Nvidia’s China Zero Becomes Both a Valuation Drag and an Earnings Option

Nvidia has reset the market’s China debate. Investors no longer have to underwrite Chinese data-center compute sales to believe in the AI boom. The company’s fiscal second-quarter results and $108.0 billion fiscal third-quarter revenue outlook excluded any assumed China data-center compute revenue, even as Data Center revenue reached $89.0 billion, up 117% from a year earlier.¹

That caveat matters because it makes China two things at once. It remains a valuation overhang, with U.S. and Chinese policy still limiting Nvidia’s access to one of the world’s largest AI markets. But it is also becoming an out-of-the-money earnings option: if Washington and Beijing allow a meaningful category of compliant sales, incremental China revenue would not need to rescue guidance. It could add to a framework that already assumes zero.

The stock-market implication is subtle but important. Nvidia is telling investors that non-China demand from hyperscalers, AI clouds, enterprises, sovereign buyers and frontier labs is enough to support another revenue step-up. CFO commentary filed with the SEC showed shipments of Data Center Hopper products to China were less than 1% of Data Center revenue in the July quarter, while management said there was no China data-center compute revenue in the forward outlook.³

The earnings model has become less dependent on China. The multiple, however, may still be capped by uncertainty over whether Nvidia can recover durable access to the market.

The base case is no longer China-dependent

Nvidia reported fiscal second-quarter revenue of $96.2 billion, up 106% year over year, and non-GAAP diluted EPS of $2.22. Its Data Center segment, now the company’s dominant driver, generated $89.0 billion of revenue, up 18% sequentially and 117% from a year earlier.¹ The company’s Form 8-K furnished the results and CFO commentary to the SEC on Aug. 26, making the no-China caveat part of the official disclosure package presented to investors.²

The near-term guide was even more striking. Nvidia projected fiscal third-quarter revenue of $108.0 billion, plus or minus 2%, and explicitly stated that the forecast does not include Data Center compute revenue from China.¹ On the earnings call, management also said it expects fiscal 2028 revenue to grow roughly 70% year over year, while supply remains a bottleneck at least through the end of that fiscal year.⁵ Reuters reported that the annual growth forecast was unusually long-range for Nvidia and well above analysts’ average expectation for the period.⁷

For investors, the message is that the central constraint is not China demand. It is supply, memory, power, land, data-center shells and financing. Nvidia’s 10-Q said supply and capacity commitments rose to $279 billion as of July 26, 2026, from $119 billion in the prior quarter, primarily for data-center infrastructure systems, memory and manufacturing capacity.⁴ That supply-chain posture supports the view that the company sees demand well beyond the restricted Chinese market.

Why China still hangs over the multiple

China has not disappeared as a risk. Nvidia’s own risk disclosure says the company is effectively foreclosed from competing at scale in China’s data-center compute market under the current export-control and geopolitical environment.⁴ That is not a normal regional slowdown. It is a policy-imposed structural limit on a market where Chinese cloud providers, AI labs and state-backed buyers would otherwise be large potential customers.

The problem is compounded by the need for approval on both sides. Nvidia said it is unable to create and deliver a competitive product for broad China data-center distribution that satisfies both U.S. and Chinese authorities under current rules.⁴ Even when U.S. licenses allowed limited H200 shipments to specific China-based customers, the company said PRC restrictions limited sales. The resulting shipments accounted for less than 1% of Data Center revenue in the most recent quarter.⁴

That leaves investors with a difficult valuation question. If China remains closed, Nvidia can still grow rapidly, but part of its historical and theoretical total addressable market is impaired. If access reopens only in narrow, low-margin or tariff-burdened forms, revenue may return without fully restoring profitability. Nvidia disclosed that licensed H200 shipments require U.S. inspection and are subject to a 25% tariff upon importation into the United States, while management said current Hopper shipments to China are dilutive to corporate gross margins.⁴⁵

This is why China can weigh on valuation even when it does not weigh on guidance. Multiples reflect terminal market access, durability of competitive position and geopolitical tail risk, not just next quarter’s revenue line.

The upside case is cleaner because guidance starts at zero

The other side of the debate is that a zero-China outlook creates a cleaner upside setup. Kiplinger cited market commentary noting that Nvidia’s guidance excludes China data-center compute because of government restrictions, making any future contribution from that category potential upside.¹⁰

That does not mean China reopening is likely or imminent. It means the earnings debate has changed. In earlier phases of the export-control cycle, investors had to ask how much China deterioration was embedded in estimates. Now, for the data-center compute line, Nvidia has made the answer explicit: none is assumed in the forward outlook.¹³

If compliant products can be sold at meaningful scale, the impact would be incremental to a base case already supported by a $108.0 billion quarterly revenue guide and a 70% fiscal 2028 growth expectation.¹⁵ That could force investors to revisit both earnings estimates and the probability weighting assigned to Nvidia’s long-term China opportunity.

