US Considers Ban on Import of New Chinese Optical Modules, What's the Impact?

Pubblicato 2026-08-05Pubblicato ultima volta 2026-08-05

Introduzione

While domestic US manufacturers have the technical capability, their production capacity is difficult to scale up in the short term. Compatibility and reliability testing of high-speed optical modules for cloud service providers also requires a considerable amount of time. The potential ban may impact the construction progress of AI data centers in the United States.

While US domestic manufacturers have the technical capability, their production capacity is difficult to ramp up in the short term. Compatibility and reliability testing for high-speed optical modules by cloud providers also requires considerable time. The potential ban could affect the progress of US AI data center construction.

Following bans on importing Chinese robots and inverters, the US government is now targeting Chinese optical transceiver modules (referred to as "optical modules").

According to media reports on August 4th, informed sources revealed that the US Federal Communications Commission (FCC) is drafting a new regulation plan to prohibit the import of new optical modules manufactured in China. Existing mature models that have already obtained FCC certification and are in mass supply are not subject to constraint, and products from non-Chinese manufacturers can be exempted from the ban.

The report also stated that US officials hope the ban will be announced and take effect within the year, but the draft is not yet finalized. The FCC still has the authority to modify, relax, or shelve the policy.

In response to the rumored ban, the Chinese Embassy in the US urged the US side to heed the objective and rational voices of the business communities of both countries, to stop groundlessly smearing Chinese companies and frequently using sanctions to apply pressure. Should measures be introduced that seriously damage China's legitimate rights and interests, China will take all necessary measures to respond.

High-speed optical modules handle the conversion of optical and electrical signals between GPU (AI computing chip) clusters and are the blood vessels of supercomputing networks. The annual AI (artificial intelligence) capital expenditures of US leading cloud providers, reaching hundreds of billions of dollars, ultimately translate into orders for high-speed optical modules. The US is blocking new domestic products under the pretext of computing infrastructure security. This move essentially extends the supply chain battle from the upstream chip blockade to the midstream and downstream critical hardware components for computing power.

The market is concerned that if the ban is implemented, it could impact Chinese optical module manufacturers like Zhongji Innolight, New H3C Technologies, and TFC Optical Communication, which currently hold a major market share.

A person close to Chinese optical module manufacturers told Caijing that the matter is uncertain, and it remains unclear whether the related companies will respond. Currently, the stance follows that of the national competent authorities.

Industry Impact

The reason given by the FCC is cybersecurity risk, claiming that Chinese equipment poses hidden dangers such as data theft, implantation of malicious programs, and interference with the operation of critical AI computing power facilities.

An industry insider told Caijing that optical modules are only responsible for photoelectric signal conversion and do not have functions for data storage or information collection. From a technical perspective, the risk of data leakage is almost non-existent.

"This potential ban could be a pressure tactic by the US side ahead of the US-China intergovernmental dialogue on artificial intelligence in September," analyzed one industry insider. They noted that over the past few years, US measures like proposed additional tariffs or trade bans on the Chinese optical module industry have not materialized. This reflects the importance of optical modules in North American AI construction and the benefits of Chinese manufacturing. China's optical module industry holds a global position in AI akin to South Korea's memory chips and Japan's high-end materials.

The media reports did not define the scope of the restricted "new optical transceiver modules," but the market generally anticipates the focus of control will be on next-generation high-speed optical modules like the iterative versions of 1.6T and 3.2T. This is because 800G optical modules have already been certified and are being supplied in bulk to the US market. However, the first-generation versions of 1.6T optical modules lie at the borderline between "mature legacy models" and next-generation new products. Whether they are restricted depends on how the FCC defines "new models."

The global AI industry is currently in a cycle window transitioning from 1.6T to 3.2T optical module technology. Nvidia's new computing platform and Google's TPU (Tensor Processing Unit) clusters have already begun testing and sampling of next-generation optical modules. By skipping the mature existing market and directly blocking new product access, the US intends to disrupt the first-mover advantage of Chinese manufacturers in the continuous iteration of high-end optical components, forcing North American cloud providers to turn to domestic and other alternative overseas suppliers.

Chinese manufacturers hold significant advantages in packaging, system integration, yield control, and cost control for high-end optical modules.

Statistics from the industry research firm LightCounting show that the global Ethernet data center optical module market size in 2025 is approximately $23.8 billion. Based on revenue, the top ten global companies in order are Zhongji Innolight (China), New H3C Technologies (China), Coherent (US), Accelink (China), Lumentum (US), HiSense Broadband (China), Huagong Zhengyuan (China), Source Photonics (controlled by Chinese capital, US operating entity), Marvell (US), and Cambridge Industries (China).

