Markets don't care if you beat earnings. They care if you beat expectations — and on June 3, 2026, Broadcom learned that lesson the hard way.

$AVGO entered its Q2 FY2026 earnings report at all-time highs, brushing $495 intraday. Wall Street had priced in perfection — a flawless AI growth narrative, accelerating hyperscaler demand, and a VMware integration firing on all cylinders. Then the numbers hit.

Broadcom beat earnings estimates. Revenue came in strong. The core semiconductor business held up. But the AI chip revenue forecast — the number everyone was actually watching — came in below the whisper numbers that had driven the stock to nosebleed territory. Within hours, $AVGO cratered ~15%, tumbling from roughly $479 post-market to around $405 by June 4. That's roughly $300 billion in market cap — gone. In a single trading session.

The carnage wasn't contained. AMD and Intel got dragged down sympathetically as the market suddenly questioned whether the AI chip trade had overshot. For a stock that was trading at 22-23x forward earnings and had been crowned Wall Street's top semiconductor pick for 2026, the selloff was as swift as it was brutal.

📖 Read more: The Great Rotation Is Finally Here — Dow Record, Chip Rout, and the Trade Nobody's Talking About

The Numbers: A Beat That Wasn't Good Enough

Here's the uncomfortable truth about earnings season in 2026: the bar for AI-exposed companies is no longer about growth. It's about acceleration. Broadcom grew. It beat on the bottom line. Its AI revenue — driven by custom ASICs like Google's TPU and next-generation networking chips — continued to expand aggressively. But the market had priced in something closer to a blowout, and when the forward guidance landed with a thud, the air came out of the trade.

The stock had run from a 52-week low of $241.11 all the way to $495 — more than doubling in under a year. When you've priced in that much optimism, anything short of perfection becomes a sell-the-news event. That's exactly what happened.

The AI Business: Real Demand, Inflated Expectations

Broadcom's AI story is fundamentally different from Nvidia's. $NVDA owns the GPU monopoly — off-the-shelf chips that any hyperscaler can buy and deploy. Broadcom's model is custom silicon: designing bespoke ASICs for specific customers like Google (TPUs), Meta, and reportedly a third undisclosed hyperscaler. These chips are purpose-built for each customer's AI workloads, offering better power efficiency and cost performance at scale than general-purpose GPUs.

The bull case has always been that as AI matures, hyperscalers will migrate from buying $40,000 H100s from Nvidia to designing their own chips — and Broadcom is the partner they'll call to build them. That thesis isn't broken. Broadcom reportedly has $200 billion or more in AI compute deals locked in across its customer base. That's not vapor — it's contracted revenue that will materialize over the coming years.

What the Q2 report exposed wasn't a demand problem. It was an expectations problem. The market had extrapolated near-term AI revenue growth in a straight line upward and priced the stock accordingly. When Broadcom's guidance suggested the ramp would be steadier — not slower, just steadier — the air pocket opened.

VMware: The Silent Engine

Lost in the AI panic is Broadcom's other massive transformation: the VMware acquisition. Since closing the deal, Broadcom has aggressively restructured VMware's business model — shifting customers from perpetual licenses to subscription contracts, simplifying the product portfolio, and slashing costs. The result has been a dramatic margin expansion that flows straight to the bottom line.

VMware revenue provides a stabilizing counterweight to the lumpy, project-based nature of custom ASIC deals. While AI chip revenue grabs headlines, VMware's subscription revenue is predictable, high-margin, and growing. The market, in its single-minded focus on AI guidance, may be undervaluing this cash flow engine.

What This Means for the AI Semiconductor Trade

The AVGO selloff is a Rorschach test. Bears see a canary in the coal mine: if the second-most-important AI chip company can't meet sky-high expectations, maybe the whole AI infrastructure buildout is overhyped. Bulls see something else entirely: a $300 billion overreaction to a guidance print that changes nothing about the structural demand for AI compute.

Here's what we know. Hyperscaler capex — from Microsoft, Amazon, Google, Meta — continues to accelerate. Every major cloud provider is building AI infrastructure as fast as they can pour concrete and string fiber. The chips that power those data centers have to come from somewhere. Broadcom, alongside Nvidia and a handful of others, is one of the few companies on earth capable of supplying them at scale.

Broadcom's custom ASIC model also provides a strategic hedge that Nvidia doesn't have. As hyperscalers increasingly seek to reduce their dependence on any single supplier, the demand for custom silicon designed in partnership with Broadcom should only grow. Nvidia's GPU business is enormous and dominant — but it's also a single point of failure that customers are actively trying to diversify away from.

The Silver Lining

At roughly $405, AVGO sits well below the Wall Street consensus target of ~$480 and miles below the high target of $630. The forward P/E has compressed significantly. The 52-week high of $495 — hit literally hours before the earnings report — now looks like a distant memory.

If you believe AI demand is real — and the capex numbers from every hyperscaler suggest it is — then Broadcom's long-term thesis remains intact. The selloff punished the stock for a guidance miss, not a business-model failure. The company still dominates custom AI silicon. VMware still prints cash. The $200 billion-plus in contracted AI deals hasn't vanished.

The question isn't whether Broadcom is a good company. It's whether you're willing to buy when everyone else is panicking — and whether you believe that the AI buildout is a multi-year structural shift, not a short-term hype cycle. The market just handed you a 15% discount to find out.