$AXTI $COHR $LITE $NVDA CURRENT CHINESE INP EXPORT-CONTROL REGIME
China’s formal InP control is not a blanket embargo; it is a discretionary licensing regime. On February 4, 2025, China’s Ministry of Commerce and General Administration of Customs issued Announcement No. 10, adding indium-related items to China’s dual-use export control list. The controlled indium items include 3C004.a indium phosphide, 3C004.b trimethylindium, 3C004.c triethylindium, and 3E004 technology and data for producing 3C004 items, including process specifications, process parameters, and processing procedures. Exporters must apply to the competent commercial authority of the State Council for a license before exporting the listed items, and the announcement took effect immediately.
The broader legal framework became more institutionalized shortly before the InP action. China’s State Council dual-use export control regulations took effect on December 1, 2024 and define dual-use items as goods, technologies, and services usable for civil or military purposes or capable of contributing to military potential, especially weapons of mass destruction and delivery systems. The framework includes license management, control lists, and supervision mechanisms. In practice, this gives MOFCOM and customs the ability to review not only the item but also destination, end user, end use, intermediaries, and policy sensitivity.
The item-level distinction matters. Refined indium metal itself is not listed in Announcement No. 10’s indium item text; the formal listed items are InP, trimethylindium, triethylindium, and related production technology. However, the practical control perimeter appears to be widening. Reuters reported in June 2026 that Chinese customs scrutiny of indium metal exports has increased, including requests for end-user details and longer approval times for some buyers, even though Reuters had not identified blocked indium metal shipments. This creates upstream optionality for China: even if InP licenses are granted, raw indium availability, paperwork timing, and inventory strategy can become additional pressure points.
AXT’s own disclosures show that the licensing regime has had real revenue consequences. AXT stated in its 10-K that the InP export portal opened in March 2025, that Tongmei received initial permits on June 11, 2025 to resume shipments to certain customers in Europe and Japan, and that timing for permit review and approval remained uncertain, unclear, and outside its control. The company also disclosed that it was unable to estimate when it would receive the necessary permits to resume InP substrate shipments to the U.S. In January 2026, AXT reduced Q4 2025 revenue expectations to $22.5 million to $23.5 million primarily because China’s Ministry of Commerce issued fewer InP export control permits than expected.
The U.S. destination remains the highest-friction lane. On AXT’s Q1 2026 call, management indicated that U.S.-based export permits were still pending, while permits for U.S. customers based in other global regions were being obtained more readily. Management also said MOFCOM had contacted AXT on several U.S. applications to request additional data, which was characterized as a sign that U.S. applications remained under review rather than categorically rejected. This creates a nuanced risk map: China appears willing to allow some InP exports, especially to Europe, Japan, and certain non-U.S. endpoints, but direct U.S. shipment approval remains less predictable.
INDUSTRY SUPPLY STATE
The InP substrate market is structurally concentrated and capacity-constrained. Reuters reported that AXT and Sumitomo dominate global InP substrate production, with Sumitomo at roughly 40 percent share, AXT at roughly 35 percent share, and JX Advanced Metals at roughly 10 percent. Reuters also reported that AXT is the 2nd-largest InP substrate producer, that AXT and Sumitomo together account for almost 80 percent of global InP substrate manufacturing, and that China’s export restrictions have pushed the average 6-inch InP wafer price up 250 percent to approximately $5,000.
The upstream mineral position reinforces the policy leverage. USGS estimates China accounts for 70 percent of global indium production, while the U.S. has 100 percent net import reliance for indium. USGS also notes that InP-based substrates are used in fiber-optic telecom networks, lasers, receivers, high-speed photodetectors, and optical communications, and that AI is expected to increase demand for specialized chip materials made with InP.
Demand is accelerating from multiple optical vectors: 800G and 1.6T transceivers, EMLs, photodiodes, CW lasers, CPO, OCS, and AI data center interconnects. Coherent said industry-wide InP capacity is a constraint, that it expects to double internal InP output capacity by the end of calendar 2026 1 quarter ahead of plan, and that it expects to more than double internal InP capacity again by the end of calendar 2027. Coherent also stated that 6-inch InP yields more than 4x as many devices at less than 0.5x the cost of 3-inch in its production context, explaining why 6-inch substrates carry strategic value beyond simple wafer count.
