Nvidia Hits $5.53 Trillion as U.S. Clears H200 Sales to China

Nvidia crossed approximately $5.53 trillion in market capitalisation on 14 May 2026 after U.S. authorities quietly handed roughly ten Chinese technology firms permission to purchase the H200, an AI chip that had been locked out of China for months. The approvals arrived on the opening day of the Xi-Trump summit in Beijing, framing the move as both a commercial milestone and a diplomatic signal. For investors who have watched Nvidia navigate a maze of export restrictions since early 2025, the licenses represent a measurable shift in the geopolitical calculus around AI hardware.

What follows breaks down exactly which companies were approved, what the licenses are worth in revenue terms, what Wall Street’s named analysts are saying today, and what the continuing restrictions on next-generation chips mean for how long this tailwind lasts.

The ten Chinese companies now cleared to buy Nvidia’s H200

The U.S. Department of Commerce has authorised the following ten firms to purchase Nvidia’s H200 AI chips under the case-by-case licensing regime, with initial orders capped at 50,000 H200 units per company for Q2-Q3 2026:

• ByteDance (social media and AI platform operator)
• Alibaba Cloud (cloud computing and enterprise AI)
• Tencent Holdings (gaming, social media, and cloud services)
• Baidu (search, autonomous driving, and AI software)
• Huawei (telecommunications and semiconductor design; scope disputed)
• SenseTime (computer vision and AI research)
• iFlytek (speech recognition and natural language processing)
• Cambricon Technologies (AI chip design)
• Biren Technology (GPU design)
• Moore Threads (GPU design)

ByteDance reportedly requested approximately 450,000 units, with roughly 200,000 approved in the initial tranche.

Disputed: Pending official clarification. The scope of Huawei’s approval remains contested. Some reports indicate the licence is limited to the HiSilicon semiconductor design unit only, not the broader Huawei corporate entity. The precise terms have not been officially resolved.

At a midpoint price of $35,000 per unit, the estimated hardware value for known approved volumes is as follows:

Alibaba (BABA) gained 8.18% on 14 May, reflecting the market’s reading of which Chinese equities benefit most from restored chip access.

From outright ban to licensed access: how U.S. chip policy got here

Today’s approvals are the product of a specific chain of policy reversals, not a sudden shift. The arc from outright ban to licensed access unfolded in six stages:

The asymmetric risk profile facing semiconductor stocks at the summit was quantified before the session opened: Bloomberg Intelligence placed a 70% probability on partial de-escalation while projecting only 8-12% upside for that scenario, against a 25% probability hardening case carrying 10-20% downside for the SOXX and SMH.

1. January 2025: The Biden administration’s AI Diffusion Rule banned H200 and H100 exports to China.
2. Mid-2025: Nvidia pivoted to the H20, a modified chip designed to comply. The H20 was then halted in August 2025 under further restrictions, recording zero sales in Q2 FY2026.
3. Approximately 15 January 2026: The Trump administration reversed the Biden-era bans, enabling case-by-case H200 licensing with conditions: 25% tariffs, 50% domestic supply priority, U.S. testing mandates, and performance caps (TPP less than 21,000; memory bandwidth less than 6,500 GB/s).
4. March 2026: AMD received approval for MI300X exports to five Chinese firms, estimated at approximately $2.5 billion in value.
5. April 2026: Intel’s Gaudi3 AI chips were cleared for ByteDance and Baidu, capped below 15,000 TPP.
6. 14 May 2026: First mass H200 licenses issued to approximately ten Chinese firms, timed with the Xi-Trump summit.

The BIS revised licensing policy for semiconductor exports to China, published on 13 January 2026, established the case-by-case review framework under which today’s H200 approvals were processed, including the performance caps, domestic supply priority requirements, and end-use compliance conditions that define each licence’s scope.

What the B200 ban means for Nvidia’s China ceiling

The B200/Blackwell architecture remains fully banned for export to China as of May 2026. The H200 sits approximately two generations behind Nvidia’s current leading-edge product. That context matters: the commercial upside from these approvals is real, but it is bounded. Washington is calibrating access, not abandoning controls, and that distinction sets a ceiling on how much China revenue Nvidia can recover through this licensing regime.

What the H200 is and why Chinese AI labs need it

The H200 is Nvidia’s second-most capable AI processor available for commercial sale, positioned between the older H100 and the banned Blackwell B200 in performance terms. Priced at $30,000-$40,000 per unit, it is designed for large-scale model training and inference at data centre scale.

The demand from ByteDance, Alibaba, Baidu, and Tencent is business-driven. Each firm is competing in a domestic large language model (LLM) and generative AI market that is expanding rapidly, and each needs GPU-class hardware to train and deploy models at competitive scale. A large language model is an AI system trained on vast amounts of text data to generate and understand language. Without access to chips at the H200’s performance tier, these companies face a hardware bottleneck that slows development timelines.

The approved performance caps constrain what buyers actually receive:

• What H200 enables: Large-scale model training, data centre inference, multi-model deployment
• What the licence restricts: Total Processing Performance capped below 21,000; memory bandwidth capped below 6,500 GB/s

Chinese buyers receive a deliberately constrained version of the chip’s full capability, making the demand structural (driven by competitive necessity) rather than opportunistic.

What Wall Street analysts are saying about the revenue upside

The analyst consensus is genuinely bullish. Three named strategists raised price targets within hours of the approval news.

