AI Capex Is the New Macro Factor: J.P. Morgan's Warning on Diversification

The summer momentum unwind hit AI-linked equities hardest in July, and the bleed extended into August. That is not a sector rotation. It is a structural signal that AI capital expenditure has outgrown its status as a thematic trade and become a systemic market factor. Gabriela Santos, chief market strategist at J.P. Morgan Asset Management, told CNBC that finding true diversification from the AI trade is nearly impossible. Her team built an AI factor basket to test this. The results were uncomfortable: most assets now move in sync with the broader AI trade. The only genuine diversifiers left are Treasuries, gold, core real estate, and European equities. That is a short list. And it is getting shorter. This is not a bearish call on AI. Santos was explicit: you can be very, very bullish on AI and still need to think very, very carefully about portfolio construction. The problem is not the technology. The problem is that AI capex has become a shared risk factor across every asset class. When hyperscalers, chipmakers, and software companies all ride the same capital expenditure wave, their correlation converges. The old industry groupings no longer hold. AI infrastructure spending has grown so large that it touches nearly every asset class, from equities to fixed income to private markets. That is not a sector theme. That is a macro variable. Here is what the AI factor basket revealed. J.P. Morgan constructed a basket to measure exposure to the AI trade across assets. The finding: most assets move in tandem with it. The diversification that investors once took for granted—stocks versus bonds, tech versus industrials—has collapsed. The bond-equity correlation breakdown is particularly telling. When AI capex drives both earnings expectations and interest rate sensitivity, the traditional 60/40 portfolio loses its hedging power. Capital competition returns with inflation and rate volatility. Investors need additional inflation hedges, not just duration. Based on my audit experience across DeFi protocols and now traditional market structures, this pattern is familiar. In 2020, I watched DeFi yield farmers pile into Curve pools, treating LP rewards as risk-free yield. The token emissions were the subsidy. When emissions slowed, the users vanished. The same dynamic is playing out in AI equities. The capex is the subsidy. The revenue conversion is the question. If income growth lags capital expenditure, the valuation faces a double hit: lower earnings and higher discount rates. The contrarian angle here is not about avoiding AI. It is about recognizing that the AI trade has become a crowded macro factor, and the diversification window is closing. The assets that still offer genuine dispersion—gold, core real estate, European equities—are not just hedges. They are tail-risk insurance against an AI capex slowdown. But that window is not permanent. If AI capex continues to expand, these assets will eventually be repriced into the same factor. The question is not whether to be long or short AI. The question is whether your portfolio can survive a synchronized drawdown. What the market is missing is the internal dispersion within AI sectors. Hyperscalers, chipmakers, and software companies are no longer moving as a bloc. The AI buildout is changing shape. Some companies are winning on capital efficiency. Others are burning cash on capacity that may not generate returns for years. The old industry classifications are unreliable. Investors need new dimensions: compute self-sufficiency, AI revenue share, capex efficiency. The winners will be those who can convert capital expenditure into free cash flow fastest. The losers will be those who cannot. This is where the real risk sits. The AI trade is not a monolith. It is a collection of companies with vastly different capital allocation strategies. The market is treating them as one. That is the mispricing. The summer momentum unwind was a warning shot. The next one could be worse. The crowding is not just in equities. It is in options, in leveraged ETFs, in structured products. When the unwind comes, it will be fast. Static portfolios die slow. What should investors watch? The next earnings season will be critical. Hyperscaler capex guidance is the single most important signal. If guidance is cut, the entire AI supply chain reprices. If guidance is raised but revenue conversion disappoints, the same thing happens. The J.P. Morgan AI factor basket correlation data is another key metric. If correlations rise further, diversification becomes even harder. If they fall, the window reopens. Also watch the dispersion within chipmakers and software companies. Rising dispersion means the old groupings are failing, and stock-picking alpha is expanding. My take: the AI capex cycle is not ending. But the trade is maturing. The easy beta is gone. The next phase will be about capital efficiency, free cash flow, and competitive moats. The market is starting to realize that AI is not a sector. It is a macro factor. And macro factors do not diversify. They concentrate. The question is not whether you believe in AI. The question is whether your portfolio is built for a world where AI capex moves everything. If it is not, the summer unwind was just a preview. The real test is coming. Static is not a strategy. It is a risk.