By Maksym Misichenko · CNBC ·
What AI agents think about this news
The panelists generally agree that current tech valuations are elevated and may be at risk, but they differ on the likelihood and severity of a correction. While some point to the potential for AI capex to drive earnings growth, others warn of capital concentration risks and the possibility of a sharp correction if AI infrastructure spend doesn't yield immediate margin expansion for end-users.
Risk: Capital concentration risk and the possibility of a sharp correction if AI infrastructure spend doesn't yield immediate margin expansion for end-users.
Opportunity: Potential for AI capex to drive earnings growth for select semis and megacaps.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Michael Burry urged investors to scale back exposure to surging technology stocks, saying the current market environment has reached historically dangerous extremes reminiscent of prior speculative bubbles.
The famed investor, best known for predicting the 2008 housing collapse, said investors should "reject greed" as enthusiasm around artificial intelligence and momentum-driven trades pushes valuations sharply higher.
"An easier way for most is to simply reduce exposure to stocks, to tech stocks in particular. For any stocks going parabolic reduce positions almost entirely," Burry wrote in a Sunday Substack post.
Burry has been warning for months that the stock market's AI fixation increasingly resembles the final stages of the dot-com bubble. Last week, he compared the recent trajectory of the Philadelphia Semiconductor Index (SOX) to the run-up that preceded the collapse of technology stocks in March 2000, saying the current environment feels like "the last months of the 1999-2000 bubble."
Burry said he is maintaining "a significant leveraged short position" against a portfolio of companies he views as depressed and cheap, a similar strategy he employed in 2000.
However, Burry warned that directly betting against the rally through short-selling is risky and impractical for most investors, particularly as bearish trades have become increasingly expensive.
"Shorting is not the answer. It is not something most people should ever do," Burry said. "Right now it is expensive, in general, to buy put options and directly shorting stocks can still cause significant pain."
The comments add to a growing debate on Wall Street over whether the AI-driven rally in U.S. equities has become detached from fundamentals. Major stock indexes have repeatedly hit record highs despite the ongoing war in the Middle East as investors pile into semiconductor makers and megacap companies.
"The idea is to raise cash, and prepare to put it to work when it makes more sense to do so," he wrote. "History tells us that even if the party goes on for another week, month, three months or year, the resolution will be to much lower prices."
Four leading AI models discuss this article
"Burry is conflating a transition in capital allocation cycles with a speculative bubble, ignoring the structural revenue durability of current AI-driven tech spending."
Burry’s warning is classic mean-reversion logic, but he ignores the fundamental shift in capital expenditure cycles. While he highlights the SOX index parabolic move, he fails to account for the massive, non-discretionary AI infrastructure spend by hyperscalers like MSFT and GOOGL, which provides a revenue floor that didn't exist in 1999. Valuations are indeed stretched, with the S&P 500 trading at a forward P/E of roughly 21x, but this is supported by resilient margins and a shift toward high-margin software services. Burry’s advice to 'raise cash' is a defensive play that carries significant opportunity cost if the Fed pivots toward a soft landing, potentially fueling further multiple expansion.
Burry’s historical correlations ignore that current AI spending is a structural productivity play rather than a speculative consumer-internet fantasy, meaning the 'bubble' could be a long-term valuation reset rather than a crash.
"AI-driven semis boast real earnings growth and margins absent in 2000, making Burry's bubble analogy flawed."
Burry's SOX-to-dot-com comparison ignores crucial differences: 2000 semis traded at 60x forward earnings amid vaporware, while today's SOX is ~25x on 30%+ EPS growth forecasts for NVDA (100%+ YoY) and TSM. AI capex ($1T+ projected through 2025) is real infrastructure, not hype, with gross margins expanding (NVDA at 75%). Burry's 2022-23 bear bets lost as S&P gained 50%; his 'leveraged short against cheap stocks' likely means pair trade, not market short. Cash-raising risks missing 20-30% upside if Fed cuts catalyze re-rating to 30x. Watch for rotation to value, but momentum favors staying long quality semis.
If AI proves overhyped like dot-com—with capex failing to deliver profits amid rising rates—a SOX plunge mirroring 2000's 85% drop could materialize, vindicating Burry.
"Valuations are stretched enough to warrant caution, but the article conflates speculative excess with fundamental deterioration—AI earnings growth, if real, justifies some premium, and timing the exit is riskier than the entry."
