|
Two AI giants delivered blowouts in a week, and the market sold them both
The numbers Broadcom put up last night belong in a record book. Sales reached $29.6 billion for the quarter, an 86 percent jump in twelve months. Profit nearly doubled. Free cash flow came to $13.7 billion, almost half of revenue. And the line the whole market was watching, revenue from the custom AI chips it designs for the largest technology companies, more than tripled from a year ago to $16.7 billion. On the call, chief executive Hock Tan raised the company’s 2027 target for that business to around $115 billion. By almost any standard this was one of the great quarters in the chip industry’s history. The stock fell as much as 6 percent.
It fell for a reason that says more about the market than the company. Broadcom guided next quarter’s revenue to about $34.8 billion. Analysts, on average, had penciled in $35.03 billion. The gap is roughly two-tenths of one percent. In a calmer market, a rounding error against a quarter like this would be forgotten by lunch. In this one it was the whole story, because the good news was already in the price.
This is the second time in three months Broadcom has beaten and been sold; in June it dropped about 13 percent after a similar blowout. And it is the second AI giant in a week to get the treatment, after Nvidia reported a record last Wednesday and spent its session fighting to hold a gain. We wrote on Tuesday that Broadcom carried so much success in its price that even a large beat could be met with selling. That is exactly what happened.
The bears lost the argument, the bulls lost the trade. Step back from the tape and the fundamental verdict is plain. For a year, the fear around Broadcom was that custom silicon might not scale, that the hyperscalers’ plans to design their own chips would stay a slide-deck ambition. Last night ended that debate: AI revenue up 221 percent, a fourth-quarter guide of $21.7 billion, a 2027 target lifted to $115 billion. The customers building alternatives to Nvidia are doing it faster, not slower. The demand story strengthened. What softened was the market’s willingness to pay up for it.
What a high bar costs you. When a market sells blowouts, it is telling you something specific: expectations have caught up with reality, and the easy gains from a rising theme are behind it. That does not mean the theme is finished; Broadcom and Nvidia may compound for years. It means the way you make money in them has changed. A stock that rises on great news rewards optimism. A stock that falls on great news punishes it, and forces an investor to be right not only about the company but about a bar that keeps lifting. For the index owner this matters, because Nvidia and Broadcom sit near the top of the S&P 500. When the market’s biggest engines stop climbing on good news, the index they anchor loses its lift.
Where that leaves you. The takeaway is simpler than buy-or-sell: notice what the reaction reveals. A market that cannot rise on two historic quarters is running on a very high bar, two days before a jobs report that could shake the rate backdrop beneath it. That is a moment to value durability over momentum, businesses whose prices already assume ordinary results rather than flawless ones. The AI story remains intact; the market’s patience with anything short of perfect is what has thinned. None of this is investment advice.
Broadcom just proved the AI buildout is bigger than the skeptics thought, and its stock fell anyway. That is what a market at a very high bar looks like: the news keeps getting better and the reward keeps getting smaller. Friday’s jobs number will test whether the ground underneath can hold the weight.
| 