Two of the year's most-watched companies reported earnings this week, both posted numbers that would have been the envy of any boardroom a decade ago, and both were promptly punished by investors. SpaceX, in its first quarterly report since June's record IPO, grew revenue 92 percent to $7.81 billion and comfortably beat estimates. The stock fell about 8 percent after hours. AMD reported record revenue of $11.5 billion, up 50 percent, with data-center sales more than doubling. Its shares dropped roughly 7 percent. The AI trade is discovering an awkward truth: when a stock is priced for a miracle, merely doing extremely well starts to look like a disappointment.
Look closer at SpaceX and the strain is obvious. Elon Musk's company merged with his AI venture xAI in February, and the cost of that ambition is now visible on the balance sheet. Its AI segment lost $1.26 billion in the quarter, and the space business lost another $542 million; only Starlink, the satellite-connectivity arm, actually made money, with $1.66 billion in operating income. The company burned through $4.9 billion last year, largely on AI infrastructure, and its shares are down about 16 percent since the June debut. Investors got a revenue beat and focused instead on the capital pouring out the door.
AMD's report told a subtler version of the same story. Data-center revenue surged 107 percent to $6.7 billion, and the company guided to another strong quarter. But gaming revenue plunged 31 percent, and the stock trades at an eye-watering 124 times earnings, against roughly 30 for the S&P 500. At that valuation, as one market watcher noted, a company has to beat expectations "by a mile" and promise to beat them again next quarter just to stand still. AMD did neither emphatically enough, so it fell. The revenue was real; the priced-in perfection was the problem.
Zoom out and the nerves have a name. In its July survey, Bank of America ranked a bursting of the AI bubble as the single biggest risk to financial markets. The Bank for International Settlements has warned that AI valuations exceed earnings by a margin not seen since 1929. Investor Michael Burry, of "Big Short" fame, has cautioned that markets are in "rare air, so extreme that the consequences will be unavoidable." And an analysis by Bain and Company concluded that tech companies will need to find some $2 trillion in fresh AI revenue just to turn a profit on what they are spending to build the infrastructure.
Against all that stands the industry's most reliable optimist. Nvidia's Jensen Huang told Bloomberg he expects the semiconductor industry to grow perhaps tenfold over the next decade, to nearly $7.9 trillion, to feed an oncoming boom in autonomous AI agents, and Nvidia's own revenue is still climbing 85 percent a year. That is the real question hanging over this week's numbers. The growth is genuine and enormous; nobody is faking the revenue. What no one yet knows is whether it can grow fast enough, and long enough, to justify the trillions already committed. Beat-and-fall is not a crash. But it is the market quietly admitting that the easy part of the story is over.