Broadcom's $230 Billion AI Bet Hinges on a Stock You Can't Buy

Broadcom just delivered one of the biggest forward AI guides in chip history — and the stock sold off anyway. Shares fell about 4% on Thursday after the semiconductor giant's fiscal Q3 report, even though the quarter itself beat expectations on nearly every line.

Revenue for the quarter came in at $29.6 billion, up 86% year over year and ahead of the roughly $29.4 billion consensus. AI semiconductor revenue more than tripled to $16.7 billion, up 221% year over year, now making up 56% of total sales. Adjusted EPS of $3.32 beat the $3.24 estimate, extending Broadcom's beat streak to nine straight quarters, as financial outlet FinanceFeeds detailed in its earnings breakdown this week.

So why the selloff? Fourth-quarter revenue guidance came in at $34.8 billion against a $35.03 billion consensus — a shortfall of less than 1%. That's the entire story behind Thursday's drop: not a bad quarter, but a growth rate that decelerated by a hair more than the Street wanted to see.

The bigger number that got attention, though, was CEO Hock Tan's longer-range guide: AI semiconductor revenue is projected to hit roughly $115 billion in fiscal 2027, then double again to $230 billion in fiscal 2028. Tan said demand "was simply hot and we're just getting started," pointing to customer compute needs that would "inflate even more" over 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

That warning from CNBC's Jim Cramer cuts right to the risk in Broadcom's forward guide, as 24/7 Wall St. reported from his Thursday appearance. Tan named Anthropic directly as the customer behind much of that eye-popping fiscal 2027-2028 number, citing deployments of one gigawatt of Ironwood compute in 2026, another five gigawatts of TPU v8i capacity in 2027, and visibility to 10 more gigawatts in 2028.

The catch: Anthropic is a private company. There's no way for retail investors to buy exposure to it directly, and its business model depends on continued access to capital that, so far, has come from private funding rounds rather than public markets. If Anthropic doesn't file for an IPO — or scales back its infrastructure spending for any reason — a meaningful chunk of Broadcom's most bullish long-term number gets shakier.

It's worth putting Broadcom's AI ramp in context against the broader chip sector. Rivals like Nvidia have leaned on a wider customer base spanning cloud giants, enterprises, and sovereign AI projects, which is one reason Nvidia shares actually ticked up around 2% even as Broadcom stumbled on the same trading day. Broadcom's model — building custom silicon for a handful of hyperscale customers — pays off enormously when it works, but it concentrates risk in a way diversified chipmakers don't face.

For investors holding or considering AVGO, that makes any future Anthropic S-1 filing one of the most important catalysts to watch, arguably more consequential in the near term than Broadcom's own quarterly prints. It's an unusual position for a $1.6 trillion-plus semiconductor company to be in — dependent on the fundraising timeline of a customer it can't control.

None of this erases the fact that Broadcom's custom AI silicon business is real and growing fast. Nine consecutive quarters of beating estimates isn't luck, and a 221% year-over-year jump in AI chip revenue reflects genuine hyperscaler and AI-lab demand that isn't going away even if Anthropic's path to public markets stalls.

Options traders had priced in a big move heading into the print, with implied volatility suggesting a swing of more than 12% in either direction. The actual move landed within that range, and the muted extended-hours reaction relative to the size of the guidance beat suggests the market had already partially absorbed the Anthropic dependency story before earnings even hit.

The stock is still up 24% over the past year despite Thursday's pullback and remains down about 6% over the trailing month — worth watching for investors who think the guidance dip was an overreaction to a rounding-error miss rather than a real crack in the AI capex cycle.

Bottom line: Broadcom's quarter was strong, its multi-year guide is enormous, and the stock's near-term direction now hinges less on Broadcom's own execution than on whether one private AI lab goes public.