Broadcom's AI Revenue Grew 221%. The Stock Fell Anyway.

Broadcom just posted the strongest AI quarter in its history, and the market handed shareholders a 5% loss for it.

AI semiconductor revenue hit $16.7 billion in the fiscal third quarter, up 221 percent from a year ago. Total revenue reached $29.6 billion, up 86 percent, and adjusted earnings per share came in at $3.32. By any ordinary measure, this was a beat, not a stumble.

The damage came a paragraph later, in the outlook. Broadcom guided fourth-quarter revenue to about $34.8 billion — up 93 percent, still explosive — against the roughly $35 billion Wall Street had penciled in. A gap of a few hundred million dollars on a company printing $29.6 billion quarters. That gap, not the growth, is what moved the stock.

This is the second time in three months the same thing has happened. In early June the company reported record numbers, AI revenue up 143 percent, and reiterated its plan to pass $100 billion in AI chip sales in fiscal 2027. The stock sank more than 12 percent, because the guidance had not been raised. Now the same structure: spectacular results, guidance that merely confirms, and another fall.

The pattern tells you what is actually being traded. The market is not paying up to find out whether Broadcom is thriving. It is paying for a specific growth trajectory — and it stops paying when the company delivers that trajectory but nothing more.

Here is the math on the setup. The stock entered the report already more than 25 percent below its all-time high, still trading at roughly 31 times forward earnings. A multiple like that capitalizes the boom in advance. It does not reward you for good news; the good news is already in the share price. The only information that moves such a stock is news better than the forecast — and when the bar is 200 percent AI growth, there is almost nothing left that qualifies as better.

So the honest reading of the 5% drop is not "AI is failing." The underlying engine looks intact. AI revenue is contracted with six of the frontier labs — Google, Meta, Anthropic and OpenAI among them — under long-term supply agreements that stretch toward 2028. Management is guiding AI chips to $21.7 billion next quarter, up 236 percent. The falling price is the market re-marking a trajectory it had already bought, not a verdict on the technology.

That distinction should change what "buying the dip" means to you here. When a stock drops on a guidance number that is itself 93 percent growth, the decline is not a clearance sale. It is the market taking back the premium it had placed on perfection — and that premium does not just come back because the next quarter is also strong. Two missed beats in a row is evidence, not noise.

The legitimate bear case exists and deserves its weight. Reuters reported that Broadcom itself warned that intense competition could blunt gains from its custom processors. A business tied to a handful of hyperscaler customers, carrying a valuation that demands its own acceleration, has a thin margin for that kind of friction. That is the real risk in the story — not a fake AI bubble narrative, but genuine competitive and concentration pressure against an already-high bar.

The thing that would reset this setup is specific: a raised fiscal 2027 target, a new marquee customer, or a multiple that grinds low enough that simply confirming guidance becomes the surprise. Until one of those shows up, treat a green earnings report next to a red stock price as the same message delivered twice. The market already believed. Broadcom just failed to exceed belief — and a company that has been repriced for perfect execution is punished for anything less.