Nvidia’s (NVDA) first H200 shipments to China are unlikely to change its quarterly results. The reported volume is just too low.
But it matters to U.S. investors, since Nvidia’s current guidance assumes data-center compute revenue from China will be zero.
For the first quarter of fiscal 2027, Nvidia’s earnings report projects second-quarter sales of roughly $91 billion, plus or minus 2%.
Management’s choice to exclude China as a sales channel means the H200 sales channel is repeatable and might ultimately contribute revenue over the disclosed expectation. It also might expose shareholders to a familiar threat.
Previously, a surprise U.S. export ban on Nvidia’s H20 chip had led to the business booking a multibillion dollar inventory charge, CNBC reported. As a result, China policy can create earnings upside, but it can also erase profit with extraordinary rapidity.
This means the first shipments of H200 will not only be a geopolitical development. They are also an early test of whether Nvidia can turn a market it currently values at zero into an incremental source of sales, TechCrunch noted. The company must do this without losing its profits, interrupting supply for American customers, or creating another inventory problem.
The shipments won’t answer that question immediately. They do, however, bring China back into the earnings equation for U.S. stockholders.
“Our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide,” Nvidia said in a regulatory filing.
China could alter Nvidia’s earnings calculation
Nvidia is not dependent on China to fuel its expansion.
Revenue of $81.6 billion in the first quarter was up 85% from a year ago. In its most recent quarterly report to the Securities and Exchange Commission, Nvidia reported that data-center revenue rose 92% to $75.2 billion.
Its gross margin came in at 75% on a non-GAAP basis, demonstrating the profitability of the AI infrastructure boom.
Those results make the H200 development more of an optionality narrative than a rescue story.
Nvidia has already demonstrated that demand from U.S. cloud providers, sovereign AI projects, enterprises, and specialized AI companies can drive substantial growth without Chinese data-center compute sales.
On top of the existing momentum, a working H200 license path may be an additional demand source, Reuters confirmed.
The extent of the potential is still uncertain. Nvidia characterized the first shipments as “very few” and has not set a China revenue goal for the initiative. But the company offers a clear starting point for investors by assuming no China data-center compute revenue.
Small cargoes will be of little use. A recurrent sales channel may not necessarily be as rosy as it seems, if that is the case.
Related: Nvidia partner sued over five critical products
History explains why investors need to look beyond anticipated revenue.
In April 2025, the U.S. government said exports of its China-focused H20 processor will require a license, South China Morning Post indicated. The company later wrote down $4.5 billion for excess inventories and purchasing obligations.
The business also stated it couldn’t ship an extra $2.5 billion of H20 revenue in the quarter, according to its fiscal first-quarter 2026 results.
Without the penalty, its non-GAAP gross margin would have been 71.3% instead of the reported 61%, Nvidia said.
That instance showed how quickly export policy may flow from Washington into the calculation of an American investor’s earnings per share.
A solid H200 process might lessen such volatility by giving Nvidia a legal way to sell certified goods against existing demand in China.
An unstable process could repeat the H20 problem. Nvidia may commit to production based on customer interest, only to face a rule change that leaves it with inventory it cannot ship to its intended market.
Then, for U.S. stockholders, the China possibility needs to be viewed along two lines: how much income it generates, and how reliably Nvidia can produce it.
Nvidia’s H200 route is narrow by design
In January, the Commerce Department amended its export policy to allow case-by-case inspections for applications for similar processors such as Advanced Micro Devices’ (AMD) MI325X and Nvidia’s H200.
The Bureau of Industry and Security strategy did not open China up without limits.
License approval is contingent on exporters and buyers meeting security, compliance, and supply-related conditions.
The comprehensive Federal Register rule is of special interest to U.S. investors. Applicants must attest that Chinese shipments will not delay orders for U.S. customers or divert semiconductor production capacity that otherwise would serve the American market.
More AI:
- The new Chinese AI model rattling U.S. tech investors
- Anthropic restores access to Mythos 5 for select organizations
- SoftBank CEO offers stinging critique of Musk’s AI bet
Aggregate H200 shipments to China and Macau also cannot exceed 50% of the number shipped for use in the U.S. Each shipment must undergo qualified third-party testing in the U.S. before export.
And they limit the potential pace of the China opportunity. The latest customers cleared to buy sophisticated chips from Nvidia and AMD include a unit of ZTE and two other Chinese firms, Reuters said. Reports confirmed that about 10 companies, including Alibaba, Tencent, and privately held ByteDance, were certified earlier.
