About 600,000 H200 chips stranded in Hong Kong: Has China's chip industry achieved reverse chokehold leverage?
About 600,000 NVIDIA H200 chips that have already obtained U.S. export licenses are stranded in Hong Kong awaiting approval, and the author uses this as a basis to analyze China's reverse leverage capability in the chip game. The article argues that this capability is concentrated in three asymmetric nodes: gallium and germanium raw materials (in 2025 China accounts for 98.9% of primary gallium and 68.6% of germanium), H200 access rights, and memory production capacity, while advanced process nodes, EUV lithography machines, and high-end HBM remain one-way dependencies on the United States. The article also mentions that the expiration of the suspension period for civilian controls on gallium and germanium on November 27, 2026, is the key variable going forward.
Why read this
AI score 64
The article breaks down China's asymmetric leverage in gallium and germanium, market access, and memory production capacity, and points out that advanced process nodes and HBM remain one-way dependencies.
Topics
Body · AI translation
[[**About 600,000 H200 chips are stranded in Hong Kong awaiting approval, showing that China has countermeasures at asymmetric nodes such as gallium and germanium and market access, but advanced processes and HBM still depend on the United States; the expiration of gallium and germanium controls on November 27 is a key variable.**
**Key points**
**1. Nvidia's China weight is small, and approval power becomes a bargaining chip**
In Q2 of fiscal 2027, revenue from mainland China and Hong Kong accounted for 8.2%, Hopper sales to China were less than 1% of data center revenue, H200 has generated no revenue and requires a 25% tariff, so the impact of not approving it is controllable.
**2. Domestic inference substitution holds, but training remains constrained**
In H1 2026, Ascend's share was 28%, only 6 percentage points behind Nvidia's 34%, and non-GPU shipments surpassed GPU for the first time; but the 910C is about 60% to 70% of the H100, and production capacity, yield and HBM are bottlenecks.
**3. Gallium and germanium controls are the card closest to "reverse chokehold"**
China accounted for 98.9% of primary gallium and 68.6% of germanium in 2025, and prices outside China rose 9 to 10 times; the suspension covers civilian use only, the military ban has not been relaxed, and whether it resumes after the November 27, 2026 expiration is key.
**4. Hong Kong's 600,000 H200 chips expose regulatory friction and depreciation risk**
The U.S. side requires case-by-case review, China-bound volumes no more than 50% of U.S. customers, and a 25% cut, while China worries about backdoors and prepayment; the H200 lags behind the B200 and Vera Rubin, and gray transshipment takes nearly $290,000 per unit. 2026-11-11 15:03 Hot topic | Has China's chip industry already begun to choke others in reverse? Skai has something to say Quick overview
This article comes from the WeChat public account: Skai Suisui Kan, author: Skai, original title: "Hot topic | Has China's chip industry already begun to choke others in reverse?"
November 9, 2026, a regular Foreign Ministry press conference. An AFP reporter asked: Reuters reports that about 600,000 advanced Nvidia artificial intelligence chips are stored in Hong Kong, waiting to be shipped to mainland China; despite demand from tech companies, the Chinese side refuses to approve large-scale imports. Spokesperson Guo Jiakun answered with only one sentence: "China's position on the issue of U.S. chips exported to China is consistent. For specific details, I suggest you ask the competent Chinese authorities."
On the same day, a Bloomberg reporter followed up on "how China would respond if the United States adopted stricter chip restrictions," and received a word-for-word identical answer. No confirmation of the number, no explanation of the standards, no response on "where it is stuck."
But this ambiguity is more informative than an explicit statement. Over the past three years, the keyword for China's AI industry was "unable to buy chips," with the restriction in Washington; now about 600,000 H200 chips that have already obtained U.S. export licenses are sitting in Hong Kong warehouses, and the signature line is in Beijing. The United States has cleared them, but Beijing has not signed. U.S. Commerce Secretary Lutnick said bluntly at a Senate Appropriations Committee hearing on April 22, 2026: "I can tell you clearly that, to date, they have not bought a single chip."
This picture is easily read as a clean narrative reversal. But "reverse chokehold" as a judgment needs to be precisely taken apart: China has indeed gained the ability to say "no" at several asymmetric nodes in the chip supply chain, but this ability is still a long way from reciprocal countermeasures in the sense of technological sovereignty.