But the quality of any upside matters. A narrow license regime that produces small shipments, elevated inspection costs, tariffs or gross-margin dilution would be less valuable than broad approval for a competitive product. Nvidia’s disclosure makes clear that the China problem is not simply demand. It is whether the company can ship a product that customers want and regulators on both sides will permit.⁴

Policy risk is moving from chips to compute access

The latest policy signals suggest Washington may keep broadening the definition of restricted access. Tom’s Hardware reported that the Trump administration is considering controls aimed at Chinese remote access to advanced AI compute through servers in nearby countries such as Thailand and Singapore, with a rule potentially shared with industry groups as early as September.¹³ That would extend the policy question beyond physical chip shipments to who can use Nvidia-class compute once it is installed elsewhere.

That matters for Nvidia because the company’s growth model increasingly depends on global AI factories, neoclouds and sovereign infrastructure. If export controls follow usage rather than hardware location, compliance complexity could rise for cloud customers and data-center operators serving multinational demand.

Enforcement pressure is also visible. Taiwanese prosecutors charged nine people, including individuals from Nvidia and Super Micro, over alleged illegal exports of high-end AI servers to mainland China, according to the Associated Press.¹² Such cases strengthen the case for tighter monitoring and know-your-customer regimes, even if regulators loosen some direct-sales pathways.

For investors, the China option is not a simple binary. A reopening of compliant chip sales could add revenue, but tougher rules on remote access, diversion risk and end-use verification could limit the scale or timing of that upside.

Non-China demand is the stronger signal

The most bullish read from the quarter is that Nvidia’s non-China demand is strong enough to carry the story. The company said Blackwell Ultra drove the July-quarter Data Center ramp, while hyperscale revenue more than doubled year over year and AI Clouds, Industrial and Enterprise revenue rose 138%.⁴ Management also said Vera Rubin had begun production shipments and could account for about 20% of Data Center revenue in the fiscal third quarter.⁵

That demand strength changes how investors should frame China. It is no longer the swing factor needed to validate the next few quarters of growth. Instead, it is a source of uncertainty around long-term market share, competitive ecosystems and geopolitical risk.

This distinction matters because Nvidia’s valuation rests on two overlapping claims: first, that AI infrastructure spending remains durable; second, that Nvidia can continue capturing a large share of that spending. The July quarter strongly supported the first claim outside China. The second remains exposed to export controls, China’s domestic accelerator push and the possibility that restricted markets help local competitors build developer ecosystems.

Nvidia explicitly warned that being foreclosed from China’s data-center compute market helps competitors build larger customer and developer bases that could challenge it globally.⁴ That is the deeper overhang: lost China revenue today could become lost ecosystem influence tomorrow.

The investor takeaway

China is both less important and more interesting after Nvidia’s Aug. 26 report. It is less important because Nvidia’s current growth algorithm — $96.2 billion of quarterly revenue, $89.0 billion of Data Center revenue and a $108.0 billion next-quarter guide — does not require China data-center compute to beat expectations.¹ It is more interesting because any policy-compliant reopening would now arrive against a zero-revenue assumption.

The base-case conclusion is that Nvidia has largely de-risked near-term estimates from China. The valuation conclusion is less clean. A closed China market still reduces total addressable market and increases competitive risk. A partially reopened China market could add upside, but only if the products are competitive, approvals are reliable and margins are acceptable.

For public-market investors, the right framing is not “overhang or catalyst.” It is both. Nvidia has shown the AI boom can keep compounding without China. The next reset in the stock may come if investors decide China is no longer a hole in the model, but a call option on top of it.

Nvidia’s China Zero Becomes Both a Valuation Drag and an Earnings Option

More from Siften Finance

Baowu’s Jimblebar Interest Is About Commodity Security, Not BHP Housekeeping

Things to know

Comments

Source Material

The base case is no longer China-dependent

Why China still hangs over the multiple

The upside case is cleaner because guidance starts at zero

Policy risk is moving from chips to compute access

Non-China demand is the stronger signal

The investor takeaway

Key topics covered

UK Retailers Face Footfall and Margin Squeeze Before October Budget

Spire’s £1.03bn take-private bid puts UK healthcare valuations to the test

China’s Treasury Buying Is a Carry Trade With a Currency Signal