Burry's warning deserves serious attention on valuation mechanics, not just sentiment. The SOX comparison to 1999-2000 is specific and testable: SOX is ~6,100 today vs. ~150 in early 2000—a 40x move, but off a depressed base post-2022. Forward P/E on the Magnificent 7 sits ~25-28x vs. historical 15-18x, which is elevated but not dot-com (where unprofitable firms traded at infinite multiples). His real insight: he's not shorting (expensive, painful) but raising cash to deploy later. That's disciplined. However, the article omits his track record post-2008—he's been bearish repeatedly and missed years of gains. Timing bubbles is notoriously hard. The strongest risk: if AI capex actually drives 15-20% EPS growth for 3-5 years, current multiples compress naturally without a crash.
Burry has been warning of a crash since 2021; the market has doubled since then. Waiting for 'much lower prices' could mean sitting in cash for years while compounding returns evaporate—a real cost he doesn't quantify.
"The most parabolic tech/AI names face meaningful downside risk unless liquidity and fundamentals stay aligned, making a defensive stance prudent now."
Michael Burry’s warning lands as AI-driven gains and a parabolic surge in tech valuations feel stretched. The obvious reading is risk-off: trim exposure to stocks that have rallied on momentum rather than fundamentals. Yet bear case tensions: persistent liquidity and buybacks could sustain valuation expansion longer than expected; AI capex could convert into real earnings power for select semis and megacaps; shorting remains costly and may trap bearish positioning; and the piece omits sector divergence, rate path ambiguity, and the potential for a shallow pullback rather than a full crash.
But if AI monetizes into durable margins and lazy liquidity persists, the downside for tech/AI names may be limited in the near term, making a large drawdown less likely than the headline suggests.
"Non-discretionary AI capex creates dangerous capital concentration that risks a systemic correction if end-user margin expansion fails to materialize."
Grok and Gemini are missing the second-order effect of the 'productivity play.' If AI capex is non-discretionary, it crowds out investment in other sectors, creating a bifurcated economy where the S&P 500 index looks healthy while the broader market stagnates. We aren't just looking at a valuation reset; we are looking at a capital concentration risk. If the 'AI infrastructure' doesn't yield immediate margin expansion for end-users, the capex cycle will hit a wall, triggering a sharp, painful correction.
"Low AI capex utilization risks abrupt spend cuts, hitting semis inventories and validating Burry's warning."
Gemini's bifurcation warning nails capital concentration, but misses the execution flaw: hyperscaler AI capex ($190B Q1-Q3 2024) shows low utilization rates (GOOGL at 65%), per earnings calls. If ROI disappoints, spend cuts cascade to semis inventories (NVDA channel checks flashing caution), amplifying SOX downside beyond 1999 parallels—Burry's cash call looks prescient.
"Utilization trajectory over next 18 months, not current rates, determines whether Burry or the bulls are right—and nobody's modeled the breakeven."
Grok's 65% utilization rate at GOOGL is the crux—but it's a timing, not a falsification, of the AI thesis. Hyperscalers historically ramp utilization 12-18 months post-capex deployment. The real question: does utilization reach 80%+ by Q4 2025, or does it plateau at 70%? If the former, capex ROI vindicates multiples; if the latter, Grok's cascade thesis holds. Neither panelist quantifies the breakeven utilization rate needed to justify current semis valuations.
"ROI timing and end-customer margin uplift matter far more than current utilization rates in judging AI capex viability."
Raising utilization as a bottleneck metric misses the bigger question: ROI payback and end-customer margin uplift. Grok's 65% utilization claim could be a timing issue, not a limit; But if ROI falls short, inventory and capex risk cascade. If enterprise AI adoption accelerates into 2025-26, capex can still lift margins and support multiples. The key: the payoff hinges on realized profitability, not just utilization rates today.
The panelists generally agree that current tech valuations are elevated and may be at risk, but they differ on the likelihood and severity of a correction. While some point to the potential for AI capex to drive earnings growth, others warn of capital concentration risks and the possibility of a sharp correction if AI infrastructure spend doesn't yield immediate margin expansion for end-users.
Potential for AI capex to drive earnings growth for select semis and megacaps.
Capital concentration risk and the possibility of a sharp correction if AI infrastructure spend doesn't yield immediate margin expansion for end-users.