With AI Revenue Set to Surge 400% Over the Next 2 Years, Broadcom Stock Looks Like a Buy on Recent Dip

Broadcom projects AI revenue of $115B in fiscal 2027 (vs prior ~$100B guidance) and then $230B in fiscal 2028—implying roughly a 2x step-up each year. In fiscal Q3, revenue rose 86% YoY to $29.6B and adjusted EPS jumped 96% to $3.32, topping LSEG expectations (EPS $3.24; revenue $29.36B). The article also highlights valuation support with the stock trading at 11.5x forward P/E on its fiscal 2028 outlook and calls the post-dip setup an attractive AI-infrastructure buy.

Analysis

The market is likely underappreciating that AVGO’s AI thesis is less about top-line growth and more about becoming embedded in hyperscaler roadmaps, which makes the next 12-18 months a supply execution story rather than a demand story. That is bullish for visibility, but it also means the stock will start trading on whether the company can actually convert multi-generation design wins into shipped units without sacrificing mix or margin.

Second-order winners are the hyperscalers building custom silicon stacks, especially GOOGL and META, because every percentage point of performance-per-dollar improvement reduces dependence on merchant GPUs and can defend cloud economics over time. The biggest structural loser is probably not NVDA outright, but the broader merchant accelerator ecosystem: if inference shifts more toward custom ASICs, the addressable share for off-the-shelf compute and some adjacent networking vendors compresses, even if total AI spend keeps rising.

The contrarian risk is that the “cheap on forward EPS” framing may be misleading because the market is paying for a very high-quality software multiple on an increasingly semis-heavy mix. As semis dominate the revenue mix, consolidated gross margin can keep drifting lower, so the multiple deserves some haircut unless the software segment re-accelerates or AI networking proves more margin-accretive than expected. If customer capex guides flatten, or if gross margin fails to stabilize above the low-70s, the de-rating could happen fast over 1-3 months even if the 6-18 month AI runway remains intact.

My base case is positive but not chaseable after a strong move: use weakness to build, not strength to add. The cleanest falsifier is a 1H27 spending delay from GOOGL/Meta or evidence that AVGO’s supply commitments are stretching beyond its current manufacturing envelope.

Key Decisions for Investors

• Initiate a staggered long AVGO position on 3-5% pullbacks over the next 2-4 weeks; hold for 6-12 months. Risk/reward is attractive if AI revenue visibility converts into sustained EPS compounding, but trim if gross margin trends below the low-70s.
• Pair trade: long AVGO / short NVDA over 1-3 months as a relative-value expression of custom-silicon share gains versus merchant GPU exposure. Thesis works if inference workloads keep shifting toward bespoke ASICs; invalidate if NVDA re-accelerates on training demand and AVGO order cadence slows.
• For convexity, use a 6-12 month AVGO call spread rather than outright stock if position size is limited by valuation risk. This captures upside from execution while capping downside if the market de-rates the multiple on margin dilution.
• Watch GOOGL and META capex commentary as the key catalyst set; if either signals a slowdown or deferral in AI buildouts, reduce AVGO exposure immediately. That would be the first high-signal warning that the 2027-2028 revenue bridge is too aggressive.
• No urgent short recommendation on NVDA here, but consider a small hedge in SMH or SOXX if already long AI semis broadly. The risk is not sector collapse, but AVGO-specific outperformance squeezing the relative basket.