Broadcom Beat Every Number, So Why Did the Stock Drop 4%?

Broadcom just posted one of the strongest quarters in its history, and the stock fell anyway. That disconnect is worth understanding if you own AVGO or are thinking about buying the dip.

The chipmaker's fiscal third-quarter results, released September 2, beat Wall Street on every headline number. Revenue hit $29.6 billion, up roughly 86% year over year and above the $29.4 billion analysts expected. Non-GAAP earnings per share came in at $3.32, topping the $3.24 estimate and extending Broadcom's beat streak to nine straight quarters.

AI semiconductor revenue was the real headline: $16.7 billion, up 221% year over year and 54% from the prior quarter. Guidance for the current quarter calls for total revenue near $34.8 billion, 93% higher than a year ago, with AI semiconductor revenue projected to hit $21.7 billion.

Shares still fell as much as 6% in after-hours trading before paring the drop to around 3.5%. By the next afternoon, AVGO had slid roughly 4% to trade near $352, even though the stock remains up 24% over the past year, as Meridians' Money Desk detailed in its earnings breakdown this week.

That kind of reaction — a beat-and-raise quarter met with a sell-off — usually means the market is pricing in something the headline numbers don't capture. In this case, it's concentration risk buried in the guidance Broadcom gave for the next two years.

"This is now Anthropic. It's Anthropic or bust… They need Anthropic to have a blockbuster IPO." — Jim Cramer, CNBC's Squawk on the Street

CNBC's Jim Cramer offered the clearest explanation for the selloff during Squawk on the Street. Broadcom CEO Hock Tan told analysts the company expects AI semiconductor revenue to reach roughly $115 billion in fiscal 2027, then double again to $230 billion in fiscal 2028. That's the kind of number that should send a stock soaring on its own.

The problem, according to Cramer, is who's driving it. Tan named Anthropic as on track to become Broadcom's largest AI chip customer in 2027, deploying one gigawatt of custom silicon this year and scaling toward five more gigawatts in 2027, with line of sight to another 10 gigawatts in 2028. Anthropic is a private company. There's no way for a retail investor to buy shares in it directly, and it hasn't filed for an IPO.

OpenAI is the second-largest customer in the roadmap Tan laid out, with 1.3 gigawatts planned for 2027. Google, Meta, and two unnamed buyers round out the customer list. Broadcom's own regulatory filing flags "dependence on a limited number of significant customers for AI semiconductor demand" as a formal risk factor — this isn't speculation from outside analysts, the company disclosed it itself.

Compare that to Nvidia, which reported Q2 revenue of $96.22 billion and guided to $108 billion next quarter, spread across a broader mix of frontier AI labs including its own direct partnership arrangements. NVDA actually traded up 2% the same day Broadcom fell, because its customer concentration risk looks lower by comparison to investors weighing similar exposure.

Here's the practical takeaway: if you own AVGO because you believe the $230 billion fiscal 2028 number, you're effectively making a bet on an Anthropic IPO that hasn't been filed yet. That's not necessarily a bad bet — Anthropic's growth trajectory and model demand are well documented elsewhere — but it's a specific, identifiable risk rather than a diversified AI-chip thesis spread across many buyers.

There's an added wrinkle worth knowing about. Anthropic recently disclosed that foreign AI labs used roughly 24,000 fraudulent accounts to distill its Claude models — a reminder that model-security and IP risk is now a real factor for any lab preparing to eventually go public.

What to watch next: any Anthropic S-1 filing would be the clearing event that validates this entire growth story, and Broadcom's fiscal Q4 report on December 9 will show whether that customer mix is holding up as promised. Until then, this is a stock where the fundamentals and the price are telling two different stories — and right now the market is listening to the risk, not the revenue.