If the ban is issued, the Chinese manufacturers primarily impacted would be Zhongji Innolight and New H3C Technologies.

Among the global leading optical module manufacturers, Zhongji Innolight (with a market share between 23% and 27%) derives about 90.58% of its revenue from overseas markets. Nvidia, Google, Meta, and Amazon are its core customers. In Q1 2026, approximately 61.7% of the company's income came from the US market. The US Department of Defense included Zhongji Innolight on its 1260H list in June this year, prohibiting US government and military procurement of its products.

New H3C Technologies would be impacted more significantly. According to its Q1 2026 report, over 94% of the company's revenue comes from overseas business, with customers essentially all being North American cloud giants. Its flagship LPO low-power 1.6T product is the main model procured by Meta. If its next-generation 3.2T products are barred from US certification, the company's growth trajectory over the next two to three years will face a hard constraint.

In comparison, Accelink (with overseas revenue accounting for about 25%), which has a dual focus on domestic computing power and overseas orders, and Huagong Technology Group (whose wholly-owned subsidiary Huagong Zhengyuan specializes in optical module business, with the group's overall overseas revenue accounting for about 10%) and is deeply tied to domestic internet giants, would be less affected.

Another focal point of market attention is that, based on currently available information, it is impossible to determine whether optical modules produced overseas by Chinese companies would be restricted.

Chinese manufacturers have long begun hedging against supply chain risks. Since 2025, Zhongji Innolight, New H3C Technologies, and Accelink have successively put production bases in Thailand into operation. High-end 1.6T modules specifically supplied to North America are shipped from Southeast Asian plants to avoid tariffs and geopolitical regulatory risks. Some companies are simultaneously planning branch factories in Malaysia, building a multi-location overseas manufacturing network to disperse policy risks associated with a single region.

However, if US regulators determine the origin by piercing through corporate shareholding nationality, the risk-hedging effect of overseas factories could be significantly diminished.

Impact on Capital Markets

Affected by this news, US-listed domestic optical module concept stocks collectively rose. AAOI (NASDAQ: AAOI) saw a pre-market surge of up to 18%; Coherent (NYSE: COHR) had a pre-market peak gain of 19%; Lumentum (NASDAQ: LITE) rose up to 13% pre-market. After the market opened, the gains somewhat retreated, with their intraday highest gains recorded at 15% and 8.92% respectively. Upstream optical fiber component stock Corning (NYSE: GLW) rose nearly 8% pre-market. Funds are betting that if new domestic high-speed optical module access is restricted, procurement orders from North American cloud providers will flow to domestic optical communication manufacturers.

On August 5th, during the A-share market call auction session, Zhongji Innolight, New H3C Technologies, and TFC Optical Communication all fell by the 20% daily limit. However, at market open, their share prices rebounded. At the time of writing, Zhongji Innolight's stock price was down 13.89% to 880 yuan per share, New H3C Technologies' was down 10.71% to 400 yuan per share, Dongshan Precision's was down 9.54% to 162 yuan per share, Yuanjie Technology's was down 8.75% to 1200 yuan per share, and TFC Optical Communication's was down 6.6% to 198 yuan per share. The optical module index fell by 3.5%.

Regarding the impact of this event on A-share optical module companies, CITIC Securities stated that this policy is difficult to implement and its impact is limited. On one hand, the policy is hard to enforce, as the vast majority of global high-speed optical modules currently come from Chinese manufacturers. Relying solely on overseas manufacturers cannot support North American demand. Completely replacing the current stock of optical modules would be unimaginably costly. Related policies have ultimately been much ado about nothing. On the other hand, leading manufacturers have overseas production capacity. Zhongji Innolight, New H3C Technologies, and others actively began布局 (laying out) Southeast Asian capacity early on, and current shipments to North America basically come from overseas capacity. Therefore, even in extreme scenarios, Chinese manufacturers have ample room for response.

CITIC Securities analyzed that currently, domestic optical module leaders significantly outpace overseas counterparts in terms of technical strength, product progress, cost control, and capacity layout. The gap cannot be closed in the short term. A comprehensive decoupling of optical modules would only cause comprehensive落后 (lagging behind) in North American data center construction, which would be a case of penny wise and pound foolish.

