Nvidia: Trading At A Rare Discount Just As Its Next AI Supercycle Kicks Off

The near-term setup still favors NVDA because scarcity does more than lift ASPs: it reduces customer bargaining power and shortens sales cycles, which tightens revenue visibility into the next two product turns. The more important second-order effect is that a clean ramp signal at the next platform should pull forward capex commitments from hyperscalers that have been waiting for proof-of-supply, potentially steepening orders into the next 2-3 quarters rather than just preserving current demand.

The competitive loss is not just share leakage to Chinese silicon, but a gradual erosion of NVDA’s default status in local inference workloads. Once a domestic stack gets to roughly 40% share, the adoption curve can become self-reinforcing through software localization, procurement policy, and system integration, even if raw performance still trails. That means BABA’s AI exposure should be treated less as a single-chip winner and more as an enabler of a national ecosystem that can capture workload share at the margin for years, especially in sovereign and regulated deployments.

The market is likely underpricing the option value of the next-gen ramp while overestimating the immediacy of the China threat. China can pressure the low-to-mid end first; it is much harder to displace NVDA in frontier training and the highest-value cluster budgets, so the share loss path is probably gradual rather than cliff-like. The real risk to the bullish NVDA view is not competition alone, but any evidence that the new platform ramp is supply-constrained or slips by even one quarter, because that would convert scarcity from a pricing tailwind into a shipment ceiling.