Nature of the plan: Current information comes from media-disclosed sources; officials have not yet publicly confirmed. Restrictions may focus on new models, but the criteria for determining new models, the continued validity of existing certifications, and transition arrangements are all unclear.

  Direct impact: Taking a domestic manufacturer with the largest global data communications market share as an example, in 2025, revenue from the US market accounted for 57.3%, further rising to 61.7% in Q1 2026. During the same period, revenue from mainland China accounted for less than 4%. The risks associated with revenue and customer structure should not be generalized by sentiment.

  Difficulty of substitution: Domestic manufacturers collectively account for nearly 50% of the global data communications optical module market, while the largest single US-based supplier accounts for about 12% to 13%. The formation of substitute supply is constrained by capacity building, customer validation, and yield ramp-up.

  Boundaries are shifting: On July 22, the (Federal Communications Commission) FCC voted to include logic devices produced by entities on the covered list within the scope of restrictions. Equipment containing such devices, even if produced by a third party, will not be able to obtain certification. Two sets of criteria, by producer and by place of production, are expanding in parallel.

  Allocation judgment: The current event is not sufficient to alter the existing framework. It is necessary to track whether the determination criteria and applicable categories extend to index-heavy components, as well as the progress of the US-China presidential meeting in mid-September.

Event Review and Boundaries Awaiting Confirmation

On August 4, 2026, Reuters, citing four informed sources, reported that the (Federal Communications Commission) FCC is drafting a measure to restrict new models of data center optical transceivers produced by Chinese companies from entering the US market. Officials hope to introduce it within the year. Neither the White House nor the FCC has publicly commented. The plan is not yet finalized, and the effective date and transition arrangements are also unclear.

According to the currently disclosed content, three details determine the practical implications of this plan. Firstly, restrictions may focus on new models, with existing approved products experiencing relatively limited immediate impact. However, how new models are defined, whether existing certifications continue after product upgrades, and whether a transition period will be set, have not been announced. Therefore, future delivery and revenue risks cannot be ruled out based on this. Secondly, the implementation method is reportedly to first impose principled restrictions on all new models, then grant exemptions to a large number of non-Chinese suppliers, similar in structure to previous measures targeting drones and routers. Thirdly, the official stated reasons are to prevent data theft, malware injection, and service interruptions, focusing on infrastructure security, meaning proactive defense before large-scale deployment, rather than remediation for existing damage.

The form of market reaction is more informative than its magnitude. The day after the news broke, the leading domestic optical module manufacturer's stock price once fell by 13% to 16% during intraday trading. Subsequently, the company clarified that relevant rules had not yet been introduced, and the stock price quickly recovered, even turning positive by over 9%. Peer manufacturers saw declines of approximately 5% to 10% on the same day, while suppliers primarily focused on upstream optical components, who were only indirectly affected, experienced smaller declines. The significant intraday volatility and rapid recovery reflect the market's difficulty in pricing an unfinalized plan, rather than a corresponding change in the profitability path on that day. In terms of direction, the first stage of pricing primarily reflects the reallocation of supply share between Chinese and non-Chinese manufacturers, rather than a decline in (Artificial Intelligence) AI terminal demand. However, if insufficient alternative capacity drives up procurement costs or delays data center deliveries, the second stage of impact could still transmit to the pace of (Artificial Intelligence) AI capital expenditure.

Directly Affected Segments: US Revenue Exposure and Difficulty of Substitution

Revenue structure determines the nature of risk in this segment. Taking InnoLight Technology (300308.SZ), the global leader in data communications optical module market share, as an example: in 2025, the company's revenue from the US market was 21.897 billion yuan, accounting for 57.3% of operating revenue; revenue from mainland China was 3.603 billion yuan, accounting for 9.4%; and other overseas regions accounted for 12.740 billion yuan, or 33.3%. In Q1 2026, this structure further tilted towards the US, with revenue shares from the US, other overseas regions, and mainland China being 61.7%, 34.6%, and 3.7% respectively. Customer concentration is also high; in 2025, the top five customers accounted for 24.1%, 18.3%, 14.2%, 11.3%, and 8.1% of operating revenue, respectively. Over 60% of revenue directly corresponds to US data center customers, while domestic revenue share has fallen to less than 4%. This structure implies that if restrictions are ultimately implemented and cover its main R&D models, the risks to revenue, customer structure, and profit margins are substantial and should not be simply attributed to emotional fluctuations.

