On 2 November 2026, the artificial intelligence industry delivered one of the strongest earnings days in its history — and investors sold anyway. Anthropic, Apple and Palantir all posted record numbers, Marvell announced a $115 billion acquisition, and OpenAI signed four new deals in Abu Dhabi. Yet share prices fell across the board. The message from the market was clear: in the AI boom, big revenue is no longer enough. Investors now want to see profit, discipline and a believable path to returns on enormous spending.
Anthropic was the headline act. The company reported about $17 billion in revenue for the third quarter, up from $11.6 billion in the previous quarter and just $1.4 billion a year earlier. Growth, however, had slowed to 47%, and the company swung to a $567 million operating loss. The filing behind those numbers told a deeper story: for the first nine months of 2026, Anthropic earned $33.2 billion against $18.8 billion in computing costs alone. In other words, more than half of every dollar of revenue was consumed by the cost of running its models. For a company preparing to list on the stock market, that ratio matters more than the headline growth rate, and reports suggested future investors were pushing for a lower valuation than first expected.
Apple’s result told a similar story from a different angle. The company posted a record September-quarter revenue of $114.5 billion, up 12% year on year. But a softer margin forecast was enough to send the shares down about 3% in after-hours trading. Apple is not an AI pure play, yet its result was read through the same lens: even the world’s most profitable hardware company is spending heavily on AI, and investors want proof that the spending will protect margins rather than erode them.
Palantir delivered the most dramatic contrast between growth and reaction. Revenue nearly doubled, rising 96% to $2.313 billion. But the value of new contracts it signed grew only 6%, and the shares fell anyway. The market was asking a forward-looking question: how fast can the order book keep growing once the first wave of AI contracts has been signed? A company can report a superb quarter and still disappoint if the pipeline behind it looks thinner.
Marvell’s news was different again. It agreed to buy Lumentum, a laser maker, for $115 billion in stock, paying a premium of about 24%. Its shares closed down 12.72%. Investors were not questioning the strategic logic — lasers are a critical part of AI data-centre hardware — but they were questioning the price and the timing. In a market already nervous about AI valuations, a giant all-stock deal looks like a bet that today’s prices will still look cheap tomorrow.
OpenAI, still private, added to the mood of caution. It signed four deals in Abu Dhabi, yet at the same time cut its latest funding round from $30 billion to $20 billion, with its valuation reported down from $1.4 trillion to $1.25 trillion. The deals show demand; the smaller round shows that even the most famous name in AI cannot raise money on any terms it chooses.
Behind all of this sits a broader market turn. The Nasdaq had fallen about 10% from its recent high, and 5.2% across October alone, its worst month since March 2025. Alphabet had already warned investors with its own numbers: revenue of $127.6 billion, but quarterly capital spending of $51.4 billion against $24 billion a year earlier, and negative free cash flow of $9.1 billion. The AI build-out is real, but so is the bill.
For businesses that buy AI rather than sell it, the lesson is practical rather than financial. First, expect AI prices and contracts to keep changing as suppliers absorb their own computing costs. Second, judge vendors on unit economics — what a tool costs per task, per customer or per hour saved — not on how fast the vendor itself is growing. Third, avoid locking a whole operation into a single model or supplier while the market is still repricing risk. The companies that use AI well in 2027 will not necessarily be the ones that bought the most of it in 2026. They will be the ones that bought it carefully, measured the return, and kept the freedom to switch.
The record results of 2 November were not fake. They were real, audited and in some cases extraordinary. What changed is the question investors are asking. A year ago it was: how fast can AI grow? Now it is: how much will that growth cost, and who will actually profit from it? Until companies can answer the second question as convincingly as the first, even record quarters may be greeted with a sell-off.
Sources: company filings and reporting by Fortune, the Wall Street Journal, CNBC and Investing.com, as summarised by Van Data Team (vandatateam.com) and Working Intelligence.
