Over the past quarter, three laser manufacturers — Lumentum, AAOI, and MACOM — reported synchronized supply crunches. Lumentum cited "unexpected demand surge" with pricing power. AAOI flagged a 20–40% gap and weekly customer escalations. MACOM noted tight InP DFB inventory. The data points are consistent. But the market is pricing in a CPO revolution that is at least two years away. This divergence is a signal.
Context: Why a laser shortage matters for crypto.
Blockchain networks are scaling. Layer-2 rollups, zk-proof aggregation, and decentralized storage (Filecoin, Arweave) demand high-throughput data centers. Co-packaged optics (CPO) — where lasers sit directly next to switching ASICs — is the proposed cure for power and bandwidth bottlenecks. The CW (continuous wave) DFB laser is the heart of CPO. It must deliver high power, narrow linewidth, and survive near-switch temperatures. This is not a silicon race. It is a compound semiconductor game: indium phosphide (InP) epitaxy, grating design, hermetic packaging. The know-how is concentrated.
Core: The supply chain layers — and the China gap.
| Tier | Players | Current State | |------|---------|---------------| | 1 | Lumentum, Coherent, Broadcom | Mass production of high-performance CW lasers, deep CPO ecosystem involvement | | 2 | AAOI, MACOM | InP DFB products exist, but focus on pluggable/NPO | | 3 | Sivers, Chinese vendors | Technology in development, not certified for first-gen CPO |
Based on my audit of hardware supply chains for crypto mining operations (2018–2022), the bottleneck pattern is familiar. ASIC manufacturers faced similar wafer capacity constraints. The difference here is verification: Lumentum’s earnings call transcript is a data point, not a proof. I have cross-referenced their statements with AAOI’s — the direction is consistent. But the magnitude is unverified. The 20–40% gap is a claim, not a measurement.
The "China lag 2–3 years" narrative is particularly suspect. Both AAOI and Lumentum have incentives to maintain pricing power. Chinese firms have caught up in pluggable optics — they are not incompetent. The gap is real in high-end InP epiwafer and automotive-grade reliability, but the timeline is softer than executives imply. Verification is the only trustless truth.
Contrarian: The real bottleneck is not CPO — it is legacy.
Every laser manufacturer is currently shipping for pluggable 800G/1.6T transceivers, not for CPO. Lumentum explicitly stated CPO revenue will not materialize until H2 2027. The current shortage is driven by AI data center demand for existing optics, not future CPO. The market is conflating two time horizons. The supply chain stress today is real, but it is a stress test for the old architecture. The new architecture (CPO) is still a lab experiment with certification timelines.
The hidden detail: Lumentum claims they can increase laser output — but that statement is fuzzy. Increase by how much? At what cost? Without raw numbers, it is noise. Silence in the supply chain data speaks louder than hype.
Takeaway: Position for divergence, not linearity.
Investors should treat the CW laser bottleneck as a real but temporary event. The key risk is over-valuation of CPO-exposed stocks based on a narrative that will not deliver revenue for 2+ years. The null hypothesis: the shortage will ease before CPO volume ramps, as legacy demand peaks and Chinese capacity catches up. I trust the null set, not the influencer. Watch actual CPO certification announcements and capacity expansion plans. The code — the supply chain data — is the only truth. Metadata is just data waiting to be verified.