The formation of substitute supply is subject to multiple constraints. According to market research firms, in 2025, major domestic manufacturers collectively accounted for nearly 50% of the global data communications optical module market, while the largest single US-based supplier accounted for about 12% to 13%. For domestic capacity to take over the substituted share, it needs to simultaneously complete capacity building, customer validation, and yield ramp-up. Policy promotion can accelerate investment but cannot eliminate the time required for this process. On the other hand, domestic leaders' high-end capacities are already distributed in Thailand, Malaysia, and domestic bases such as Suzhou and Tongling. A Mexican base may commence production in Q4. Whether overseas capacity constitutes a buffer depends entirely on the final rules—whether determined by place of production, ownership, or component origin. This cannot be determined before the text is released and is a prerequisite for all conclusions in this article.

Segments in China's (Artificial Intelligence) AI Hardware Chain Related to Holdings

Returning to the segments directly related to the tools we recommend: the construction of (Artificial Intelligence) AI data centers goes beyond chips, also requiring servers and complete machine assembly, printed circuit boards, power supplies and cooling, as well as upstream semiconductor equipment and materials. China possesses strong large-scale delivery capabilities in several of these segments, but still relies on external sources for high-end computing chips, advanced storage, and some core components. The significance of this event is to further illustrate that the advantages of China's supply chain are not solely concentrated in terminal assembly, but extend to several critical infrastructure segments, which happen to constitute the main weight sources of the aforementioned indices.

Regarding the demand basis for these segments, under supply security and cost constraints, the proportion of new domestic computing power demand met by domestic chips, servers, and basic software is expected to continue to increase. This demand comes from domestic model training, inference, and application deployment, and does not directly correspond to the access conditions for Chinese optical modules entering the US market. Historically, past rounds of external restrictions have objectively strengthened domestic customers' emphasis on supply chain security and accelerated the domestic adoption of equipment, materials, and computing chips. At the same time, in segments still reliant on overseas technology, restrictions may also increase R&D costs, extend validation cycles, and suppress short-term supply capacity. The direction of domestic substitution is thus clearer, but the speed of realization still depends on technological maturity and commercialization capabilities. Related analysis has already been conducted in 'Global Computing Power Resonance and Accelerated Domestic Substitution'. It should be noted that progress in the model and application layers, as well as domestic advantages in power supply and computing-power-electricity synergy, belong to a broader industrial background and are not direct arguments at the level of the aforementioned index holdings.

Allocation Framework and Exposure Check

The allocation framework we provided in 'Market Dynamics: Q2 (Gross Domestic Product) GDP 4.3% Growth Momentum Shifting to (Artificial Intelligence) AI and Semiconductors' is divided into three layers: GF CSI All Share Information Technology (Exchange Traded Fund) ETF (159939.SZ) corresponds to hardware realization, Hang Seng Tech (Exchange Traded Fund) ETF (3067.HK) corresponds to ecosystem monetization, and T. Rowe Price China Evolution Equity Fund (LU2187417386) corresponds to the physical infrastructure supporting large-scale (Artificial Intelligence) AI deployment. Whether this event changes the profitability path corresponding to any of these layers can be checked at two levels.

Firstly, industry classification and index constituents: According to the current CSI industry classification, data communications optical module manufacturers are categorized under communications equipment-related industries and are not included in the CSI All Share Information Technology Index. Similarly, no companies primarily engaged in data center optical modules are found among the existing constituents of the Hang Seng Tech Index. Therefore, under the current index structure, the direct holding exposure of these two passive instruments to this event is low. It should be noted that industry classifications, companies' main businesses, and index constituents may all be adjusted, and the results of regular rebalancing still need to be tracked. Actively managed products should be based on the most recently publicly disclosed holdings, and holdings may change after the disclosure period.

Secondly, holding structure and revenue sources: The core weights of the aforementioned indices are concentrated in segments such as semiconductor equipment and materials, domestic computing chips, panels, and printed circuit boards. Their demand is primarily driven by the domestic market and domestic substitution orders. This import restriction targeting optical modules does not constitute a direct revenue transmission path. Diversification provides another layer of buffer. As of December 31, 2025, the number of index constituents is close to 200, with the top ten collectively accounting for about 30%, and the weight cap for a single stock being approximately 4%. Many constituents already on various restriction lists are in the domestic substitution direction and have achieved revenue or profit growth in recent years supported by domestic demand, indicating that external restrictions have not completely interrupted their commercialization path. These list labels are existing information already priced by the market. However, some constituents still have overseas customers and exposure to global (Artificial Intelligence) AI capital expenditure. If subsequent rules expand to areas such as printed circuit boards, server components, power supplies, or cooling, the impact path will shift from sentiment transmission to earnings transmission.

