SAT, OCTOBER 24, 2026
Independent · In‑Depth · Practitioner‑Tested
✎ News

The AI Trade Is Splitting Between Capex Promises and Actual Earnings

The Nasdaq Composite dropped more than 600 points on 22 October, its worst single session since 5 June, as investors weighed enormous AI capital commitments against near-term earnings. Lam Research fell 10.3% to $234.08 despite beating on earnings, marking its thirteenth loss in fourteen sessions.

By AIToolsRecap October 25, 2026 4 min read 30 views
Home › Articles › News › AI Capex Doubts Hand the Nasdaq Its Worst Day S...

The Nasdaq Composite fell more than 600 points on 22 October, its worst single session since 5 June, with the Dow and S&P 500 firmly down and the VIX rising after a three-day lull. Lam Research dropped 10.3% to $234.08 on a disappointing December outlook despite beating on earnings - its thirteenth loss in fourteen sessions. Meta weighed on large-cap tech and CoreWeave dragged the neocloud names lower.

The Lam Research reaction is the one worth reading closely, because beating on earnings and falling 10% means the market is no longer pricing the quarter. It is pricing the forward order book, and a weak December outlook from an equipment supplier is a statement about whether the announced buildouts actually convert into purchase orders.

The context is the week's disclosure run. Meta put its future AI-related commitments at about $825 billion, up from roughly $700 billion in July, against 2026 capex guidance of $135-145 billion on negative free cash flow. SK Hyper is seeking about $350 billion for one 5GW phase of a 15GW programme. Google Cloud is planning for up to 960,000 Rubin GPUs. None of those are earnings; all of them are obligations or ambitions, and the split among investors is now less about whether AI works than about who funds the gap and how long it stays open.

Sources

Schaeffer's · CNBC · XTB

Tags
AI NewsNvidia2026
⚑

Spot an inaccuracy?

We verify facts before publishing and correct errors promptly. If something in this article is wrong or outdated, let us know.

Report an error →