Broadcom delivered record Q2 fiscal 2026 revenue of $22.2 billion - up 48% year over year - with AI semiconductor revenue hitting a record $10.8 billion. The stock fell about 15%. Wall Street wasn't upset about the results. It was upset that CEO Hock Tan didn't raise the company's $100 billion annual AI semiconductor revenue target.
A false narrative developed around the selloff. The market heard "not raising the target" and decided the AI boom is stalling. The actual data says the opposite: the business is accelerating.
Here's what matters:
1. AI revenue is accelerating, not plateauing.
Broadcom's AI chip revenue went from $8.4 billion in Q1 to $10.8 billion in Q2, and the company guided Q3 AI semiconductor revenue to $16 billion - for a single quarter running through July. That's a $5.2 billion sequential jump. Each quarter is a bigger step than the last. The Q3 number annualizes to $64 billion on its own. The $100 billion target is for fiscal 2027, which runs through October 2027 - there's still two quarters remaining. The trajectory isn't flattening. It's steepening.
2. The $73 billion backlog is contracted demand, not aspirational talk.
Tan stated Broadcom has "line of sight to achieve AI chip revenue in excess of $100 billion in 2027". The $73 billion in AI backlog is the proof. Backlog means orders already placed by customers - it's committed revenue, not management optimism. When a company has $73 billion in contracted AI orders, the risk isn't demand. The risk is whether manufacturing can keep up. Broadcom isn't chasing customers. Customers are chasing Broadcom.
3. Not raising the target is discipline, not a warning.
Investors wanted Tan to raise the $100 billion number after Q2 AI hit $10.8 billion. Holding steady isn't a red flag - it's how management avoids becoming its own ceiling. Broadcom has a track record of delivering or exceeding guidance. Sticking to a public number while the business accelerates underneath it is the opposite of a problem.
4. EPS beat, and the stock just got a forced valuation discount.
EPS came in at $2.44 versus $2.39 expected. The stock still fell 15% because the market wanted a bigger promise, not more proof. Before earnings, Broadcom's forward P/E (the stock price divided by expected future annual earnings) was roughly 37.7x. A 15% selloff on record results compresses that multiple toward the low-30s. For a company growing revenue at 48% with $73 billion in backlog, that compression created a disconnect between price and earnings power. The math didn't change. The multiple did.
5. The bear case is real but distant.
Competition in custom AI accelerators is intensifying. Nvidia's custom silicon and cloud providers building in-house could pressure Broadcom's long-term market share. That's legitimate, but it's a 2028-2030 concern, not a 2026 problem. With $73 billion in committed orders already in the pipeline, the near-term revenue path is secured. Wall Street is pricing in a competitive threat that hasn't touched the backlog yet.
The valuation disconnect:
Broadcom was trading near 38x forward earnings before earnings. After a 15% drop on record results, the multiple compressed sharply. The company is growing at 48%, has $73 billion in contracted AI orders, and management hasn't changed its $100 billion target. The target is intact. The acceleration is real. The price moved down because the market didn't get a bigger cheer.
The stock may need to find a bottom before investors dive in - battered names often need stabilization time. But the setup is clear: when a record quarter produces a 15% drop because management played it steady instead of speculative, the gap between price and fundamentals sets up a buying opportunity, not a risk signal.
The break condition: if the next two quarters of AI chip revenue stay on this accelerating trajectory and the backlog converts as expected, the multiple re-rates. The key risk is a material slowdown in hyperscaler AI spending that forces management to revise the $100 billion target downward. If that happens, the thesis breaks.
But right now, the data says accelerate. Wall Street is selling because it didn't get a bigger promise. That's not investing. That's impatience.