor demand is highly cyclical and subject to the "bullwhip effect"—once demand slows, today's hardware shortage could quickly turn into a severe glut.

Trigger 6: Regulatory and Safety Risk

Public perception of AI technology continues to deteriorate. About 55% of Americans want to slow the pace of AI development, while only 13% oppose it, and this holds across major political groups. According to the 2026 IBM/Ponemon Cost of a Data Breach Report, one in four malicious corporate data breaches globally involved some form of AI-assisted attack. A major safety incident could trigger a strong political backlash and tighter regulation.

The "Second Derivative Problem": Slowing Growth Is Enough to Detonate the Crisis

The report places special emphasis on the "second derivative problem" of the AI boom: AI capex does not need to fall in absolute terms—merely slowing growth is enough to trigger financial stress.

This is exactly the logic of the subprime crisis in the 2000s—what triggered the collapse then was not falling house prices, but house prices ceasing to rise rapidly. In AI build-out, frontier labs sign contracts committing to pay billions of dollars to data center operators, who use those contracts as collateral to borrow and build data centers. But these labs themselves have no profits, and their ability to keep paying depends entirely on continued fundraising. Hyperscaler capex is like a loan book: once it slows, the whole boom unravels.

Quantifying the Market Shock: $20 Trillion in Wealth Destroyed, Credit Losses Manageable

Tech stocks currently account for as much as 40% of U.S. stock market capitalization, with semiconductors alone close to 20%. The U.S. tech sector currently trades at about 30 times earnings (down from 40 times a year ago), far below the 70-75 times peak during the dot-com bubble. Under the AI crash scenario, the report forecasts:

The S&P 500 falls 30% overall; tech stocks fall 40%; other sectors fall about 20%, for a combined market value loss of about $20 trillion.

The report says these estimates pull the tech weighting in the U.S. market back to levels seen before ChatGPT's launch.

By the end of 2027, the non-hyperscaler AI ecosystem is expected to have about $545 billion in outstanding bonds, $920 billion in bank loans, and about $650 billion in private credit. Total AI-related debt is roughly $2.9 trillion, of which about 25% is exposed to risk.

The report assumes a cumulative default rate of 35% over 2028-2030 and recovery rates of 25%-50%, yielding overall credit losses of $125 billion to $190 billion. While sizable, this does not pose a systemic threat to the overall U.S. financial system. Equity holders will bear the largest losses (about $20 trillion), but there will be no 2008-style cascade of credit losses and widespread bankruptcies.

The most vulnerable links include: emerging cloud providers (neoclouds), Oracle, independent model developers (OpenAI, Anthropic), and construction/equipment suppliers.

Mild Recession, Similar to 1990 or 2001

By early 2028, U.S. tech-related capex will have risen by about 3 percentage points relative to 2023 (as a share of real GDP). The report assumes a full reversal, but because data center investment is highly import-dependent, falling imports will partially offset the GDP hit from lower capex. Taking all effects together, the end of the AI data center boom will reduce U.S. GDP by about 2 percentage points (the IMF's latest World Economic Outlook arrives at almost the same estimate).

The report says that, overall, the severity of this recession will be comparable to 1990 or 2001—a mild recession by historical standards—and gives three reasons:

First, private sector balance sheets are healthy. Private credit growth is less than 10 percentage points of GDP, more similar to the late 1980s and far below pre-subprime-crisis levels, so there is no risk of a systemic credit event.

Second, monetary policy ammunition is ample. The Federal Reserve has rebuilt sufficient monetary policy space for the first time since the global financial crisis, with plenty of room to cut rates without relying on large-scale fiscal stimulus.

Third, rate cuts will activate the bottom of the "K-shaped economy." Low-income groups, consumer durable goods demand, and the residential real estate market (where there is substantial pent-up demand) that have long been suppressed by high inflation and tight monetary policy will benefit directly from rate cuts. Historically, after the dot-com bubble burst, the Fed's easing quickly spawned a new bubble in the housing market, and a similar scenario could repeat. Just as the Fed's easing after the dot-com bust fueled a housing boom, rate cuts will quickly stimulate growth outside AI—and that is the core reason the recession will remain mild.

On the labor market, the report notes that the AI build-out boom has created about 500,000 jobs so far, expected to rise to 800,000 by the end of 2027. If all those jobs disappeared, the U.S. unemployment rate would rise by about 50 basis points, without considering the broader impact on business confidence.

The report points out that the biggest tail risk is that an AI crash could actually accelerate corporate adoption of AI automation to cut costs. History shows that the large-scale push for internet-related automation happened during the 2009 recession, not during the boom. If an AI crash triggers a wave of automation, the labor market shock could exceed expectations, leading to a slow "jobless recovery."

How to Preserve Assets in a Bubble

The report says an AI crash is a classic demand shock, not a 1970s-style supply shock (stagflationary recession). Long-term interest rates will fall significantly, and the correlation between bonds and stocks will turn negative, favoring the safe-haven properties of Treasuries. But because the recession is mild, there will be no return to zero rates or permanent quantitative easing.

The key difference between the AI boom and the dot-com bubble is that this tech boom is highly concentrated in a handful of U.S. and Asian companies, whereas the dot-com bubble involved broad participation across developed markets, including Europe. South Korea, as the core of the AI supply chain, has seen tech exports as a share of GDP rise by 10 to 30 percentage points over the past 12 months and will be hit hard in a crash.

But markets like Europe that did not deeply participate in the AI boom will significantly outperform the U.S. after the initial panic selling. These economies bore the negative effects of the AI boom (high inflation, high interest rates) without fully enjoying its benefits, so the end of the AI boom may actually bring relief.

The report argues that the impact of an AI crash on the dollar will be far greater than the dot-com bubble and the subprime crisis. Because the root of the crisis lies in the U.S. financial system, capital will flow to other currencies such as the euro, the yen, and even the pound. Global investors will not need to rush into U.S. Treasuries for safety; their own domestic bonds can serve the same purpose.

The report explicitly states that shorting U.S. tech stocks or significantly extending duration is not advisable at this stage—if the AI boom still has several years to run, premature positioning will lead to severe underperformance.

The optimal strategy follows the "easy bubble" trading framework proposed by GMO's Ben Inker: hold a diversified global equity portfolio that benefits if the boom continues and outperforms concentrated bets on U.S. tech stocks in a crash. Only gradually add government bonds when clear signs of stress appear in the AI ecosystem.

TS Lombard's scenario report sends a sober, not doomsday, message: the AI bubble bursting will not be 2008, but more like 2001. The U.S. economy will experience a mild recession, the financial system will not collapse, and the Federal Reserve has ample tools to respond.

But for investors, $20 trillion in wealth destruction, a 40% drop in tech stocks, and severe shocks to AI supply chain countries and regions are enough to deal a devastating blow to any portfolio concentrated in AI.