Microsoft's record day
The week's defining moment belonged to Microsoft. Shares closed up 15.5% on Thursday, adding roughly $450 billion in market value — the largest one-day increase in market capitalization any company has ever posted. By Fortune's accounting it was Microsoft's best single day since 2008, though the exact scale of the move depends on who is doing the counting.
The catalyst was earnings. The Economic Times reported the stock surged 16% as the company's AI cloud bet "starts showing up in numbers" — the phrase that mattered most to a market that has spent months funding AI capital expenditure on faith. The timing was pointed: Wednesday had been brutal for AI-linked stocks, and Microsoft's print yanked the whole trade back from the brink, carrying the rest of Big Tech with it.
The read-through ran straight to the picks-and-shovels crowd. The Motley Fool argued on August 1 that Microsoft's data center plans amount to another vote of confidence for Nvidia shareholders, delivered by one of Nvidia's largest customers. When a hyperscaler guides capacity higher, the spending lands somewhere — and Nvidia is still the default destination.
Amazon follows
Amazon closed up 15.32% on Friday, its biggest one-day jump since April 2012. Bloomberg's Spencer Soper reported the move followed accelerating cloud revenue growth at AWS, alongside a chip business now running at roughly $25 billion. That second number is the one worth sitting with: Amazon is no longer just a buyer of AI silicon, it is a meaningful producer of it, which complicates the tidy story of a single-vendor AI supply chain.
The chip round trip
Semiconductors took the scenic route. The slide began in Asia, and Yahoo Finance reported U.S. chip stocks extending losses in Tuesday premarket trade, building on Monday's declines. Then it reversed — violently — with the Kospi posting a record day on the rebound. Crash and recovery inside a single stretch of sessions is the signature of a sector where positioning, not fundamentals, sets the short-term price.
A $45 billion bet unravels
No story captured the whiplash better than Leopold Aschenbrenner's. The 24-year-old former OpenAI researcher, whose essay collection "Situational Awareness" made him a minor celebrity in AI circles, spent the first half of 2026 as one of the most celebrated investors in finance. Financial Express reported his fund of the same name was up 439% year-to-date by the end of June.
Then July happened. The Wall Street Journal reported the fund fell roughly 67% in a single month, giving back close to a year of gains in a matter of days. The lesson is not that the AI thesis was wrong; it is that concentration cuts symmetrically, and the same leverage that produced a 439% run produced the drawdown.
Where the money goes next
Underneath the volatility, the composition of AI spending is shifting. The Motley Fool argued on August 1 — in a piece syndicated through Yahoo Finance — that the center of gravity is moving from training models to running them, with inference representing a prize the article sizes at $1.3 trillion. Wall Street is repositioning accordingly. Training is a capital project with a finish line; inference is an operating cost that scales with every user, every query, every day.
The next test
Nvidia reports on August 26, and analysts are, per an MSN report, "nearly unanimous at strong buy" — while remaining genuinely split on what the quarter holds. The bar has been set around $91 billion. After a week where two megacaps each added double-digit percentage gains in a session and one hedge fund lost two-thirds of its value, that number is the one the whole trade will be graded against.