Here’s Why Brad Gerstner Remains Bullish on NVIDIA (NVDA) Despite Valuation, Bubble Fears

NVIDIA is described as benefiting from scarcity of high-quality AI chips, strong hyperscaler demand, and management's view of total sales potentially reaching $1 trillion through 2027. The article cites Blackwell revenue of $184 billion in 2025 and an expected $320 billion in 2026, while also noting ongoing valuation and bubble concerns. Overall, the piece is constructive on NVDA’s long-term growth outlook but is primarily commentary rather than a new company-specific catalyst.

Analysis

The market is still underpricing how a true supply-constrained AI platform shifts value capture from unit shipments to system-level economics. If compute remains scarce, the margin pool should migrate further upstream into networking, interconnect, power delivery, and software attach rates, which means beneficiaries extend beyond the chip vendor to a tighter cluster of infrastructure suppliers with better operating leverage. That also raises the bar for would-be competitors: they do not just need a better accelerator, they need an ecosystem that can scale deployment velocity under real-world power, cooling, and integration constraints.

The second-order winner is hyperscale capex quality, not just quantity. As spending rises, the key question becomes return on deployed capital over 12-24 months; if customers start seeing diminishing incremental model performance or slower monetization, the procurement cycle can elongate even while headline budgets stay elevated. In that scenario, the market could rotate from pure semiconductor beta into picks-and-shovels names tied to data center buildout, while high-multiple application-layer AI stocks remain vulnerable to any evidence that inference economics are not improving fast enough.

The main risk is that the bullish narrative becomes self-defeating: if everyone prices in perpetually scarce supply and aggressive pricing power, expectations can outrun shipment cadence and gross margin realization. The next reversal point is likely not a demand collapse but a digestion phase over the next 2-3 quarters, when customers rationalize inventory, redeploy capex toward internal efficiency, or push harder on multi-sourcing and custom silicon. If that happens, the stock can still be fundamentally strong while multiple compression does most of the damage.

Contrarianly, the market may be too focused on the chip monopoly and too little on the eventual decentralization of AI infrastructure spend. Once model training becomes more standardized and inference dominates usage, pricing power can diffuse toward network, power, packaging, and software orchestration layers rather than pure GPU scarcity. That argues for owning the enablers of scale and hedging the highest-duration beneficiaries if the trade has already crowded into a single-name AI consensus.

Key Decisions for Investors

• Maintain a core long NVDA, but finance it with short-dated covered calls into the next 1-2 quarterly catalysts; the setup favors upside continuation, but implied volatility is likely overpriced relative to the probability of a digestion phase.
• Pair trade: long AVGO / short a basket of high-duration AI software names over 3-6 months; if AI capex stays strong, semis and infrastructure monetize first, while software re-rating depends on proof of downstream ROI.
• Add exposure to data center infrastructure beneficiaries over the next 1-2 quarters, especially networking/power/cooling supply chain names; these should have better earnings durability if GPU scarcity persists.
• Use any post-earnings gap up in NVDA to trim 10-20% of the position and redeploy into less consensus AI infrastructure; reward-to-risk deteriorates once the market fully capitalizes the 2026-2027 growth path.
• For tactical traders, buy 3-6 month put spreads on NVDA as a hedge against capex digestion or inventory normalization; the thesis is not a collapse, but multiple compression can still produce a 10-15% drawdown.