Coherent’s parallel U.S. capacity build does not make the AXT agreement redundant. Coherent announced a proposed $50 million CHIPS Act award to expand its Sherman, Texas 6-inch InP facility, including doubled production space and quadrupled wafer production capacity. That reinforces the strategic direction but also highlights the bottleneck: Coherent is scaling downstream and internal InP production aggressively, yet still prepaid AXT for external 6-inch substrate capacity. The rational interpretation is that Coherent needs redundancy, near-term substrate insurance, and supplier diversity while its own capacity ramps.
IMPACT ON THE AXTI-COHR DEAL
The key risk is not whether Coherent wants the wafers. Coherent’s demand appears robust, its AI optical roadmap is accelerating, and the cash prepayment is small relative to Coherent’s size. The risk is whether AXT can turn Beijing-based capacity into exportable wafers for Coherent’s required destination and end use. If the wafers are shipped directly from China to Coherent’s U.S. operations, the permit risk is high because AXT has disclosed persistent uncertainty around U.S. permits. If the wafers can be shipped to Coherent affiliates, contract manufacturers, or supply-chain nodes outside the U.S., risk is lower but still not eliminated. If the wafers support China-based or Asia-based optical production, export friction could be lower, but the value to Coherent’s U.S. Sherman expansion and CHIPS narrative would be less direct.
The agreement may signal commercial confidence, but it should not be interpreted as evidence of license approval. Coherent’s CEO reportedly raised InP license delays during a U.S. business delegation visit to China in May 2026, and InP export controls were reportedly discussed in U.S.-China trade talks. The deal was signed after those discussions and after Coherent’s CHIPS announcement, so it may reflect improving visibility or a negotiated path. However, the 8-K contains no disclosure that MOFCOM has approved Coherent-specific permits, and the structure gives Coherent refund rights if AXT cannot meet capacity commitments for more than 6 successive months.
For AXTI, the agreement is strategically bullish on customer validation and pricing power but not fully bullish on revenue certainty. It confirms that a major photonics supplier is willing to fund AXT’s 6-inch InP capacity, which supports AXT’s claim that capacity is the binding industry variable. It also strengthens the case that AXT’s Beijing InP assets have strategic scarcity value. However, because the capacity expansion is in Beijing, export approvals remain the conversion gate. If permits are granted smoothly, AXTI can convert backlog into revenue, improve utilization, and potentially sustain higher gross margins through richer InP mix and price increases. If permits remain slow, AXTI may carry a refundable customer liability, add capacity that is harder to monetize with Western customers, and face a mismatch between investor expectations and shippable revenue.
For COHR, the agreement is best viewed as an option-like supply hedge. Coherent is paying a modest amount relative to its revenue base to reserve a scarce upstream input that could gate much larger AI optical revenue streams. If permits are granted, Coherent gains secured 6-inch substrate access at agreed pricing and improves its ability to ramp 1.6T, CPO, and other InP-based products without relying solely on internal capacity, Sumitomo, or JX. If permits are delayed or denied, Coherent’s downside is partly protected by refund rights, but its real cost is lost time, constrained internal fab utilization, and potential inability to satisfy hyperscaler demand on desired schedules.
The deal also creates a geopolitical contradiction. Coherent is receiving U.S. industrial-policy support to expand domestic InP photonics manufacturing in Texas, while this agreement depends on 6-inch InP wafer substrate capacity being expanded at AXT’s Beijing facility. That does not invalidate the agreement, because global optical supply chains remain deeply cross-border and high-quality InP substrates are not readily substitutable. It does mean that COHR’s “domestic supply chain resilience” narrative remains incomplete until either internal substrate/device capacity is sufficient or non-China sources can support the required scale.
