JPMorgan Strategist: The AI Rally Is Expanding Far Beyond the Magnificent Seven

JPMorgan’s strategist argues the AI-led rally is broadening beyond the Mag-7, with infrastructure/backlog momentum driving new leadership. Quanta Services reported Q2 adj. EPS of $4.24 vs $3.03 (+39.93%), revenue of $9.56B (+41.1% YoY), and record backlog of $53.44B, while GE Vernova posted orders of $24.2B (+88% organically) and backlog of $176B; Cisco booked $4B AI infrastructure orders in Q4 and guided FY2027 AI infrastructure revenue of $7.5B. Walmart also beat with Q2 FY27 adj. EPS of $0.81 vs $0.74 and gross margin up 96 bps, supporting the view that AI spend is translating into wider, profitable demand.

Analysis

This is less an AI trade than a capacity trade: the scarce asset is now kilowatts, interconnects, switchgear, and field labor, so the economic rent accrues to companies that can turn constrained infrastructure into priced backlog. That favors PWR, GEV, and ETN over software vendors whose AI spend is still largely an expense line; the second-order effect is margin expansion for the bottleneck owners and margin pressure for hyperscalers if power and equipment inflation stays sticky. The same dynamic should spill into adjacent names like EMR, HUBB, NVT, and STRL as the supply chain reprices lead times rather than end-demand.

The near-term catalyst is not “AI adoption” but whether backlog converts into revenue without gross margin leakage over the next 1-3 quarters. The main risk is a classic capex air pocket: if a few hyperscalers pause budgets, order books can remain full while revenue growth decelerates because delivery, permitting, or labor constraints push recognition out 2-4 quarters. That is the key falsifier—book-to-bill rolling over, pricing power fading, or management guiding to slower conversion despite record order intake.

The consensus may be underweighting how cyclical this can still be: infrastructure stocks can look secular until they’re valued like utilities, then any normalization in lead times compresses multiples fast. I would not chase the group indiscriminately here; the cleaner expression is to own the best balance sheets and pricing power while hedging market beta. WMT and JPM are more confirmation than alpha: if consumer and credit stay resilient, they validate the broader-risk backdrop, but they are not the purest way to express the AI infrastructure thesis.

Key Decisions for Investors

• Long PWR and ETN on any 3-5% pullback; use a 3-6 month horizon and require continued backlog growth plus stable margins as the setup remains intact.
• Long GEV vs short XLK as a relative-value hedge: express the view that the AI dollar is migrating from software narratives into physical bottlenecks; cover if GEV order growth or gross margin disappoints next quarter.
• Add a watchlist alert on CSCO rather than a fresh initiation: if AI infrastructure orders reaccelerate and the market starts paying for recurring networking demand, CSCO can re-rate further, but much of the obvious move is already in the stock.
• Avoid chasing the whole basket after strength; if you want a tighter risk/reward, wait for any sign of slower backlog conversion in PWR/GEV before fading the group, because the thesis breaks only when orders stop turning into revenue.