Broadcom's AI Revenue Just Jumped 143%. So Why Is the Stock Sliding?

Broadcom shares are down ~22% from an all-time high after a post-earnings 13% drop, despite fiscal Q2 results: revenue rose 48% Y/Y to a record $22.8B, net income increased 88% to $9.3B, and adjusted EPS rose 54% to $2.44 (beating estimates). AI is the key driver, with AI semiconductor revenue up 143% Y/Y to $10.8B and management guiding AI semiconductor revenue to rise >48% sequentially and 200% Y/Y to $16B in fiscal Q3. The sell-off is attributed mainly to profit-taking and high valuation (P/E >80 at the peak; still high at 60 now but forward P/E ~19), with a projected gross margin dip from 77.1% in Q2 to ~74% in Q3 due to a higher mix of lower-margin AI chips.

Analysis

The setup is less about fundamental deterioration and more about the market refusing to pay peak multiple for a name that already became the default AI winner trade. When a stock is priced for flawless upside, even a still-strong guide can trigger de-risking if the next step-up in growth looks smaller than the last one; that is a sentiment/positioning event, not an operating one. The opportunity is that the business is transitioning from narrative scarcity to backlog conversion, which usually supports a slower but more durable rerating once quarterly execution catches up.

Second-order, the important issue is mix. A higher share of AI silicon that carries worse gross margin can compress near-term quality screens, but it also widens the addressable wallet share inside hyperscaler capex and should pull through networking, packaging, and software attach over time. That favors the broader AI infrastructure complex more than pure compute names: NVDA still owns the premium AI GPU narrative, but AVGO’s custom accelerator exposure is a latent competitive threat if large customers keep optimizing away from merchant silicon.

Risk is that the de-rating persists for 1-3 months if AI spend headlines flatten or if the next quarter confirms slower sequential growth and margin compression. The thesis breaks if management re-accelerates sequential AI revenue, raises the multi-year target, or shows gross margin stabilization despite the mix shift. Longer term, if the backlog is real, the current move looks like valuation compression ahead of earnings power being recognized, not the start of an earnings miss cycle.

Key Decisions for Investors

• Go long AVGO on continued post-earnings weakness, using a 3-6 month horizon; prefer staged entry rather than chasing a one-day bounce. Risk/reward is attractive if the stock re-rates even halfway back toward prior highs on backlog conversion evidence; thesis is falsified if next quarter sequential AI growth drops below management’s implied cadence.
• Pair trade: long AVGO / short SMH or XLK for a relative-value expression over the next 1-3 months. This isolates AVGO-specific rerating from a broader AI beta unwind; cut the trade if semiconductor multiples expand again on renewed capex enthusiasm.
• If options liquidity is acceptable, buy a 6-12 month AVGO call spread instead of stock to express upside while limiting the risk that valuation stays compressed for another quarter. The catalyst is the next earnings print; the trade loses if gross margin remains stuck in the low-74% area without offsetting guide raises.
• Monitor NVDA as the competitive read-through, but do not short it outright unless hyperscaler commentary shifts toward more custom silicon adoption. AVGO is the cleaner long; NVDA remains the higher-quality momentum name and can keep outperforming on narrative alone.
• Set an alert for any upward revision to FY2027 AI revenue or signs that sequential growth re-accelerates after Q3. That would invalidate the 'peak optimism' thesis and likely force a fast multiple expansion in AVGO over days, not months.