First, see clearly where this business sits in Nvidia's books: small enough that it can "not approve"
600,000 H200 chips stranded in Hong Kong should intuitively make Nvidia anxious. But breaking down the financial data, this intuition needs correction.
In the second quarter of fiscal 2027 (ended July 26, 2026), Nvidia's revenue was $96.221 billion, of which data centers were $89.023 billion, accounting for 92.5%, up 117% year on year. Shipments of Hopper products to China were less than 1% of data center revenue, estimated at less than $890 million at the literal upper limit; mainland China and Hong Kong combined were $7.88 billion, 8.2% of total revenue. Jensen Huang disclosed at the annual shareholders meeting that China's share of revenue had fallen to about 9% in fiscal 2026, while before export controls, in fiscal 2022, that figure was 26.4%.
Harder than these numbers are two sentences in the annual report. First, over the past six months, the company recorded about $400 million in charges for excess H200 inventory and purchase obligations; second, the company assessed that, without approval from both governments, it "has effectively been foreclosed from competing in China's data center computing market," and this guidance no longer assumes any data center revenue from China. The word "foreclosed" appearing in a company's annual report says more than any analyst's judgment.
Jensen Huang said even more directly in a CNBC interview on May 21, 2026: "Huawei is very, very strong... we have effectively ceded that market to them."
There is another layer that is easily overlooked: the H200 was approved for small-batch exports to China in February 2026, but the annual report states that "the program generated no revenue," and H200 chips must pay a 25% tariff when shipped back to the United States. The production and sales chain of this product line itself has to cross a U.S. tariff, and its cost structure is no longer just the price of a chip.
This means that the marginal damage of "not approving" to Nvidia's current-period revenue is manageable. With 600,000 chips stuck in Hong Kong, its global AI computing business is still expanding at more than double the rate. But the symbolic significance of this business far exceeds its financial significance: if even chips approved for export cannot get in, Nvidia's presence in China's data center market will accelerate toward zero, and Huawei's ecosystem will continue to consolidate in a "no-competition" environment. One of the premises for Beijing daring to slow the approval pace so much is that Nvidia's commercial weight in the Chinese market has already become too small to constitute reverse pressure.
What domestic substitution has caught, and what it has not
The reason the H200 standoff in Hong Kong can happen is another premise: domestic AI chips can already fill in for a considerable portion of scenarios.
IDC data for the first half of 2026 gives a clear scale: China's data center AI accelerator market shipped 2.008 million units in total, with Nvidia at 678,000 units, a 34% share, down 44% year on year; Huawei Ascend at 572,000 units, a 28% share, up from more than 300,000 units in the same period last year, with the gap between the two narrowing from 31 percentage points a year earlier to 6 percentage points. Alibaba Pingtouge took 7%, Cambricon 5%, Kunlunxin 4%, and Hygon 3%. More telling is the structural change: non-GPU accelerators shipped 1.016 million units, surpassing GPUs' 991,000 units for the first time, growing nearly 80% year on year, and these four domestic vendors together accounted for 95% of the non-GPU market. Of the 11 Chinese chipmakers tracked by IDC, 9 saw year-on-year shipment growth in the first half. Morgan Stanley expects China's AI chip self-sufficiency rate to rise from 35% in 2025 to 45% in 2026, and to jump to 68% in 2027.
However, the matter of "share" depends on who is counting and what is being counted. Huawei rotating chairman Xu Zhijun said in September 2026 that, based on data that can be counted, Ascend's shipments in the Chinese market have already surpassed Nvidia's; IDC's measure is 34% versus 28%. The former refers to all accelerator cards, while the latter is limited to data center accelerators. Both numbers are real, only the denominators differ—this itself is a feature of the market reconstruction period: even the most basic statistical boundaries are still moving.
Beijing's approval stance on the H200 confirms a layered strategy: prioritize domestic chips for inference, and mainly reserve imported chips for training. This strategy has already been implemented at the software layer. DeepSeek's V4 runs entirely on Ascend; on September 30, 2026, it open-sourced a full set of infrastructure components for Ascend—the TileLang compiler tool, the DeepGEMM operator library, and the DeepEP communication library—existing in parallel with the corresponding components for Nvidia's platform. The same DeepInfer code supports both Ascend 950 and Nvidia SM100, and switching hardware requires only changing an environment variable. This matters far more than "catching up in performance": it turns "not using CUDA" from a slogan into a deployable, measurable, and comparable engineering option. In other words, the part of "reverse chokehold" that holds on AI chips is "for inference, I can choose not to buy yours."
