Michael Hartnett: I coined “Mag7” - Here’s What Ends Their Run

Michael Hartnett says he coined “Magnificent Seven” for companies that generated substantial cash while spending relatively little. That profile has changed: he expects the Mag Seven, or hyperscalers, to spend $1.2 trillion on capital expenditure next year, up from $100 billion three or four years ago, while free cash flow moves from about $300 billion to a projected negative $200 billion. The AI 10 now make up roughly 40% of the S&P 500, a concentration he calls bubble-like but not proof that stocks cannot rise further. He compares the market with past peaks in technology, the Nifty Fifty and Japan, while noting railroads once reached about 63% of the index.

The central financial risk to AI investment is the bond market. Higher yields, wider credit spreads and borrowing costs could constrain hyperscalers’ spending; Hartnett warns that a disorderly move in bonds could make Mag Seven stocks and related assets dangerous. He sees AI and “anything but bonds” as linked trends, and says bonds have ended bubbles, booms and bull markets before. A 5% Treasury yield is not a magical threshold, but conditions could feel too tight as nominal yields approach 5%, real yields approach 3% and the curve nears inversion. Yield increases have already hurt REITs, banks, duration stocks and small caps, even as US equities stayed near records because earnings have not yet been compromised. By late September, more than half of S&P 500 companies were below their 200-day moving averages.

Hartnett does not expect five- or ten-year Treasury yields to surge through 5% toward 6%, and says investors have not concluded that bonds are at a major secular turning point. He expects yields to stabilize in the coming months but remains alert to a policy error or system break that could trigger a disorderly rise and a regime change. He questions whether 2026 will bring such a change for bonds or AI. A credible central bank is needed to anchor long-term yields, although investors doubt policymakers would raise rates into a recession to contain inflation. Hartnett is currently long equities and short bonds, while watching for one final rise in yields; if yields hold around 5%, he sees potential for small caps, real estate and some duration assets. Long bonds are his biggest contrarian call, followed by long China.

The other major threat to AI’s run is political resistance. Hartnett likens the railroad boom’s end to a period of inequality, job losses, social upheaval and worker strife, while saying conditions today have not reached that level. He frames bonds as the “Wall Street” threat and voters as the “Main Street” threat, with voter responses to AI harder for officials to control. Opposition to data-center expansion is emerging in Texas, where the Democratic candidate for governor has made a halt to expansion a central issue and the race is described as neck and neck. Governor Abbott, a champion of data centers, has proposed a moratorium or pause as his polling has weakened, though Hartnett doubts Texas will turn Democratic. He cites a 60% Polymarket probability of a broad Democratic win in November, but much lower odds of Abbott losing. A major Democratic victory could make data centers a campaign target and trigger a substantial reversal in big tech and the dollar.

Beyond the Mag Seven, market gains have broadened mainly into banks and healthcare, which Hartnett characterizes as AI adopters with large technology budgets, alongside semiconductor and data-center builders. Still, he expects the next equity move to become more concentrated and “AI only,” partly depending on developments at the long end of the bond market. Investors are long equities, hold no bonds and have favored broader participation; yields holding at 5% could renew incentives for AI. Hartnett favors small caps over the longer term, especially small-cap technology firms able to use AI to grow, but acknowledges that recent yield rises have hit them hard. He also sees potential for policies responding to voter anger to benefit smaller companies rather than directing all gains to large firms.

Hartnett expects commodities to remain in a secular bull market, supported by AI demand and scarce resources, and expects energy to keep outperforming. Energy is more exposed than gold or industrial metals to geopolitics: peace deals could pull prices down, while lasting damage to Russian or Middle Eastern supply could support them, and oil could reach $200 before falling. Gold, which he views as a hedge against populism, anarchy and currency debasement, has shed frothy positioning; he considers a move above $5,000 before the midterms unlikely but expects it to return above that level. He also favors international equities, including European defense, Asian technology, Chinese AI and Latin American resources, and sees scope for diversification as fiscal spending grows outside the US. He expects better relative performance from non-US markets over five years, not a dramatic separation from still-correlated US stocks. His longer-term rule is to “buy humiliation and sell hubris”: consider neglected China, bonds, commodities and real estate, and take some profits in popular US equities, AI and technology.