Hook: A Supply Squeeze That Speaks in Numbers
In the third week of January, Lumentum’s CEO reported something unusual: “We are seeing a sudden surge in demand for CW DFB lasers that we did not anticipate.” Not a vague nod to industry trends. A specific confession. On the same earnings call, AAOI stated a 20% to 40% supply gap, with customers calling weekly. MACOM echoed the sentiment, calling indium phosphide DFB lasers “tight across the board.” Three independent sources. One pattern. The question is not whether the bottleneck exists—it does. The question is whether it is a real structural barrier or a carefully managed narrative.
Context: What Tight Supply Actually Means
CW—Continuous Wave—DFB lasers are the unsung workhorses of silicon photonics. In a CPO (co-packaged optics) architecture, the laser is the light source. It must deliver high output power to compensate for silicon modulator losses, maintain narrow linewidth and low relative intensity noise for PAM4 modulation, and survive the thermal hell of being seated next to a switching ASIC. This is not a commodity component. It requires indium phosphide substrates, precision epitaxial growth, DFB grating engineering, hermetic sealing, and die-to-die coupling with silicon photonic integrated circuits. The barrier to entry is not a silicon wafer fab. It is a decade of accumulated process knowledge.
Core: The Three-Way Verification
Let the data speak. Lumentum reported that demand for its CW lasers exceeded internal forecasts, allowing the company to raise prices. AAOI quantified a 20% to 40% gap between customer demand and available supply, with some clients escalating to weekly calls. MACOM confirmed that its entire indium phosphide DFB portfolio is under pressure. Three revenue cycles, all pointing to the same conclusion: the current supply of high-performance CW lasers is insufficient to meet the immediate demand from 800G and early 1.6T optical modules.
But here is where the detective work gets interesting. Lumentum stated it expects first-generation CPO laser shipments to begin only in the second half of 2027. That is two years away. The current bottleneck is for pluggable modules, not for CPO. The narrative that “CPO laser supply is tight” conflates two different markets. The technical challenges are related, but the timelines are not. The lasers being rushed today are for existing form factors, not for the next-generation light engine.
Contrarian: The Optimism Discount
Both AAOI and Lumentum publicly stated that Chinese manufacturers are 2 to 3 years behind in qualifying for first-generation CPO laser supply. This sounds like a fact. But it is a statement from parties with a financial incentive to maintain a perception of high barriers. Chinese optical module companies have demonstrated rapid catch-up in the pluggable space, sourcing indium phosphide chips from domestic fabs. The 2-to-3-year gap may be real for the specific use case of CPO—where reliability certification and system-level co-design are critical—but it is not a static gap. It is a gap that will shrink as Chinese fabs invest in higher-end epitaxy and packaging.
The hidden variable is not capability but certification. Even if a Chinese supplier can produce a CW laser with comparable performance, the qualification cycle for a hyperscaler’s CPO platform is 12 to 18 months. The gap is real, but it is a timeline gap, not a technology gap. The market is pricing in a permanent advantage for incumbents. I would argue that is overconfidence.
Takeaway: Follow the Fab, Not the Hype
The supply bottleneck is real, but it is a temporary bottleneck in a specific segment—pluggable optics—not a structural moat for CPO. The real signal to watch is not the price of CW lasers today, but the expansion plans of Lumentum and Coherent. If they break ground on new indium phosphide capacity within the next two quarters, the bottleneck narrative shifts. If they rely on pricing alone, the slowdown is manufactured. History repeats, if you read the chain. Or in this case, the fab. Ledgers don’t lie. Anomaly detected. Look closer.