CICC stated that there will definitely be short-term emotional impact, but the medium to long-term impact is likely limited. Their analysis noted that the core function of optical modules is photoelectric signal conversion; they do not themselves handle business data storage. Compared to active equipment, their risk boundaries and attack capabilities are clearly different. Therefore, whether the logic of this policy holds up, and whether the proposed restrictions comply with the principle of necessity, is itself debatable.

Simultaneously, CICC mentioned that similar stories have been told many times since 2020. The market has consistently worried that the share of Chinese manufacturers would be taken by North American competitors, but discussions have actually subsided in the past two years. This is because years of verification have shown that even with continuous industrial policy倾斜 (tilt) towards the US domestic supply chain, the share of Chinese manufacturers has not decreased but rather increased. The reasons are practical: high-speed optical modules are not something that can be delivered immediately just by having factories and equipment. Customer certification, mass production yield, and delivery stability all require time to accumulate. Currently, Chinese suppliers' share in high-speed optical module orders for overseas major customers averages seventy to eighty percent or even higher. If a one-size-fits-all ban is applied, North American manufacturers might not be able to fill the gap for several years.

"This matter is about trading on sentiment in the short term, observing rule interpretations in the medium term, and competing on product, yield, and delivery in the long term. Policies can change shipping addresses, but may not change who is best at making optical modules," CICC stated.

Guosheng Securities, meanwhile, stated that the optical module industry has developed to form a supply chain pattern centered around Zhongji Innolight + New H3C Technologies + Coherent + Lumentum. In the medium term, it is difficult to achieve capacity replacement for this 2 Chinese + 2 US core. As the "screws and bolts" for photoelectric conversion in data centers, optical modules do not承载 (carry) data storage or forwarding functions, leaving little room for concern regarding information security. Simultaneously, Zhongji Innolight and New H3C Technologies have already achieved批量交付 (batch delivery) in Thailand and are accelerating their北美布局 (North American layout). They have relatively rich experience in the structural design and compliance aspects of going global. While the impact of geopolitics on the globalization of optical module manufacturers should not be ignored, ultimately, the focus must return to the business logic itself—supply and demand dynamics, technological iteration, and product delivery—rather than short-term emotional disturbances.

An industry insider told Caijing that while US domestic manufacturers have technical capability, their production capacity is difficult to ramp up in the short term. Compatibility and reliability testing for high-speed optical modules by cloud providers also requires considerable time, which could affect the progress of US AI data center construction. This may influence the formulation of detailed rules for the FCC ban. However, the general direction of restricting the Chinese optical module industry and扶植 (fostering) US domestic enterprises is unlikely to change.

"The game between China and the US in the technology field is常态化 (normalized). Moreover, the reported FCC action is only a proposed ban, and subsequent complex博弈 (game-playing) is still needed. The event may die down amid反复的博弈 (repeated game-playing)," analyzed the insider.

They stated that the position of Chinese optical module companies globally is irreplaceable. If the ban is强制实施 (forcefully implemented), it would significantly delay the progress of US AI construction,反而给 (instead giving) China an opportunity to catch up in AI. Chinese optical module suppliers碾压 (overwhelm) overseas manufacturers in R&D capability, production capacity, yield rate, and cost advantages. Overseas manufacturers completely lack the ability to替代 (replace) Chinese manufacturers within five years.

Letture associate

Meme Coin with $60 Million Market Cap Plunges 65% in One Minute, FOMO Faces Renewed Scrutiny

A Solana-based meme token, $CATE, which had surged from a $20M+ to over $80M market cap in about a week, experienced a dramatic 65% crash within one minute. This flash crash has intensified scrutiny on the trading app 'fomo' and highlighted the speculative nature of the current meme coin market. The crash coincided with two events: the token's X account being suspended and the fomo app experiencing downtime, preventing users from trading. While the X suspension was straightforward, the fomo outage raised significant questions. $CATE's primary narrative driver was the open endorsement by Poorgoat, a top-ranked trader on fomo with over 200,000 followers, who had turned a ~$45,000 investment into over $2M at the peak. The token itself had no novel fundamentals, being a "cat sister" to Doge, a concept already existing on Ethereum without success. The crash, triggered by less than $1.5M in selling volume despite over 60,000 holder addresses, exposed a harsh reality: purely "organic" community-driven meme tokens (excluding past successes like $SPX) may now have a market cap ceiling around $17M, as exemplified by the long-term chart of $neet. This incident has fueled existing controversies surrounding fomo. Critics have grown skeptical of the app, alleging that rankings dominated by KOLs who receive lucrative token airdrops could be manipulated to create "pump-and-dump" schemes, luring in retail users before a rug pull. The timing of the crash during fomo's outage—preventing many of its users (who represent over 60% of $CATE holders) from reacting—was viewed as highly suspicious. Further controversy arose when another popular fomo trader publicly sold near the peak, and concerns were raised about the security of accessing private keys during the app's downtime. Fomo's official explanation of server overload due to surging user traffic was met with skepticism, given its substantial funding. The event serves as a stark reminder of the risks in meme coin speculation and the potential vulnerabilities of relying on a single trading platform during market volatility.