But the gap on the training side cannot be ignored, and the bottleneck is not only in chip design. Ascend 910C uses dual-die packaging, with an FP16 peak of about 800 TFLOPS and 96GB of integrated HBM; its overall performance is about 60% to 70% of the H100, and the DeepSeek team's measured inference performance is about 60% of the H100. The gap of a single card is made up for by the system: the 384 supernode uses self-developed HCCS to connect 384 cards into one computing pool, with more than 500 sets already deployed; the CloudMatrix 384 cluster has computing power of about 300 PFlops, at the cost of roughly four times the power consumption.
The real bottlenecks are in two places. First, production capacity: the 2026 production target for the Ascend 910C is about 600,000 units, and the target for all series is 1.6 million wafers, nearly double that of 2025, but the wafer demand corresponding to 600,000 units is about 35,000 wafers per month, which already exceeds the current total capacity of SMIC's N+2 line; either capacity must be expanded before the end of the year, or output must be diverted to other nodes. Second, yield: estimates of SMIC's N+2 yield vary from 20% to 70% across institutions—about 20% in early 2024, about 40% in early 2025, and 65% to 70% in 2026 according to industry figures cited by Bloomberg—all below TSMC's roughly 90% for the same process. The divergence in figures itself shows one thing: the real data for this production line is not transparent, and the cost is hidden precisely in that opacity. As for HBM, in the second quarter of 2026 the global HBM revenue share was 50% for SK Hynix, 33% for Samsung, and 18% for Micron; Chinese manufacturers are still at the sample-sending stage and have no share.
So "reverse chokehold" does not hold on the training side. China still needs a limited import window to fill the high-end computing power gap, but this window has been deliberately narrowed.
What comes closest to a "reverse chokehold" is precisely not the chips themselves
If "reverse chokehold" means having veto power over a link that the other side cannot replace, then China's closest thing to this definition in chip-related fields is not any single chip, but the raw materials on which chip manufacturing depends.
First, let us state the status of this card accurately. On December 3, 2024, the Ministry of Commerce issued Announcement No. 46 of 2024: Paragraph 1 prohibits the export of dual-use items to U.S. military users or for military use; Paragraph 2 states that "in principle, no licenses will be granted for the export to the United States of dual-use items related to gallium, germanium, antimony, and superhard materials," and imposes stricter end-user and end-use reviews on graphite. On November 9, 2025, Ministry of Commerce Announcement No. 72 of 2025: the implementation of Paragraph 2 is suspended until November 27, 2026.
In other words, only the civilian half has been suspended; the ban on military users and military use has never been relaxed. And the civilian half is currently in its suspension period, with seventeen days left until it expires. The card is on the table, with a hand resting on it but not playing it. Whether it will be restored or the suspension extended after seventeen days is the variable more worth watching right now than the "600,000 units."
The numbers are readily available. The consultancy Project Blue estimates that in 2025 China will account for 98.9% of global primary gallium supply and 68.6% of germanium supply. S&P Global projects that by the end of 2026, gallium supply capacity outside China will be about 20 tons, with a shortfall of about 678 tons; germanium metal output will be about 31 tons, 177 tons less than demand. Reuters reported in September that since the controls began in 2023, prices of these two metals in markets outside China have risen to 9 to 10 times their levels at that time.
The most intuitive comparison is not in the shortfall, but in the physical form of production capacity. Gallium is not mined; it is a byproduct of alumina production. Industry research gives a ratio of about 200,000 tons of alumina production capacity to support every 1 ton of gallium produced; China possesses more than 55% of global alumina production capacity. Sanmenxia Aluminum alone has a metal gallium production capacity of 290 tons per year—fourteen times the projected supply capacity of all regions outside China by the end of 2026. The same applies to germanium: it is a byproduct of zinc smelting, and to extract germanium, one must first have zinc smelters. Even under the most optimistic path: if the eight announced projects proceed smoothly, gallium supply outside China will increase from about 5 tons at present to 386 tons by 2030, and Reuters, based on S&P data, calculates that the non-Chinese market will still need to obtain about 65% of its gallium from China at that time; germanium refining capacity will rise to 126 tons, which would still cover only about 48% of local demand.