Changes in Control Pathways and Key Tracking Points

The above checks address current exposure, but the more noteworthy aspect of this event lies in the change of the pathway itself. Unlike traditional export controls that restrict the outflow of US technology and products to China, the pathway involved in this report is closer to US domestic equipment authorization, import access, and supply chain security reviews. Reuters previously reported that the US Department of Commerce had shelved a set of import restrictions targeting China, including data center equipment, after a softening of trade relations. Subsequently, the (Federal Communications Commission) FCC advanced measures targeting categories such as drones and routers through its covered list and equipment authorization mechanisms. Regulatory tools targeting Chinese tech products are showing a trend of expanding from entity restrictions to product access, but the specific relationship between different departmental measures and trade negotiations still awaits formal document confirmation.

Boundary shifts are not hypothetical; precedents already exist. On July 22, 2026, the (Federal Communications Commission) FCC voted to approve a report and order that includes logic devices produced by entities on the covered list within the scope of restrictions. Equipment containing such devices, even if produced by a third party and not assembled in China, will also be unable to obtain equipment certification. The rule applies from the date of its approval, with no transition period. Software and firmware are not yet included. It also needs to be pointed out that this device rule currently only applies to list items identified by producer, and not yet to new categories identified by place of production. The current proposal targeting optical modules is closer to the latter. Two sets of determination criteria are operating in parallel and expanding independently. This is precisely the specific meaning of 'boundaries may shift'. What needs continuous tracking is not the measure targeting optical modules itself, but whether the determination criteria and applicable categories extend to index-heavy components.

Observation windows in the opposite direction also exist. It is reported that the heads of state of China and the US will meet in the US on September 24, with (Artificial Intelligence) AI listed as a discussion topic. Previously, foreign ministers and vice foreign ministers from both sides had conducted preliminary communications on this. If both sides reach an understanding on (Artificial Intelligence) AI-related issues, the pace of escalating restrictions might slow down. However, disagreements surrounding advanced model regulation, intellectual property, and export controls could also complicate the meeting. Therefore, it should not be assumed as a unilateral positive. Based on this, we define three conditions for the invalidation of our current judgment: if the final measures extend to complete machines containing relevant components or upstream devices; if restrictions expand to index-heavy components such as printed circuit boards, server components, power supplies, and cooling; or if supply chain adjustments significantly slow down the pace of global (Artificial Intelligence) AI capital expenditure. Should any of these situations occur, we will re-evaluate the profitability exposure and allocation judgment of relevant tools.

Short-term Impact and Medium-to-Long-term Direction

Synthesizing the previous two sections, short-term impacts require distinguishing the subject. For domestic optical module enterprises directly supplying US data center customers, the risk lies on the revenue side. For the passive index tools discussed in this article, based on the current constituent structure, the impact is expected to be indirectly transmitted primarily through risk appetite, sector valuation, and fund rebalancing. The nature of these two is different and should not be combined in description.

The medium-to-long-term direction has not changed due to this event; rather, it has become clearer. Global (Artificial Intelligence) AI is presenting two parallel ecological paths: one centered on large-scale capital expenditure by cutting-edge chip manufacturers and leading cloud providers, and the other, under conditions of restricted advanced processes, focusing on cluster scale, cluster-level effective computing power, and software-hardware synergy. These two paths are not mutually exclusive but form two exposures that should be covered in parallel. When access and certification systems diverge, the parallel operation of two supply chains may transition from a temporary phenomenon to a long-term state. This is also the process by which self-reliance and control move from an option to a prerequisite. It can be reasonably expected that domestic (Artificial Intelligence) AI infrastructure construction and application deployment will proceed at their own pace, and the proportion of required hardware supplied by domestic supply chains is expected to continue to increase. The divergence itself will raise construction costs on both sides and structurally divide their respective accessible markets. This is a process that unfolds over years, and its impact may not be reflected in short-term financial data.

Focus Targets and Allocation Framework

In summary, this event is currently insufficient to warrant an adjustment to the allocation structure. We maintain the existing three-layer framework unchanged.