marsbit11 min fa

Meme Coin with $60 Million Market Cap Plunges 65% in One Minute, FOMO Faces Renewed Scrutiny

marsbit11 min fa

Arthur Hayes' 10,000-Word New Article: When Will the AI Wave Peak? Will the Bubble Burst? Why Am I Frantically Buying ETH?

Arthur Hayes' latest article explores the intersection of AI investment, credit markets, and cryptocurrency. He argues that the massive capital expenditure (CAPEX) in AI infrastructure resembles a real estate-driven credit bubble, akin to the 2008 financial crisis, rather than a tech growth story. Hayes predicts that the AI bubble's peak will arrive in 2027, triggered not by failing AI profits but by a slowdown in the growth rate of data center construction and capex, revealing systemic over-leverage. He explains that banks, incentivized by government policy and steep yield curves, will continue lending to AI projects despite rising risks, expecting eventual government bailouts. Hayes further speculates that the US government might directly intervene by creating a sovereign wealth fund to buy AI equities, injecting permanent liquidity into the system. For crypto investors, this misallocation of capital is bullish. Hayes believes Bitcoin will bottom as AI capex growth decelerates but credit continues to flow, setting the stage for a major rally fueled by the inevitable large-scale monetary stimulus. He sets a rough year-end 2026 target of $5,000 for Ethereum, citing its potential role as a settlement layer for tokenized real-world assets (RWA) as a key narrative driver. The eventual government response to the AI credit bubble, he concludes, will unleash a cryptocurrency bull market unlike any seen since 2021.

marsbit25 min fa

Arthur Hayes' 10,000-Word New Article: When Will the AI Wave Peak? Will the Bubble Burst? Why Am I Frantically Buying ETH?

marsbit25 min fa

Goldman Sachs Details Korea Memory "8 Key Focus Points": Valuation, Long-Term Agreements, Inventory, CXMT Impact, Buybacks, etc.

Goldman Sachs maintains a "Buy" rating on Samsung Electronics and SK Hynix, stating their recent sharp share price corrections reflect excessive market pessimism unsupported by fundamentals. The report addresses eight key market concerns: 1. **HBM Pricing:** GS forecasts HBM blended average prices for Samsung and SK Hynix to nearly double by 2027 due to persistent tight supply and a need to re-establish price premiums over standard DRAM. 2. **Long-Term Agreements (LTAs):** LTA terms are shifting in favor of suppliers, featuring longer terms, higher coverage targets (60-70% of capacity), "floor-price" protections, and significant upfront customer deposits. 3. **Inventory:** While module maker inventory is elevated, this represents a small portion of the overall market. Supplier and end-customer inventories remain healthy at low levels. 4. **NAND Supply:** Fears of a NAND oversupply are overblown. Strong enterprise SSD demand is expected to offset consumer weakness, with the supply-demand gap widening through 2027. 5. **Shareholder Returns:** GS expects both companies' actual dividends to exceed current market consensus. Share buybacks would be well-received, especially for SK Hynix to counter ADR-related dilution. 6. **SK Hynix ADR Premium:** The ADR's ~30% premium over the local share is likely persistent due to conversion constraints but aids long-term valuation re-rating by improving global investor access. 7. **SK Hynix Q2 '26 Results:** The earnings miss was due to one-time factors like slower HBM4 mix transition. Q3 is expected to see a strong recovery with higher volumes and better pricing/mix. 8. **CXMT Impact:** The competitive threat from China's CXMT is limited to the domestic market due to its technology lag and different product mix, posing minimal risk to the global supply-demand balance. GS concludes the current low valuations (approx. 3.5x 2027 P/E) imply an unrealistic lack of profit sustainability.

marsbit26 min fa

Goldman Sachs Details Korea Memory "8 Key Focus Points": Valuation, Long-Term Agreements, Inventory, CXMT Impact, Buybacks, etc.

marsbit26 min fa

Trading

Spot
活动图片