Nvidia May Have Just Eliminated Its Biggest Risk

Nvidia reported revenue jumping 106% to $96B and profit surging 126% to $59B, reflecting a continued earnings surge tied to AI compute demand. Despite competitive concerns (AMD, Intel, and customer in-house chips like Amazon), Nvidia reduced the risk by expanding its AWS partnership: AWS agreed to buy an additional 2 million GPUs over the next two years (and potential millions of CPUs). The article frames this as evidence of Nvidia’s platform staying central to major customers’ AI buildouts, supporting a bullish “buy and hold” stance for NVDA.

Analysis

The market mechanism here is not “more AI demand,” it is de-risking the bear case that hyperscalers will rapidly substitute away from Nvidia. Large cloud buyers are still willing to allocate incremental budget to Nvidia even while they build internal silicon, which implies Nvidia remains the gating technology for frontier training and high-end inference. In the next 1-3 months, that should support the multiple more than the revenue line; the order book signal matters because it weakens the argument that custom chips will cap Nvidia’s growth rate this cycle.

Second-order, Amazon is the subtle winner: custom chips lower its own compute cost while Nvidia keeps absorbing the hardest workloads, so AWS can arbitrage both paths instead of choosing one. That is more bearish for AMD and, to a lesser extent, INTC than for Nvidia, because they need share loss at the margin to justify a rerating. If Nvidia keeps expanding the “platform” beyond GPUs into CPUs/networking, the addressable basket widens and the competitive threat becomes more about workload mix than outright displacement.

The contrarian risk is that investors may overinterpret an order announcement as proof of durable pricing power. The key falsifier over 1-2 quarters is any slowdown in data-center growth or gross margin compression from mix, rebates, or elevated supply chain costs. Over 6-18 months, the real threat is that hyperscalers reserve Nvidia for training but migrate inference to in-house silicon, which would keep revenue growing but pressure the long-duration multiple.

Key Decisions for Investors

• Long NVDA on any 3-5% pullback into the next earnings/capex cycle; use shares or a 1-3 month call spread to express upside from reduced customer-cannibalization fears.
• Pair trade: long NVDA / short AMD in beta-adjusted size. Rationale: Nvidia is proving it remains the preferred supplier even to customers building custom silicon, while AMD’s relative share-gain narrative is more vulnerable to disappointment.
• Modest long AMZN as a secondary beneficiary of AI capex intensity; AWS can monetize demand with both internal silicon and Nvidia, preserving margin flexibility. Best entered on any post-news weakness rather than chasing strength.
• Set a watch item on NVDA gross margin and data-center growth in the next two quarters; if either shows meaningful deceleration, reduce exposure because the market will likely reprice the platform premium quickly.