The Receipts Problem

Hike odds collapsed, inflation cooled, yields fell — and the AI trade lost ground anyway. Wednesday night, Alphabet and Tesla have to show the receipts.

Something unusual happened last week. The macro case for owning long-duration growth assets improved substantially — inflation cooled, hike odds collapsed, yields fell — and long-duration growth assets fell about 3%. The reflex that has worked for fifteen years did not fire.

Understanding why is the whole job going into Wednesday night.

The rate picture genuinely changed

Per Bloomberg rate probabilities, the odds of a hike at the July FOMC fell to roughly 10% from 33% a week earlier. September moved down to about 60% from 86%. The first meeting carrying a theoretical certainty of a hike shifted back to December from October. Clearbrook Global noted hike odds collapsing from above 40% to below 17% in a single session on the CPI release.

That is a substantial dovish repricing, and it arrived alongside a soft labour market — June payrolls came in at 57,000 — which normally reinforces it. The July 28–29 FOMC is now shaping up as one of the more consequential meetings of Chair Warsh's short tenure. Nobody expects a move. Everybody is watching the language for whether June's hawkish dot plot survives contact with the new data.

So why didn't growth rally?

Because the thing being repriced is not rates. It is competitive position.

The standard framework says lower discount rates lift the present value of distant cash flows, so falling hike odds should be worth the most to exactly the assets that fell hardest. That logic holds only if the distant cash flows themselves are unchanged. Last week the market was handed a reason to doubt them: a credible frontier model from a Chinese startup, arriving in a month when AI infrastructure names were already roughly 20% off their highs.

The receipts problem

Analysts expect S&P 500 earnings to rise roughly 23% year-over-year this quarter, which would mark a second consecutive quarter of 20%-plus growth — a feat achieved only once in the past two decades, in 2018, and then only because earnings had not declined the prior year.

But the composition of that growth is fragile in two specific places.

The first is energy. Its outsized earnings contribution is a direct function of the war premium in crude, and it is definitionally temporary; it unwinds the moment supply normalises. The second is the AI complex, where the imbalance is more durable but no less real. Hyperscaler margins are already contracting under the weight of capital expenditure. That resolves in one of two directions: either monetisation catches up with spending, or spending slows — and if spending slows, semiconductor earnings growth cools with it.

Which means the earnings outlook from here rests on a narrower question than the headline growth rate suggests: how quickly can the hyperscalers show the receipts?

Wednesday night

Alphabet and Tesla both report Wednesday after the close, on what is the largest day of the week by market capitalisation.

Alphabet is the one that matters for the complex. Evercore's Mark Mahaney, who heads the firm's internet research, expects an acceleration in cloud growth and another lift to capital expenditure guidance. Alphabet has already raised full-year 2026 capex guidance to a range of $180–190 billion; Mahaney suggested next year's figure could reach roughly $300 billion, somewhere in a $275–325 billion band, while cautioning that the company may not put a number out at all. Investors will also want an explanation for the reported delay to its flagship model rollout.

The asymmetry here is worth stating plainly. If Alphabet delivers strong cloud growth and holds capex guidance, it removes the central objection that drove the Nasdaq down 2.9% last week, and the chip complex's 20% drawdown starts to look like a reset rather than a top. If either name disappoints materially, the rotation into energy, health care, staples and REITs that began last week accelerates.