The S&P 500 gained 1.2% and the Nasdaq rose 0.6% for the week ended October 9, but those calm closing numbers hide a sharp reversal in AI-linked stocks. Both indexes reached records on Tuesday, technology shares sold off on Thursday, and a Friday rebound kept the week positive. The trigger was not a public earnings release. It was conflicting reporting about OpenAI's annualized revenue expectations.
A private-company number moved public markets
CNBC reported that a Financial Times article suggested OpenAI's annualized revenue at the end of September was roughly $20 billion below previously indicated expectations. Broadcom fell 4.4% on Thursday, although it finished the week higher. On Friday, Bloomberg sources said OpenAI expected annualized revenue of at least $70 billion by year-end, helping the AI trade recover.
Annualized revenue is a run-rate measure, not audited full-year revenue or free cash flow. It can be useful for a fast-growing private company, but it is highly sensitive to the measurement date and assumptions. Investors repriced public suppliers before either figure appeared in a filed financial statement. That is the central risk: a private customer's reported growth can affect listed chip and infrastructure companies whose actual revenue is disclosed on different schedules.
Concentration magnified the move
AI suppliers share exposure to a small group of large customers, data-center budgets and financing conditions. When one revenue story changes sentiment, correlated holdings can decline together even if their near-term orders have not changed. Diversification across business models cannot eliminate market losses, but it can reduce the chance that a single narrative forces a portfolio-wide decision.
Other developments showed that the market was not trading on one factor. SpaceX reportedly explored a $40 billion debt raise to buy more Nvidia chips and expand AI compute, then announced an $8 billion spectrum purchase intended to strengthen Starlink mobile service. Telecom shares fell on the competitive threat. Starbucks, meanwhile, received improving consumer feedback in a 29,000-person HundredX dataset, illustrating how company-specific evidence can diverge from the dominant AI theme.
What investors should watch
The more reliable tests are supplier revenue, backlog, customer concentration, capital spending and cash conversion. The next week's inflation data and earnings reports could also change interest-rate expectations, which affect long-duration technology valuations.
BTI's bottom line
The week's lesson is not that AI demand disappeared or was restored within 24 hours. It is that investors briefly priced two conflicting private-company claims as if each were definitive. Public filings and cash economics should carry more weight than an unaudited run-rate headline.