‘Blood in the streets’: Legendary investor Jeremy Grantham pulls back the curtain on the AI wars to reveal a ‘brutal, competitive world’

Jeremy Grantham argues AI is shifting markets from a monopoly environment to a brutal competitive one, potentially capping aggregate profit margins despite $725 billion of planned capex from Amazon, Google, Meta, and Microsoft. He also flagged GMO's emerging market fund as up about 70% over 12 months versus roughly 25% for the S&P 500, implying continued mean reversion away from expensive U.S. mega-cap tech. The piece is mainly a bearish macro/valuation warning rather than a direct catalyst.

Analysis

The key second-order implication is not simply that AI spending is large, but that it is likely to be margin-destructive for the very cohort investors assume will monetize it best. When four hyperscalers simultaneously spend at scale, the economic rent from AI infrastructure migrates upstream to chips, power, cooling, networking, and datacenter REITs, while downstream software/platform owners face faster feature parity and weaker pricing power. In other words, AI may be a capex transfer mechanism more than an earnings compounding engine for the largest index weights.

That creates an important cross-asset setup: the near-term support from elevated capex can persist for quarters, but the valuation multiple expansion thesis becomes more fragile as markets start discounting lower incremental ROIC and rising depreciation drag. If AI adoption turns from scarcity to ubiquity over the next 12-24 months, the winners should diversify away from the “platform monopoly” bucket into picks-and-shovels, utility capacity, and non-U.S. cyclicals that benefit from U.S. mega-cap capex without bearing the full valuation risk. The market is still pricing an unusually clean translation from spend to profit, and that is where the asymmetry lies.

The more interesting contrarian read is that the consensus may be underestimating how long the spending boom can postpone a broader growth slowdown, even if it doesn’t improve aggregate margins. That argues against chasing an immediate short on the entire group; the better expression is relative value and convex downside after a valuation catalyst, not outright timing. A recession scare, capex guidance reset, or evidence of AI monetization plateauing would be the most obvious triggers, but absent that, this is a multi-quarter setup rather than a days-long trade.

Emerging markets deserve attention not just as a mean-reversion trade, but because lower U.S. mega-cap leadership would mechanically force global allocators to rebalance into cheaper international beta. If the Mag 7 cease to be the only earnings-growth refuge, EM and ex-U.S. quality cyclicals should capture incremental flows quickly. The risk is that AI capex remains a macro prop longer than expected; the reward is that once the market doubts margin durability, crowded positioning can unwind faster than fundamentals deteriorate.

Key Decisions for Investors

• Pair trade: short QQQ / long EEM or IEMG on a 3-6 month horizon. The trade expresses mean reversion out of concentrated U.S. mega-cap duration into cheaper international beta; target 8-12% relative return, cut if Mag 7 capex guide is revised higher and leadership broadens again.
• Rotate part of AI exposure from platform equities into picks-and-shovels: long AMT or EQIX against a basket short of AMZN, GOOGL, META, MSFT. The thesis is that infrastructure owners capture the spend while margin compression eventually hits the spenders; review after next two earnings cycles.
• Buy medium-dated downside convexity in QQQ or MAGS, preferably 6-9 month puts financed by selling upside calls. This is best viewed as an event-driven hedge against an AI monetization miss or capex fatigue; aim for 3:1 payoff if valuation de-rates 10-15%.
• Add to EM quality/cyclicals on weakness rather than chasing U.S. tech strength. Prefer selective exposure in India/Taiwan/Latin America proxies where cheaper valuations and weaker positioning can attract reallocation if U.S. tech breadth narrows further.
• Avoid fresh longs in the biggest hyperscalers until there is evidence that incremental AI dollars are producing measurable operating leverage. The risk/reward is poor here because the market already capitalizes future margin expansion that may instead be spent away.