Focus Targets	Code	Market/Domicile	Benchmark Index	Allocation Positioning
GF CSI All Share Information Technology (Exchange Traded Fund) ETF	159939.SZ	Shenzhen Stock Exchange	CSI All Share Information Technology Index	Hardware Realization
Hang Seng Tech (Exchange Traded Fund) ETF	3067.HK	Hong Kong Stock Exchange	Hang Seng Tech Index	Ecosystem Monetization
T. Rowe Price China Evolution Equity Fund	LU2187417386	Luxembourg (Undertakings for Collective Investment in Transferable Securities) UCITS	MSCI China All Shares Index	Physical Infrastructure

Source: Public information from fund managers and index companies, compiled by iFAST Research.

The above allocation corresponds to a medium-to-long-term perspective. Current (Artificial Intelligence) AI capital investments are expected to gradually translate into revenue and profit over the next two to three years. Investors can determine the relative weights of each layer based on their risk tolerance and investment horizon. Short-term price fluctuations triggered by a single policy announcement, if not accompanied by substantial changes in profitability paths and valuation centers, are usually insufficient to solely justify adjusting a long-term allocation framework. However, if fluctuations significantly alter valuation levels, position concentration, or the portfolio's risk budget, a rebalancing assessment at the portfolio level is still required.

Related Articles

   Global Computing Power Resonance and Accelerated Domestic Substitution

   Structural Bull Market Logic Unchanged: From Valuation Repair to (Artificial Intelligence) AI Engine

   Market Dynamics: Q2 (Gross Domestic Product) GDP 4.3% Growth Momentum Shifting to (Artificial Intelligence) AI and Semiconductors

Risk Warning

   Uncertainty of Rule Boundaries: The plan has not been publicly finalized. The criteria for determining new models, the continuity of existing certifications, the treatment of overseas capacity, and whether upstream components are covered, all remain unclear. If the final scope is stricter than currently reported, the affected range will expand accordingly.

   Spillover of Controlled Categories to Index-Heavy Components: Both determination criteria (by producer and by place of production) are expanding. If extended to areas such as printed circuit boards, server components, power supplies, or cooling, the impact on relevant constituent stocks will shift from valuation and sentiment to the revenue level.

   Two-way Volatility in Geopolitical Dynamics: The outcome of the US-China presidential meeting in mid-September is uncertain. Both understanding and escalating friction are possible, which could have opposing effects on sector risk appetite.

   Slowdown in Global (Artificial Intelligence) AI Capital Expenditure: Supply chain divergence raises construction costs on both sides. If this delays data center delivery progress, earnings forecasts for related sectors will face downward revision pressure.

Disclaimer

Important Notice

This report is compiled by the iFAST China Research Team for general informational purposes only and is publicly distributed to general investors. This report does not constitute, and should not be construed as, an offer to sell, a solicitation to buy, or investment advice regarding any securities, funds, or investment products, and does not take into account the investment objectives, financial situation, or specific needs of any particular recipient.

The information, views, forecasts, and valuations contained in this report are based on public information and independent judgments that our team believes to be reliable. However, the Company makes no express or implied warranty as to their accuracy, completeness, or timeliness. Forward-looking statements in this report are subject to uncertainties, and actual results may differ materially.

Past performance is not indicative of future results. The value of investments and the income derived therefrom may fall as well as rise. Investors may not get back the original amount invested. If an investment is denominated in a foreign currency, changes in exchange rates may also adversely affect the value of the investment.

Investors should independently and carefully assess the relevant risks before making any investment decisions and, if necessary, consult independent financial, legal, or tax professional advice regarding their individual circumstances.

The Company and iFAST Group's related licensed entities, including (iFAST Platform Limited) IFPL, are regulated by relevant jurisdictional regulatory bodies. For more information about the Company and this report, please visit the official iFAST website.

Analyst Disclosure

As of the date of this report, the author of this report, Ian Li, (Chartered Financial Analyst) CFA, and his immediate family members, hold no positions or interests (NIL) in the securities or related investment products mentioned in this report. The author's compensation is not directly linked to the specific views expressed in this report.

Artificial Intelligence Usage Statement

This report may have utilized (Artificial Intelligence) AI-assisted tools during the writing, data compilation, and chart creation processes. All related outputs have been independently reviewed and verified by our research team.