What these materials share is that they are used in extremely small quantities, yet without them an entire production line halts. Gallium is the foundation of gallium nitride RF devices and advanced radar; germanium is the core of infrared optics and optical fiber. Changing suppliers takes years and enormous certification investment, and China's concentration of production capacity means there is no second option in the short term. What it restricts is not design capability or manufacturing processes, but those few elements on the periodic table for which China happens to possess both reserve advantages and refining capacity. This is currently the most tangible counter-card, but its nature is a veto power over the physical supply chain, not dominance derived from technological leadership.
Memory is another asymmetric node that is taking shape. In the second quarter of 2026, CXMT's DRAM revenue share rose from 4% a year earlier to 10%, ranking fourth globally, with revenue up 716% year on year; YMTC's NAND shipment share reached 17%, rising to third globally, with a bit-based share of 14%, and its revenue share also rose from 9% to 14% within four quarters. Samsung, SK Hynix, and Micron together still account for about 90% of the DRAM market and more than 60% of the NAND market.
It does not "choke" anyone, but it weakens the ability of a few giants to price at will. When Samsung and SK Hynix shift capacity toward HBM and choose to control conventional memory supply to maintain prices, the incremental supply from CXMT and YMTC is the marginal force pushing prices down. This is a kind of "unavoidability in the price mechanism," which is not the same as the "physical unavoidability" of gallium and germanium, but it is likewise asymmetric.
Hong Kong's 600,000 chips: the physical sediment of regulatory friction
The 600,000 H200s will not rot in warehouses, but their circulation path itself is a specimen for understanding this game.
According to a Reuters report on November 5, about 600,000 chips are stored in Hong Kong awaiting shipment to the mainland; in recent months some have been released, with several large Chinese internet companies each obtaining about 10,000 chips, and a small number entering through middlemen and indirect channels. Price signals explain the situation better than quantities: by the end of October, the quoted price of an eight-card H200 server on the mainland had exceeded 6 million yuan, about three times the level a year earlier, while Blackwell servers were quoted at more than 20 million yuan. Hong Kong is not a low-price dumping ground; it is a dammed lake of high prices and shortages.
The huge profits of the gray channel are documented. On August 24, the Keelung District Prosecutors Office in Taiwan indicted nine people: a company used a Japanese company as a relay break point to obtain whitelist eligibility, forged on-site inspections of the computer room, and obtained 130 servers equipped with B300 from Supermicro, of which 74 were transshipped to the mainland via Indonesia, Japan, and Hong Kong, yielding a profit of $21.2 million after deducting purchase costs, equivalent to about NT$640 million, with a per-unit cut of nearly $290,000. The "toll" on a single server explains better than the markup on the chips themselves why this gray chain cannot be welded shut.
Both ends of regulation are tightening. On January 13, 2026, the U.S. side changed H200 exports to China from "presumption of denial" to "case-by-case review," while attaching four hard conditions: a third-party laboratory must verify that its AI capability is "below H200"; the number of chips Chinese customers obtain must not exceed 50% of the total purchased by U.S. customers; it must be proven that the U.S. market supply is sufficient and that U.S. customer capacity is not squeezed out; and the recipient must have adequate security procedures and must not use the chips for military purposes. A 25% share of sales revenue is also added. On May 31, BIS issued further enforcement guidance: as long as the counterparty's headquarters, or the headquarters of its ultimate parent company, is located in Country Group D:5 (including mainland China and Hong Kong) or Macau, then even if it is registered in a third country, exporting 3A090 and 4A090 advanced computing items to it still requires prior authorization, and the review standard remains "presumption of denial." This clause explains why Hong Kong can only serve as a warehouse and not as a channel—transshipment entities registered in third countries are pierced all the way through to their headquarters.
China's concerns on its side are not without basis either. In July 2025, the Cyberspace Administration of China summoned Nvidia over the "backdoor risk" of the H20 and required it to provide explanatory materials; Reuters reported in January this year that Nvidia requires Chinese customers to pay in full in advance, with orders non-cancelable and non-refundable, and no configuration changes allowed after ordering. Under the expectation that supply may be cut off at any time, full prepayment amounts to leaving all the risk of supply cutoff with the buyer. Negotiations are therefore stuck on the terms, not just on approvals.
Thus Hong Kong has become an institutional buffer zone: the two major powers impose regulatory requirements on the same batch of goods that point in opposite directions but are both extremely complex, and inventory settles there.
