Photonics Stocks Jump on Reuters-But the Real Story Is the $5,000 Wafer Squeeze

InP scarcity, not headline hype, is driving the move

This looks like a scarcity trade rooted in an InP chokepoint, with Reuters amplifying an existing supply bottleneck rather than creating demand from scratch. The clearest signal is not the headline itself but the price of the input: a 6-inch InP wafer has jumped 250% to US$5,000. That matters because InP has no substitute in the high-speed optical chips that power AI data centers. When a hard-to-replace base material starts repricing that sharply, access to feedstock becomes the real asset.

Why bulls see a moat

Bulls see a moat forming around firms with credible InP flow. Coherent warned of an InP shortage on an early-May earnings call, and within a week CEO Jim Anderson was in China with a U.S. delegation to raise delays in China's export licenses. If Beijing keeps the material tightly allocated, companies closest to permitted feedstock could gain pricing power and shipment priority while others fall behind.

Why bears see near-term risk

Bears are not arguing from ideology; they are arguing from logistics. The same export-license delays that create scarcity can also pressure margins, push shipments out, and leave investors holding a political problem that can overheat faster than supply can recover. If license friction eases before access converts into volume, the premium can compress quickly.

For now, though, the setup still looks more like a scarcity trade than a broad AI-theme rally.

Nvidia's photonics bet raises the stakes

The market is not reacting to a vague photonics theme. It is reacting to a major customer with real routing power deciding that light-based interconnects are worth helping build out. Nvidia's combined $4 billion investment in Lumentum and Coherent matters because it signals that optical networking is becoming more central to the AI stack. That is a strong validation of demand, even if it does not guarantee every invested company will benefit equally.

Why this changes the hierarchy

The mechanism is straightforward: higher bandwidth, lower power, and more resilient supply all support Nvidia's broader architecture as AI systems scale. By investing in companies developing optical and semiconductor technologies for future networking hardware, Nvidia is helping align supply around interfaces that matter for next-generation AI systems. For investors, that matters more than the Reuters headline itself, because it shows where strategic buyers see durable need.

There is also an alignment angle. Coherent warned of an InP shortage in early May, and Anderson subsequently traveled to China with a U.S. business delegation to raise delays in export licenses. Bulls see that as management putting its backlog and Nvidia-backed capacity on the line. Bears can reasonably counter that one trip does not restore flow. In a chokepoint market, diplomatic attention usually follows real allocation pressure, but it does not automatically resolve it.

China controls the bottleneck

This is where the scarcity gets real. China produces nearly 70% of the world's indium, and AXT and Sumitomo control nearly 80% of substrate manufacturing. That means photonics demand cannot simply jump to an alternative basin the way it might with a more commoditized component. Beijing already put indium phosphide on an export control list in February 2025, and Reuters reports growing scrutiny around indium exports even though the metal itself is not formally controlled yet. If the raw-material pipeline tightens further, the bottleneck moves upstream before final chip volumes fully show up in revenue.

AXT's own challenge says a lot: InP export permits are its biggest current challenge. So the real issue here is not just "AI exposure." It is who has controlled access to substrate and whether that access can turn into pricing power as demand keeps pulling harder.

Which names benefit if the squeeze lasts?

The cleanest winners are the firms with validated demand and a path through the bottleneck. Nvidia's prior combined investment in Lumentum and Coherent is the clearest alignment signal: these are not fringe photonics proxies. They are companies the system operator helped de-risk. If delays in China's export licenses persist, the market should keep rewarding companies closest to actual shipment capacity and customer allocation, not just pure substrate leverage.

That makes AXT a selective name, not an automatic buy. It sits right on the chokepoint, and management has said InP export permits are its biggest current challenge. Bulls can argue scarcity gives it longer-term pricing power. Bears will note that short-term operating pressure comes first when permits stall. My view: AXT is a high-beta exposure to the squeeze, which means it can outperform on confirmation and underperform sharply on disappointment.

What would confirm the trade

• Persistent delays in export licenses rather than a quick policy unwind.
• Wafer prices staying elevated, which would suggest scarcity is still binding.
• More evidence that demand is concentrating around companies with verified access and customer backing.

What would break it

• A rapid easing of export-license delays.
• Any clear sign of wafer price normalization.
• Longer-term progress toward Chinese self-reliance, including efforts around a coordinated national effort for greater semiconductor independence. That would not solve the shortage quickly, but it would weaken the scarcity thesis over time.

For now, the better question is not whether photonics matters. It is who actually controls flow through the bottleneck and who gets paid when scarcity stays tight.