Is It Too Late to Buy Nvidia and Broadcom? Here's What History Says.

Nvidia reported Q1 revenue of $81.6B, up 85% year over year, with data center revenue rising 92% to $75.2B; Broadcom reported Q2 revenue of $22.2B, up 48% year over year. Nvidia guided Q2 revenue to $89.2B-$92.8B despite no China data center contribution, while Broadcom said AI semiconductor revenue reached $10.8B and bookings topped $30B, supporting over $56B of FY2026 AI revenue. The article is fundamentally constructive on both companies' AI growth but warns that elevated valuations and high expectations could limit upside.

Analysis

The market is rewarding AI exposure as if every layer of the stack will monetize at similar durability, but the second-order winner profile is likely narrower. The most important read-through is that infrastructure demand is moving from a pure GPU cycle into a broader connectivity-and-systems cycle, which tends to favor suppliers with pricing power in networking, custom silicon, and platform control rather than discrete component vendors. That shifts the center of gravity from a single product refresh to a longer capex migration, but it also raises the odds that the next leg of upside comes from execution, not multiple expansion.

For NVDA, the real risk is not demand decay; it is expectation saturation. When a company is already priced for multiyear dominance, each beat must be large enough to offset the market pulling forward future growth, so the stock can stagnate even while fundamentals remain exceptional. The most underappreciated catalyst is the transition from training-driven spend to inference, agentic workloads, and sovereign deployments, which broadens the addressable market and reduces hyperscaler dependency; the danger is that any pause in hyperscaler capex or a delay in next-generation platform ramps can compress sentiment quickly over a 1-3 month window.

For AVGO, the bull case is revenue visibility, but the hidden fragility is customer concentration combined with a valuation that already embeds a lot of that visibility. If one or two large customers optimize their in-house silicon roadmaps, the downside is not just lost revenue but a slower perceived growth algorithm, which typically de-rates high-multiple semi names much faster than the fundamentals deteriorate. The contrarian point is that networking content may be structurally underappreciated: as clusters scale, connectivity intensity rises faster than compute node count, so AVGO can keep growing even if accelerator unit growth normalizes.

Key Decisions for Investors

• Maintain a tactical long NVDA position into the next 1-2 earnings prints, but use call spreads rather than stock to cap valuation risk; best risk/reward if the market is still underestimating inference and Rubin-driven upgrade demand.
• Own AVGO only as a core but hedged position: long AVGO / short SOXX or long AVGO / short a basket of high-beta AI semis to isolate customer-commitment visibility against sector multiple compression over the next 3-6 months.
• If entering new capital, favor AVGO over NVDA on pullbacks of 8-10% because revenue visibility is higher; however, keep position sizing smaller due to concentration risk and richer multiple.
• Consider a pairs trade long GOOGL / short NFLX as a second-order AI infrastructure expression: Google has direct monetization leverage to custom silicon and networking ecosystems, while NFLX is not a beneficiary and can act as a cheap short if broader tech sentiment weakens.
• Set a stop-loss/trim trigger on both NVDA and AVGO if AI capex commentary from hyperscalers decelerates for one quarter; that would be the fastest signal that expectations are outrunning near-term demand.