Optical Component Supplier Lumentum Sold Out Through 2029
nashnova research
Lumentum CEO Michael Hurlston says the company's optoelectronic capacity is completely sold out, with supply tightness lasting into 2029 — the AI data-center buildout is pushing the optical supply chain to its limits.
How big is the capacity gap?
Around 70% of demand for some products cannot be met next year; other products face a 30% shortfall lasting through 2028.
Hurlston had previously said capacity would be sold out before 2028. This latest statement means the squeeze is even tighter than expected.
His words: "We are completely sold out and we can't see the end of it."
Why is new capacity so slow to come?
Lumentum's core product is indium phosphide devices — a semiconductor material used for high-speed data transmission. Building new production facilities takes three to five years.
This means → even if construction starts today, new capacity won't come online until 2028–2030 at the earliest.
In plain terms = optical components can't be scaled up like software. The factory construction cycle itself is the biggest bottleneck.
Why did Nvidia invest directly?
Earlier this year, Nvidia invested $2 billion each in Lumentum and rival Coherent Corp.
This means → optical components have become a chokepoint in AI infrastructure, and Nvidia needs to lock in supply ahead of time.
This reflects a broader pattern in the AI arms race: downstream giants are reaching further upstream — whoever controls critical components controls the pace of compute expansion.
What are the hyperscalers doing?
Amazon, Microsoft and other hyperscale cloud providers have begun absorbing part of the capital-expenditure risk, backing suppliers with long-term agreements.
Hurlston said: "These hyperscalers are willing to fund it."
In plain terms = suppliers used to build factories and bear the risk alone. Now buyers are so desperate they will put up money first, just to secure priority access.
What are investors worried about?
Some investors are cautious about the rising costs and debt levels in global data-center and AI infrastructure buildout.
This means → the hyperscalers' willingness to backstop suppliers serves two purposes: locking in supply, and signaling to the market that AI spending will not slow down over short-term cost pressure.
Suppliers and buyers are now deeply tied together — there is no sign this supply chain will loosen any time soon.
市场有风险，内容仅供研究参考，不构成投资建议。