The US is preparing to ban Chinese optical transceivers used in AI data centers, but the move exposes a dangerous supply chain vulnerability: American companies depend on Chinese components to build the very technology they're trying to restrict. The Federal Communications Commission (FCC) is drafting a rule that would block imports of new Chinese transceiver models before the end of 2026, marking the latest escalation in export controls targeting AI infrastructure.

Why Are Optical Chips Suddenly So Important to AI?

As artificial intelligence models grow larger and more complex, they require thousands of graphics processing units (GPUs) working in parallel. The problem: moving data between these chips fast enough. Copper wires, which have dominated data center connections for decades, hit a physical limit around one to two meters before signal degradation becomes unmanageable. Optical transceivers solve this by converting electrical signals into photons that travel through fiber optic cables, carrying far more data over longer distances with less power consumption.

The race to build faster optical systems has become as strategically important as securing GPUs themselves. Nvidia has committed $4 billion across suppliers Coherent and Lumentum to lock up supply. Marvell agreed to pay $3.25 billion upfront for silicon photonics startup Celestial AI, with potential earnouts reaching $5.5 billion. The money flowing into this sector signals that optical interconnects are no longer a niche technology; they're essential infrastructure for the next generation of AI.

Who Controls the Photonics Market Right Now?

China currently dominates the optical transceiver market by a significant margin. Chinese vendors account for nearly two-thirds of global transceiver unit shipments and roughly 60 percent of datacom transceiver revenue, according to market research firm Counterpoint Research. Innolight, which just completed a $6.8 billion Hong Kong initial public offering (IPO), leads with a 27 percent revenue share, ahead of Coherent's 17 percent and Lumentum at roughly 6 percent.

This dominance reflects both manufacturing scale and cost efficiency. Chinese companies have built the production capacity, cleanroom facilities, and supply chains that Western competitors are still racing to match. Analysts told the South China Morning Post that Western suppliers lack the packaging capacity and manufacturing scale to replace Chinese production within the next few years.

What Makes This Ban So Complicated?

The proposed photonics ban reveals the deep entanglement of US and Chinese supply chains in ways that simple export restrictions cannot untangle. Innolight and Eoptolink reportedly supply the majority of 800-gigabit-per-second (800G) optical modules used in Nvidia's AI clusters. But those Chinese companies pair digital signal processor (DSP) chips from Broadcom and Marvell with laser components from Lumentum, Coherent, and Mitsubishi Electric. Meanwhile, Western rivals need indium phosphide, a critical material for laser production that China has controlled since 2025.

The FCC's concern is legitimate: officials worry that Chinese optical transceivers could become vectors for data theft, malware, or service disruption, similar to how Huawei telecom hardware became so embedded in US infrastructure that removing it proved slow and expensive. However, the proposed rule overlooks a critical fact: the biggest customer of China's photonics hardware is the United States itself.

How Are Policymakers Trying to Manage AI Supply Chain Risk?

The photonics ban is just one piece of a broader strategy to control AI infrastructure. The Trump administration is simultaneously pursuing multiple export control measures:

• Optical Transceivers: The FCC is drafting rules to block new Chinese transceiver models from being imported, marketed, or sold in the US before the end of 2026.
• Foundry Due Diligence: House Select Committee on China Chairman John Moolenaar is pushing the Commerce Department to enforce existing rules requiring chipmakers like TSMC and Samsung Foundry to verify the true end customers of advanced chips, preventing Chinese companies from obtaining restricted AI processors through intermediaries.
• Hardware Bans: Recent FCC actions have targeted Chinese humanoid robots and power inverters, with reports indicating consideration of bans on Chinese AI companies and additional optical components used in data centers.

The enforcement of the Foundry Due Diligence Rule is particularly significant because it addresses a specific vulnerability: Chinese companies like Huawei obtained millions of advanced AI chips from TSMC by using a front company called Sophgo to misrepresent the end customer. Under the rule, foundries exporting chips produced on 14-nanometer or more advanced process technologies must presume those devices qualify as controlled AI processors subject to licensing requirements.

What Is China's Response to These Restrictions?

China has responded with a combination of defiance and domestic investment. The country's Foreign Ministry stated it "firmly opposes" the potential photonics restriction and accused Washington of overstretching national security concerns. Chinese officials have also rejected accusations of chip smuggling and stated that "China will take all necessary measures to firmly safeguard its legitimate rights and interests against any actions that substantially harm China's interests".

More significantly, China is building its own photonics ecosystem independent of US technology. Wuxi's CHIPX pilot line, the country's first dedicated photonic chip fabrication facility, began mass-producing six-inch photonic wafers in 2025. Shanghai opened a state photonic computing laboratory in June alongside newly listed startup Lightelligence. Crucially, Chinese officials note that photonic chips do not require extreme ultraviolet (EUV) lithography tools that Washington has cut off from China.

The broader pattern shows China strengthening domestic alternatives across multiple AI infrastructure layers. Chinese firms delivered 1.65 million AI GPUs in 2025, with homegrown suppliers like Huawei and Cambricon ramping up production as Nvidia's local market share fell below 60 percent. China has also begun blocking Nvidia H200 chip imports while certifying domestic accelerators for state procurement.

Why Are Some US Tech Companies Opposing These Restrictions?

Not all of Silicon Valley supports aggressive export controls. Nearly 200 US companies signed a letter in July urging the Trump administration to refrain from banning the use of Chinese AI models, arguing that Chinese models are cheaper and require less computing power than US alternatives. This reflects a genuine economic tension: companies that rely on cost-efficient infrastructure may face higher expenses if forced to use exclusively Western suppliers.

The photonics situation presents a similar dilemma. If the FCC bans new Chinese transceiver models, US data center operators would face immediate supply constraints and higher costs, even as they transition to Western suppliers. The rule does allow existing Chinese modules already installed in US data centers to remain operational, but this creates a patchwork of approved and restricted equipment.

What Happens Next in the US-China Tech Competition?

The photonics ban is being weaponized as a bargaining chip in broader trade negotiations. With a crucial US-China summit scheduled for September 2026, the AI supply chain has become a central negotiating point. The Trump administration's strategy appears to be using export controls as leverage to shape both technology development and geopolitical relationships.

However, history suggests that export bans often trigger circumvention attempts. Reports of Nvidia chip smuggling indicate that determined actors will find ways around restrictions. China has also demonstrated resilience by investing heavily in domestic alternatives whenever Washington tightens controls.

The photonics situation is particularly complex because it involves mutual dependencies that cannot be easily severed. Western companies need Chinese manufacturing capacity and indium phosphide supplies. Chinese companies need Western components and design expertise. A ban that ignores these realities risks disrupting US data center operations while accelerating China's investment in independent alternatives. The FCC's proposed rule may enhance national security in theory, but the practical implementation could prove far messier than policymakers anticipate.