Nvidia Stock Surges on H200 Clearance, but China Halts Orders

Nvidia CEO Jensen Huang sat at the Trump-Xi summit table on 14 May 2026 as the U.S. Commerce Department simultaneously cleared H200 GPU sales to ten Chinese technology firms, a convergence of diplomacy and commercial policy that sent the chipmaker’s stock surging toward a approximately $5.53 trillion implied market capitalisation in premarket trading. The clearance names four confirmed buyers, Alibaba, Tencent, ByteDance, and JD.com, alongside six additional firms, reopening a revenue channel that had been effectively shut since January’s export restrictions took hold. Yet the headline optimism carries a complication: China has independently halted H200 purchases despite Washington’s green light, introducing a bilateral standoff that explains why the premarket gain was measured rather than euphoric. What follows covers the exact scope of the authorisation, the technical limits that define it, the semiconductor peers riding the same wave, and the policy risks that investors pricing in durable China revenue should weigh carefully.

Nvidia closes in on approximately $5.53 trillion as H200 export clearance breaks during summit

Nvidia closed at $225.83 on 13 May 2026, up 2.29%, before pushing to $230.29 in premarket trading on 14 May, a further 1.97% gain on volume exceeding 150 million shares. At that premarket price, the implied market capitalisation approached approximately $5.53 trillion.

The implied market capitalisation of approximately $5.53 trillion would represent the highest valuation any publicly traded company has reached, surpassing Nvidia’s own prior record.

The catalyst was specific: the U.S. Commerce Department cleared H200 GPU sales to ten named Chinese companies. Four, Alibaba, Tencent, ByteDance, and JD.com, were confirmed publicly. Six additional firms were included in the authorisation, per Reuters and U.S. News reporting from the same day.

Jensen Huang at the summit table

Huang joined the U.S. delegation at the Trump-Xi summit, with Reuters reporting the mission focused on AI chip access and American technological leadership. A corporate CEO seated at a heads-of-state negotiation signals how tightly commercial semiconductor policy has become woven into geopolitical strategy. The last time chip exports carried this level of diplomatic weight was during the original October 2022 restrictions; this time, the CEO of the company most affected was in the room.

The asymmetric risk profile for semiconductor stocks at the summit was well-flagged before Thursday’s session opened: Bloomberg Intelligence had estimated a 70% probability of partial de-escalation, but the same scenario matrix placed the potential downside at roughly double the projected upside if talks hardened.

Why the H200 is the only Nvidia chip China can legally buy right now

The Bureau of Industry and Security (BIS) published technical thresholds in January 2026 that determine which chips can and cannot be exported to China. Two numbers matter: a total processing performance (TPP) ceiling of 21,000 and a DRAM bandwidth cap of 6,500 GB/s. Any chip exceeding either threshold is prohibited.

The H200’s specifications place it comfortably within both limits. Its memory bandwidth of 4.8 TB/s (approximately 4,800 GB/s) sits well below the 6,500 GB/s cap, and its TPP rating falls under the 21,000 ceiling. It carries 141GB of HBM3e memory, making it the most capable AI chip legally exportable to China under current rules.

The H200 HBM3e memory specifications confirm 4.8 TB/s of memory bandwidth and 141GB of total capacity, placing the chip comfortably within the BIS export thresholds while making it the most capable AI accelerator currently accessible to Chinese buyers under the January 2026 framework.

Nvidia’s newer architectures do not qualify. Both the Blackwell (B200) and Rubin (R100) exceed both thresholds and remain prohibited.

The approved chip, in other words, is two generations behind Nvidia’s current frontier hardware. The commercial upside from this clearance is real, but it is not uncapped.

China’s counter-move complicates the clearance picture

The U.S. authorisation was met with an unexpected response from Beijing. China independently halted H200 purchases and imposed its own import restrictions, according to WION reporting on 14 May 2026. The bilateral standoff, American clearance paired with Chinese restriction, goes a long way toward explaining the 1.97% premarket gain rather than a larger surge. Markets are pricing in genuine uncertainty, not just headline optimism.

Beijing’s customs block on H200 imports has prevented any commercial shipments from completing even where U.S. licences exist, a distinction that matters for investors trying to separate authorised revenue from recognised revenue on Nvidia’s income statement.

The January 2026 framework had already placed conditions on any H200 sales to China:

• A 25% revenue surcharge on all cleared exports
• A 50% volume cap limiting total shipments
• A mandatory third-party U.S. verification requirement before any shipment

ByteDance committed upwards of $14 billion (approximately 100 billion yuan) in chip spending for 2026, according to Reuters, establishing the scale of Chinese demand now caught between American authorisation and Beijing’s counter-restriction.

