Nvidia’s Hidden $60 Billion Business Is About to Overtake Broadcom

Nvidia disclosed fiscal Q1 networking revenue of $15 billion, up 3x year over year and implying a $60 billion annualized run rate. The article argues this makes Nvidia a full-stack AI infrastructure leader, competing not just in GPUs but also against Broadcom, Arista, and Cisco in AI networking. The tone is constructive on Nvidia’s long-term positioning, though it notes valuation and AI spending slowdown risks.

Analysis

The market is still underestimating that AI networking is not a side business but the gating factor for cluster economics. If Nvidia can monetize the fabric as tightly as it monetizes compute, the implication is a much larger take-rate on AI capex: each incremental accelerator deployment now pulls through proprietary interconnect, systems, and software, compressing the addressable pool available to best-of-breed networking vendors. That creates a flywheel where Nvidia’s strongest competitive advantage is not raw performance, but the ability to subsidize hardware economics with stack-level pricing power.

The second-order loser is not just AVGO/ANET/CSCO revenue growth, but their negotiating leverage with hyperscalers. Once a customer standardizes on an integrated Nvidia stack for one generation of training clusters, switching costs rise because performance tuning, tooling, and procurement become bundled; that can push rivals into lower-margin “good enough” configurations or niche enterprise exposures. The more meaningful externality is to custom silicon programs and merchant switch markets: if Nvidia’s reference architecture becomes the default, it can slow adoption of alternative fabrics even when those alternatives are technically competitive.

Near term, the setup is about multiple expansion versus operating proof. The stock can keep re-rating over the next 1-3 months if networking revenue remains the cleanest evidence that Nvidia is becoming a full infrastructure platform, but the trade is vulnerable to any sign that AI capex is being deferred or that hyperscalers are diversifying network spend to avoid dependence. The bigger risk over 6-12 months is antitrust/regulatory scrutiny, not on GPU share alone, but on bundling and tying across compute, networking, and software.

The consensus is still too linear in its framing: it treats networking as additive to GPU demand rather than as the mechanism that expands total system spend and locks in ecosystem share. That said, the move may be partially overextended if investors extrapolate a $60B run rate without recognizing that networking is a lumpy, program-driven revenue stream and can grow less cleanly than semis once deployment cycles normalize. The best contrarian read is that Nvidia’s moat is real, but the next leg of outperformance likely comes from execution consistency, not surprise acceleration.

Key Decisions for Investors

• Go long NVDA on weakness over the next 2-4 weeks; favor a pullback entry versus chasing highs. Risk/reward is attractive if the market continues to pay up for platform-level monetization, but trim if the stock re-rates ahead of the next capex guide.
• Initiate a relative-value pair: long NVDA / short equal-dollar basket of AVGO + ANET + CSCO for 1-3 months. Thesis: Nvidia captures the highest-margin, most strategically sticky layer of AI networking; the pair isolates stack integration beta while reducing market directionality.
• For AVGO, use a hedged structure rather than outright shorts: buy puts or run a put spread 3-6 months out. Best case is a compression of multiple as investors realize its AI networking role is more exposed to hyperscaler bargaining power than Nvidia's bundled stack.
• Sell downside volatility in ANET only if you already hold core AI exposure; otherwise avoid chasing. The company remains high quality, but relative momentum may lag if customers standardize on integrated Nvidia architectures over the next two quarters.
• Watch for regulatory headlines and hyperscaler capex commentary as the key catalyst window over the next 30-90 days. If either indicates bundling scrutiny or networking spend diversification, reduce NVDA exposure and rotate into more defensive network names.