The initial deliveries demonstrate that the licensing route can operate, although they haven’t yet shown it can run quickly or consistently enough to make a difference in Nvidia’s earnings.
The method also has an associated expense. H200 processors covered must be tested in the U.S. before export. The inspection requirement exposes the chips to a 25% tariff, and Nvidia has said it may not be able to pass the full cost on to customers.
The tariff is based on a White House semiconductor proclamation that puts a 25% levy on select advanced semiconductor imports that don’t qualify for stated domestic-use exemptions.
This means headline sales from China may not create the same gross profit as a comparable transaction elsewhere.
Shipment volume is just one metric for shareholders, however. The more crucial estimate is how much profit remains after tariffs, inspections, compliance fees, and any discounts required to entice buyers to buy an older architecture.
The H200 is still a forceful processor. It features 141 gigabytes of high-bandwidth memory and 4.8 terabytes per second of memory bandwidth for demanding generative AI and large-language-model applications.
But competition is advanced.
Advanced Micro Devices is selling its MI325X accelerator with 256 gigabytes of memory and 6 terabytes per second of bandwidth. Technical specs alone may not determine adoption, but they do suggest Chinese customers have reasons to consider alternatives.
That urge to compete means time is key. The licensing process is long, and while it may allow for individual shipments, it may not provide Chinese customers the certainty to plan multiyear data-center projects.
A slow process can keep customer interactions going until Nvidia can seek clearance for fresh products.
What Nvidia investors should watch next
The first signal is shipment size.
A few deliveries won’t impact Nvidia’s $91 billion outlook. Investors want proof that the company is regularly obtaining licenses and that shipments are becoming recurring, rather than symbolic.
The second indication is advice.
If sales in China become significant, Nvidia may start mentioning data-center computing revenue from the nation in its quarterly outlook. Until then, investors should see the offer as upside potential, not guaranteed revenue.
The third indication is gross margin.
A China transaction that incurs a 25% tax, inspection charges, and other compliance costs may not be as appealing as it seems in terms of revenue contribution.
Nvidia’s margin comments will help investors assess whether H200 sales represent profitable growth or simply a method to keep a toe in the door with customers.
The fourth indication is progression of products.
The H200 is based on top of Nvidia’s Hopper architecture, and the company’s Blackwell platform is at the core of its latest artificial intelligence systems.
The China possibility cannot be a sustainable earnings driver if the performance gap between permitted CPUs and Nvidia’s leading offerings grows too much.
Investors should monitor to see if Washington will finally allow limited editions of newer designs, or if it will keep China several product generations behind.
Key takeaways for Nvidia shareholders
- Nvidia’s current outlook assumes no revenue from data-center computing in China.
- The initial H200 shipments are too small to materially affect near-term results.
- Recurring licensed sales could add revenue beyond management’s forecast.
- The 25% tariff and compliance costs could reduce the profitability of China sales.
- Export policy reversals have already caused Nvidia a $4.5 billion inventory charge.
- China access may help protect Nvidia’s global software and developer ecosystem.
The last issue is policy durability.
The sales channel for the H200 is due to Washington choosing to consider applications on a case-by-case basis. Another regulatory decision might squeeze or shut that channel before Nvidia has reliable revenue.
Chinese regulators also can affect the outcome by slowing imports, boosting domestic buying or limiting the use of American processors.
Nvidia is basically trying to balance two governments with different strategic aims. Washington wants to protect U.S. national security and prevent China from obtaining unrestricted access to advanced computing. Beijing wants to reduce its reliance on American technology. Nvidia, meanwhile, wants to serve customer demand without losing access to either market.
That tension likely won’t go away.
But for U.S. investors, the initial H200 shipments shift the financial equation, Investing.com notes.
China is now a market Nvidia can touch, and the regulatory risk of inventory charges has decreased. It’s also a potential revenue stream management did not include in its forecast.
The opportunity should not be overplayed.
The shipments are still small, the tax could squeeze profits, and another policy change could soon erase the upside.
But the investment message is now explicit. Nvidia has found a narrow path to convert zero expected China revenue to incremental sales while keeping Chinese developers engaged with its platform.
The first shipments will not affect the quarter. Still, a strong licensing channel might alter the calculation for Nvidia’s earnings.
Related: Bank of America’s unmistakable signal to Nvidia stock investors
