Nvidia Just Proved It Doesn't Need China Anymore

Nvidia guided for fiscal 2028 revenue growth of 70% y/y, citing a supply-constrained setup amid AI-chip bottlenecks (e.g., high-bandwidth memory, packaging, power, and land). The article contrasts this with Wall Street’s ~44% growth model (based on ~$397B fiscal 2027 revenue), implying a ~$100B revenue gap versus consensus expectations. It also says China exposure is limited (Hopper shipments to China were <1% of data-center revenue in the quarter and management expects no China data-center compute revenue in the forward outlook), while the stock is framed as modestly valued on a ~23x forward P/E and PEG ~0.6.

Analysis

The signal is less about a flawless demand story and more about who controls the bottleneck. If AI spend is now constrained by memory, packaging, and power, the economic rent shifts upstream to the scarce inputs, while NVDA increasingly behaves like the orchestration layer rather than the sole capacity gate. That is constructive for NVDA’s durability, but it also caps near-term upside because the market will start discounting supply-chain execution risk instead of just end-demand.

The second-order winner is any supplier exposed to AI assembly constraints, while the likely loser is the simple “China is required” bearish thesis. If China is already immaterial to the revenue mix, then the real debate is not geopolitical access but how fast non-hyperscale demand can absorb incremental supply without margin leakage. That broadening customer base also makes AMD’s share-gain narrative harder near term: when the leader is supply-limited, smaller competitors can win pilots, but they still need ecosystem scale and allocation to convert design wins into revenue.

Contrarian view: the market may be underestimating how much of this growth is being pre-booked rather than monetized, which means the next 1-2 quarters matter more than the headline 12-month outlook. If HBM or advanced packaging tightens further, NVDA’s top-line guide can hold while gross margin and delivery timing disappoint. Falsifiers are any indication of backlog conversion slowing, margin compression from component inflation, or a sudden re-acceleration in China exposure that proves the forward outlook was too conservative rather than structurally durable.

Key Decisions for Investors

• Stay long NVDA on a 3-5% pullback, with a 1-3 month horizon; use the thesis that supply scarcity is the real swing factor and the market is still underpricing forward revenue durability. Falsify if gross margin or delivery timing deteriorates in the next earnings print.
• Pair trade: long NVDA / short AMD over the next 1-3 months. NVDA has the stronger allocation power and pricing leverage; AMD remains more dependent on proving share gains convert outside pilot deployments. Exit if AMD wins a meaningful hyperscale socket announcement or NVDA guide is revised down.
• Add AVGO on weakness as a complementary AI infra expression for 6-18 months. If GPU bottlenecks persist, custom silicon and networking content should capture more of the capex budget. The risk is that hyperscalers delay ASIC ramps if NVDA supply eases faster than expected.
• Avoid shorting NVDA purely on China headlines. For the next quarter, the more relevant watch item is HBM/packaging lead times and any commentary on order mix from non-hyperscale buyers. A break in that chain is the first real bearish catalyst.