Chinese AI chip orders for the H200 were estimated at more than 2 million units, against Nvidia’s approximate 700,000-unit inventory (this figure should be treated as an estimate). The demand exists. Whether supply can legally and physically reach it remains unresolved.

Semiconductor peers catch the wave: Marvell, Micron, and ON Semiconductor

The H200 clearance lifted names well beyond Nvidia. ON Semiconductor posted the largest single-session gain of the group, climbing 11.14% to $115.71 on volume of approximately 21.55 million shares. Marvell Technology gained 3.29% to close at $183.80 on 31.86 million shares, while Micron rose 4.83% to $803.63 on 54.88 million shares. Alibaba, as both a named buyer and a proxy for the China AI demand thesis, gained 8.18% to $145.81 on 40.21 million shares.

AMD and Intel carry materially smaller China AI exposure than Nvidia, limiting their relative upside from this specific policy development. The broader market reflected where the enthusiasm was concentrated:

• S&P 500: 7,444.25 (+0.58%)
• Nasdaq Composite: 26,402.34 (+1.20%)
• Dow Jones: 49,693.20 (-0.14%), illustrating the AI-concentrated nature of the day’s gains

The breadth of the semiconductor sector response confirms this is being read as a sectoral catalyst, not a single-stock event.

The policy logic behind controlled H200 access and its durability risks

White House AI Czar David Sacks articulated the administration’s rationale: controlled sales are preferable to total denial because denial accelerates Chinese domestic chip development, particularly by Huawei and state-backed alternatives. The strategic goal is to keep American companies competitive in the Chinese market while slowing Beijing’s path to chip self-sufficiency.

Bernstein estimated that without the licensing shift, Nvidia’s share of the Chinese AI chip market could fall below 10% (approximately 8%) by end of 2026, as Chinese firms pivoted to domestic alternatives.

Former U.S. national security officials have offered a counter-argument: the approved H200 volume could meaningfully expand China’s AI compute base, potentially providing capabilities comparable to what leading U.S. frontier AI firms currently hold. The policy may preserve Nvidia’s revenue while simultaneously building the capability it aims to constrain.

What makes this framework fragile

Three specific risks undermine the durability of the current export arrangement. First, the transactional nature of the Trump-era framework means policy could shift with negotiating dynamics; what is cleared today may be restricted next quarter if summit outcomes disappoint. Second, enforcement credibility remains uncertain. Chinese firms have historically accessed restricted technologies through intermediaries, and the third-party verification requirement is the primary compliance mechanism. Third, the paradox at the framework’s core: large-scale approved shipments may accelerate the very domestic Chinese chip capability the policy intends to contain.

What the H200 clearance signals for Nvidia’s second half and beyond

The analyst consensus holds a Buy rating on Nvidia with an outperformance thesis for the second half of 2026.

Analyst consensus: Buy rating with outperformance expected in H2 2026, supported by data centre demand acceleration and the reopened China revenue channel.

That baseline view, however, depends on the bilateral standoff resolving into actual shipments, not just authorisation on paper. Three variables remain unresolved:

• The bilateral standoff: China’s import halt must be lifted before any cleared H200 units can ship
• Inventory constraints: Chinese orders of more than 2 million H200 units against Nvidia’s estimated 700,000-unit inventory imply a roughly 3:1 demand-to-supply gap that caps near-term revenue upside regardless of policy
• Policy durability: The transactional nature of the export framework introduces long-term planning risk for both Nvidia and its Chinese buyers

The two-day Trump-Xi summit (14-15 May 2026) means further developments are possible within 24-48 hours of publication. Nvidia remains in active focus through the remainder of the week.

Investors wanting to stress-test the durability of China upside against Nvidia’s actual revenue trajectory will find our deep-dive into Nvidia’s China revenue structure examines how data centre revenue collapsed from $6.0 billion in FY2024 to near zero by FY2026 Q1, why institutional investors at BlackRock and Goldman Sachs treat this market as a call option rather than a base-case driver, and what Huawei Ascend’s estimated 70-80% domestic market share implies for any recovery timeline.

A week to shape AI chip policy, with the summit still unfolding

The U.S. authorisation of H200 sales to ten Chinese firms at a diplomatic summit represents the most consequential Nvidia export policy moment since the January 2026 licensing framework was introduced. It reopens a revenue door, but it does not guarantee shipments. China’s counter-restriction means the commercial opportunity is authorised but not yet open, and investors should monitor summit outcomes and Bureau of Industry and Security follow-up communications through the remainder of the 14-15 May window.

Nvidia’s official communications and Commerce Department updates will be the most reliable signals of whether this week’s authorisation translates into realised revenue or remains a headline without a delivery date.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions. Forward-looking statements regarding Nvidia’s revenue outlook and policy durability are speculative and subject to change based on market developments and diplomatic outcomes.