The time account of the generation gap: H200 is a ticket that is depreciating
There is also a rational element that is easily overlooked: Beijing's caution about large-scale imports is not only industrial protection but also an asset depreciation calculation.
The H200 belongs to the Hopper architecture, equipped with 141GB HBM3e and 4.8TB/s bandwidth. The B200 has 192GB HBM3e and 8TB/s. These two figures cannot be directly compared on FP4—the H200 does not support FP4 at all; the comparable metric is FP8, on which the B200 is about twice the H200, and it uniquely supports FP4. By 2026, Nvidia's main push is already Vera Rubin (288GB HBM4), which had already entered full mass production by the second quarter of fiscal 2027. The H200 was the mainstay chip of 2024 to 2025; by 2026 the generation gap has already opened and will continue to widen.
The underlying logic of the U.S. approval of H200 exports to China is "second-best exports": what you get is the previous generation of products, enough for you to run but not to run fast. Senator Coons of Delaware said at a hearing that the H200's performance is six times that of the previously approved H20—that remark was meant for U.S. lawmakers, but it also illustrates another thing: what Washington is willing to sell to China is always one or even two generations behind the most advanced.
If H200s are purchased in large quantities, then two years later, when Vera Rubin is widespread, this batch of cards will depreciate faster because of the lack of FP4 support and lagging memory bandwidth. Buying a ticket to entry that is rapidly depreciating is not as good as leaving capital expenditure for domestic capacity that is still ramping up. Whether this trade-off is wise depends on whether domestic 7nm capacity and the HBM supply chain can be released on schedule in 2027 to 2028; but the logic of the decision itself is far more complex than "Beijing simply won't approve."
The precise meaning of "reverse chokehold"
Putting the four lines together, a more accurate picture emerges.
China has not achieved a "reverse chokehold" on the United States in the core links of chip manufacturing—advanced processes, EUV lithography machines, high-end HBM. These fields remain one-way dependencies, and dependencies that cannot be made symmetrical in the short term. The reverse capabilities China has formed are concentrated at three asymmetric nodes: absolute control at the raw material end (98.9% of gallium, 68.6% of germanium), access rights at the demand end (whether H200s come in and how many come in is decided by Beijing, and is subject to the hard constraint of the U.S. side's 50% ratio), and the structural impact of mature capacity on global prices (10% of DRAM, 17% of NAND).
All three share the same logic: they do not require you to lead technologically, only to occupy a position in the other side's supply chain that "cannot be bypassed and cannot find a short-term substitute." Gallium and germanium are physically impossible to bypass; H200 access is impossible to bypass in terms of market size; storage capacity is impossible to bypass in terms of the price mechanism. Conversely, America's advantages in advanced processes, high-end equipment, and HBM are likewise impossible for China to bypass.
The real change is not that the direction of the "chokehold" has reversed, but that the structure of the game has changed from "one side chokes, one side suffers" to "each side has nodes the other cannot bypass." Being able to say no on raw materials and being able to say yes on lithography machines are two completely different kinds of power. China is learning to use the former, while the answer to the latter is still written in those processes that have not yet been broken through.
The reason the Foreign Ministry's line that "our position is consistent" needs no explanation is that behind it stands not some statement, but this entire dependency structure that is slowly being reorganized—it is not yet symmetrical, but it is no longer one-way. Seventeen days later, the gallium and germanium card expires; whether it is restored or extended at that time will be the next moment when this structure is developed.
What truly changes the landscape of the U.S.-China chip game has never been the 600,000 H200s stranded in Hong Kong, but the gallium and germanium control window that expires on November 27. All of our current "reverse gaming capability" is built on an asymmetric structure in which technology still lags and raw materials hold absolute say.
The technological barriers of lithography machines and high-end HBM cannot be broken through in the short term; but the export authority over strategic raw materials such as gallium and germanium is in our hands. In 17 days, whether this key policy is extended, whether the original controls are restored, and whether it is tightened again will directly rewrite the cost and supply landscape of the global AI, RF, military, and optical industry chains. This is the biggest hidden variable in the semiconductor track by the end of 2026. AI industry signal channel: Frontier Technology
Has published 87 articles on Huxiu. This content is published with the author's authorization; the views represent only the author and do not represent Huxiu's position. If you have objections or complaints about this manuscript, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]