Moore cited the H200 clearance as adding meaningful FY2027 revenue runway. Sur forecast substantial China H200 revenue in H2 2026, with meaningful earnings-per-share uplift. Arya revised Nvidia’s FY2027 revenue estimate upward, with China contributing roughly 12% of the revenue mix.

Goldman Sachs characterised the approval as “risk-on for semis,” flagging Nvidia as leading a broader sector rotation.

Across approximately 28 tracked analysts, the average price target sits at roughly $258, with approximately 95% carrying Buy ratings. NVDA traded at $230.24 in premarket as of 7:59 a.m. EDT on 14 May.

The revenue arithmetic requires care. ByteDance’s 200,000 approved units at $30,000-$40,000 per unit represent $6 billion-$8 billion in hardware value for that single buyer. No confirmed consensus figure for all ten companies has been independently verified, and estimates range widely. The direction of the revenue uplift is clear; the magnitude is still being calculated.

Beijing’s import barrier adds a layer of uncertainty that the approved-unit figures alone do not capture: customs blocks have prevented any commercial H200 shipments from completing to date, meaning the gap between U.S. licensing approvals and actual revenue recognition may be wider than the hardware value estimates suggest.

The geopolitical wiring behind the timing

The H200 approvals were not a standalone commercial event. They landed on the opening day of the Xi-Trump summit in Beijing (14-15 May 2026), a summit covering trade normalisation and broader strategic stability. The timing is widely interpreted as a diplomatic goodwill gesture accompanying the talks.

The backdrop adds urgency. U.S. strikes on Iran on 7 May 2026 and ongoing ceasefire negotiations in mid-May have elevated energy costs (WTI crude at $101.04 per barrel on 14 May) and increased Washington’s incentive to stabilise the China relationship. Trump is reported to be seeking Beijing’s involvement in ending the conflict.

The broader summit agenda risks extend well beyond chip licences: analysts at Fidelity International flagged that markets are simultaneously pricing optimism on trade purchase commitments, AI security talks, and Iran energy negotiations, each carrying materially different probability distributions that investors should disaggregate before taking positions.

Not everyone in Washington is on board. Three categories of risk to the licensing regime are already visible:

• Congressional opposition: A bipartisan letter to Commerce Secretary Lutnick urging review of the approvals was published the morning of 14 May. Senator Pete Ricketts (R-NE) called for revival of the SAFE Chips Act.
• Huawei scope dispute: The precise terms of Huawei’s approval remain unresolved, creating a pressure point for lawmakers who oppose any affiliate access.
• Summit outcome uncertainty: If the Xi-Trump summit deteriorates or the Iran situation escalates, the licensing environment could reverse.

The Commerce Department confirmed all ten firms were vetted for non-military end-use with ongoing compliance monitoring. Investors who treat this purely as a commercial catalyst miss the political scaffolding holding it up.

Nvidia at $5.9 trillion: what investors should take away from today

Nvidia’s market capitalisation reached approximately $5.53 trillion in premarket on 14 May 2026, with shares up roughly 1.9% following a prior close of $225.83 on 13 May (up 2.29% that session). The move sits within a broader risk-on environment: the S&P 500 reached 7,444.25 (up approximately 0.58%) and the Nasdaq Composite hit 26,402.34 (up approximately 1.20%) on 14 May.

A genuine counterweight exists beneath the euphoria. The probability of a rate hike by year-end 2026 has risen above 28% on CME FedWatch, up from 20.7% one week earlier. The 10-year Treasury yield stood at 4.455% on 14 May. These conditions create headwinds for growth equity valuations that investors should factor into position sizing.

AI chip valuation comparisons across Nvidia and Broadcom reveal a market that is pricing very different risk profiles into each name: Nvidia’s forward multiple of approximately 24x reflects de-rating on competitive concerns despite strong execution, while Broadcom’s 37x reflects contract-backed revenue certainty from locked-in hyperscaler deals extending through at least 2029.

Three watch points going forward:

1. Summit outcomes: Whether the 14-15 May talks produce any indication that Blackwell restrictions may be revisited.
2. Regulatory reversal risk: Congressional pushback and the unresolved Huawei scope dispute could tighten licensing before the approved units ship.
3. Revenue recognition timing: Approved units translate to bookable revenue over Q2-Q3 2026, meaning the financial impact will appear in earnings with a lag.

The H200 approval is real and revenue-generating. But Blackwell remains banned, and the policy window is tied to geopolitical conditions outside Nvidia’s control.

The Nvidia moment that semiconductor investors have been waiting for

U.S. authorities have authorised ten Chinese firms to purchase the H200, creating a concrete near-term revenue pathway for Nvidia for the first time since the H20 was halted in August 2025. The approvals are material: billions of dollars in hardware demand from named buyers, backed by a diplomatic summit that gave Washington reason to open the door.

The constraints remain equally material. Blackwell stays banned. Performance caps limit what Chinese buyers can extract from the hardware they receive. And the entire licensing regime rests on a geopolitical foundation that could shift with one failed summit session or one congressional vote.

The next testable catalyst is specific: whether the 14-15 May summit communiqué contains any language revisiting next-generation chip restrictions, and whether Nvidia’s Q2 earnings reflect initial H200 order bookings. Both are trackable. Both will determine whether this approval is a one-time concession or the beginning of a broader reopening.

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. These statements regarding revenue estimates and policy outcomes are speculative and subject to change based on market developments and geopolitical conditions.