Hock Tan Just Put a $230 Billion Number on Broadcom's 2028 AI Revenue. That Is 4 Times This Year's.

Broadcom’s CEO Hock Tan raised AI semiconductor revenue targets to ~$115B for fiscal 2027 (from >$100B) and put 2028 “line of sight” at $230B, about 4x the ~$58B expected this fiscal year. The company says supply is already secured for both years and demand exceeds the outlook, supporting further quarterly AI revenue growth (e.g., $10.8B in 2Q to $16.7B record in 3Q, with guidance up to $21.7B in 4Q). Despite deployment-timing risk (land/power/data-center bottlenecks) and only six key accelerator customers, the stock recovered after the call and is valued at ~18x expected fiscal 2027 earnings (~$115B AI revenue), implying a relatively attractive valuation if EPS guidance toward >$30 in 2028 is achieved.

Analysis

The key shift is that the constraint has moved from demand storytelling to industrial execution. That is bullish for AVGO’s long-duration cash flow quality because it implies multi-year design wins and pricing leverage, but it also means the next leg of the story is gated by power, permitting, and data-center commissioning rather than silicon alone. In the near term, the market may still punish any quarter where shipments outpace end-customer deployment, because backlog converts to revenue only as clusters turn on.

Second-order winners are the companies that sit on the rest of the AI buildout stack: grid equipment, power management, and datacenter infrastructure. On the loser side, merchant accelerator vendors face a subtler threat than outright share loss: hyperscalers increasingly have an economic rationale to internalize more workloads into custom silicon, which caps the multiple for the most expensive AI compute names if inference and steady-state training migrate to ASICs. GOOGL is the clearest strategic beneficiary among the names given, because this reinforces TPU as a profit lever rather than a science project.

The consensus risk is underestimating concentration. A six-customer revenue base means any one customer’s capex pause, land/power delay, or packaging bottleneck can create a temporary air pocket even if end demand remains intact. Falsifiers are simple: a quarter where AI revenue growth falls materially below the implied sequential run-rate, or commentary that deployment timing—not demand—starts stretching beyond 1-2 quarters. Over 6-18 months, if those checks pass, the stock deserves a premium multiple expansion rather than a de-rating.

Key Decisions for Investors

• Long AVGO on post-earnings weakness for a 3-6 month hold; treat it as a valuation reset trade with roughly 15-20% upside if the market starts capitalizing fiscal 2028 earnings.
• Pair trade: long AVGO / short NVDA for 3-6 months as a relative-value hedge against custom-ASIC share gains and hyperscaler insourcing; cut the pair if NVDA’s lead times and revenue growth re-accelerate faster than expected.
• Add GOOGL as a secondary long only if you want indirect exposure to TPU scale-up; this is a 6-12 month position tied to cloud monetization, not an immediate catalyst trade.
• Watchlist alert: if AVGO’s next quarterly AI revenue guide fails to step up sequentially, fade the move and reassess—the thesis is deployment velocity, not order visibility.