The contract’s commercial architecture is consistent with a constrained market. Coherent receives committed capacity and access to incremental excess capacity on the same terms. AXT receives customer funding and a minimum order framework. Both parties receive downside protections. This is not a simple spot supply agreement; it is closer to a risk-sharing mechanism around scarce, strategically controlled 6-inch substrate capacity. In a normal oversupplied substrate market, a customer would have less reason to prepay $22.3 million and accept minimum order obligations. In the current InP market, the prepayment is a rational cost of allocation security.
SCENARIO ANALYSIS
In the constructive scenario, China grants Coherent-specific licenses, potentially first for non-U.S. delivery lanes and later for U.S. endpoints. AXTI then benefits from a high-quality anchor customer, customer-funded capex, improved visibility, and the possibility that other tier-1 customers seek similar long-term capacity agreements. COHR benefits from a second major external source of 6-inch InP substrates while internal Sherman, Sweden, and Zurich capacity ramps. This scenario supports higher confidence in AXTI backlog conversion and reduces the risk that COHR’s AI optical revenue is gated by substrate availability.
In the adverse scenario, China continues to delay or withhold permits for U.S.-linked Coherent shipments. AXT can still have strong demand and even strong domestic China demand, but the Coherent contract may not translate into expected Western revenue. Coherent could recover unused prepayment under the disclosed refund mechanics if AXT fails the capacity commitment for more than 6 successive months, but the strategic loss would be shipment delay rather than cash loss. AXTI would face greater risk that the market values its capacity as China-local rather than globally monetizable. COHR would remain reliant on internal capacity growth, Sumitomo/JX allocation, and alternative routing through non-U.S. manufacturing nodes.
In the mixed scenario, China approves exports selectively, with destination and end-use conditions. This is arguably the highest-probability operating pattern based on AXT’s disclosures: Europe and Japan permits have been granted, U.S. permits have remained uncertain, and U.S. customers outside the U.S. appear easier to serve. Under this scenario, the AXT-Coherent agreement still has value, but the realized value depends on Coherent’s ability to use wafers outside the U.S. or to map end-use documentation convincingly to civilian AI data center applications. The commercial impact would be positive but operationally complex, with shipment timing, routing, and compliance workstreams becoming central to quarterly revenue recognition.
KEY DILIGENCE POINTS
The next critical disclosure is the Q2 2026 10-Q exhibit. The most important terms will be whether export-control denial or delay is expressly treated as force majeure, whether capacity commitment means manufactured capacity or delivered/exportable capacity, whether Coherent’s destination is specified or flexible, whether title transfers inside or outside China, whether pricing is fixed or indexed, whether permits are conditions precedent, and whether AXT can satisfy obligations through affiliates or alternate facilities. These terms will determine whether the contract is a hard revenue bridge or a contingent allocation framework.
Subsequent permit language in AXTI commentary is more important than backlog language. AXT has already demonstrated strong demand, capacity expansion intent, and customer interest. The incremental investment debate is now about exportability. Any disclosure that AXT received permits for Coherent, resumed meaningful U.S. InP shipments, or recognized Coherent-related revenue would materially improve the quality of the deal. Conversely, commentary that permits remain pending, that shipment timing is fluid, or that Coherent deliveries are routed away from the U.S. would keep the deal positive but only partially de-risked.
Coherent commentary should be monitored for whether AXT supply is framed as substrate security, LTA-driven capex support, internal fab feedstock, or general supply diversification. If Coherent explicitly ties AXT supply to Sherman, 1.6T, CPO, or Nvidia-related demand, the strategic value rises but the China approval sensitivity also rises. If Coherent frames it as global supply-chain flexibility rather than direct U.S. capacity feedstock, the permit risk may be lower but the read-through to domestic U.S. AI infrastructure is less direct.
BOTTOM LINE
The AXT-Coherent deal is a meaningful positive data point for AXTI’s strategic relevance and for Coherent’s urgency to secure 6-inch InP. It is also a clear confirmation that Chinese InP export controls are not a theoretical risk; they are shaping contract structure, customer funding behavior, capacity allocation, and supply-chain strategy. The right interpretation is “commercially validated but regulatorily contingent.” The deal improves demand visibility and customer alignment, but the investment conclusion should remain anchored to export permit conversion, not announced capacity or prepayment alone.
