Nomura estimates that shipments of laser chips for optical communications will reach 1.542 billion units in 2028, up from 480 million in 2026, according to a chart from a Nomura note posted on X on 1 November. That is 3.2 times in two years, or about 79% a year, by our arithmetic.
The fastest-growing line in the chart is the 200G EML, from 50 million units to 271 million, or 5.4 times. CW lasers add the most units: 402 million, from 171 million to 573 million, the post says.
Business Pill · Lead Time Explained
An evergreen Business Engineer explainer on lead times. It teaches the general idea only and does not discuss Nomura’s forecast or laser chips.
The key insight: Nomura estimates optical laser chip shipments rising from 480 million units in 2026 to 1.542 billion in 2028, about 79% a year by our arithmetic, with 200G EMLs growing fastest at 5.4 times and CW lasers adding the most units, 402 million.
The chart, titled “Laser chip shipment volume forecast” and sourced to “Nomura estimates”, splits shipments into four types: 100G EML, 200G EML, VCSEL and CW laser, in millions of units.
For 2026 the segments read 192 million 100G EMLs, 50 million 200G EMLs, 67 million VCSELs and 171 million CW lasers. For 2028 they read 441 million, 271 million, 257 million and 573 million.
By our arithmetic, the 2026-to-2028 multiples are 5.4 times for 200G EMLs, 3.8 times for VCSELs, 3.4 times for CW lasers and 2.3 times for 100G EMLs. The chart also gives a 2025 base of 220 million units in total, so 2028 would be seven times that.
CW lasers account for 38% of the 1,062 million units added between 2026 and 2028, by our arithmetic, and their share of the total moves from 36% to 37%.
Laser chip shipments by type, million units, 2025F-2028F (Nomura estimates, chart as posted on X on 1 November 2026). The growth multiples and the 79% annual rate are our arithmetic.
Why CW Lasers Grow
A 404K Research note of 27 October, which summarises the same Nomura forecast, explains the split. It says “EML integrates a distributed feedback (DFB) laser with an electro-absorption modulator on a single InP die”, while silicon photonics designs “typically use an external CW laser to supply an unmodulated optical carrier”.
Its summary line is “Silicon photonics is capturing market share, but it still requires external light sources”. Per 404K, Nomura estimates that by 2027, “SiPh transceivers powered by continuous-wave (CW) lasers could capture 60% to 70% penetration in mainstream data center optical modules”.
The note adds that “Nomura expects SiPh to capture share from EML on supply chain diversification and cost advantages, but this does not signal the obsolescence of EML in high-speed optical modules”.
Where Supply Is Tight
404K reports that supply of mid-to-low-power parts is expected to improve, but that “high-power CW lasers exceeding 200 mW, 200G+ EMLs, and high-speed VCSELs could face persistent supply deficits across 2026โ2027”.
By our arithmetic, those three high-end categories in the chart (counting all 200G EMLs, all VCSELs and all CW lasers, since the chart does not separate the high-power and high-speed subsets) go from 288 million units in 2026 to 1,101 million in 2028, or 71% of 2028 shipments.
The post that shared the chart tagged Coherent, Lumentum, Applied Optoelectronics and Broadcom. As we read it, a market where unit volumes triple while the high-end parts stay short is one where pricing power sits with whoever owns laser fab capacity.
The Structural Read
As we read the chart with 404K Research’s summary, the CW laser line tracks silicon photonics: silicon-photonics designs typically use an external CW laser, so share gains for silicon photonics tend to show up as CW laser units.
As we read it, the shortages 404K describes sit inside the chart’s three faster-growing types: all 200G EMLs, VCSELs and CW lasers together make up 71% of 2028 units, by our arithmetic, though the chart does not separate the high-power and high-speed subsets.
404K Research, summarising Nomura, 27 October 2026
“Silicon photonics is capturing market share, but it still requires external light sources.”
Three Implications
FOR LASER MAKERS Nomura’s chart puts 2028 volume at 3.2 times 2026, by our arithmetic; the post tagged Coherent, Lumentum, Applied Optoelectronics and Broadcom.
FOR TRANSCEIVER DESIGNS Per 404K, Nomura estimates CW-laser silicon photonics could reach 60% to 70% of mainstream data center optical modules by 2027, with EMLs serving most of the rest.
FOR READING THE TOTAL 404K says the figures are shipment volumes, not price or revenue forecasts, and that VCSELs can be made on gallium arsenide rather than indium phosphide.
The framework describes its networking layer this way: “The interconnect that turns thousands of GPUs into a single coherent compute fabric.”
As we read it, laser chips are the light source of that layer’s optical links, and Nomura’s chart puts a unit count on its growth. The framework does not discuss the Nomura forecast.
We could not read Nomura’s note itself; the figures come from the chart as posted on X and from 404K Research’s summary. The note’s date, author and any price or revenue forecast are not shown.
The post gives 966 million units for 2027, while the chart’s 2027 segments add up to 956 million; we use the segment labels. 404K says the figures are shipment volumes, “not price, revenue, or earnings forecasts”, and that VCSELs can be made on gallium arsenide, so the total is not a measure of indium phosphide demand.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Nomura forecasts 1.542 billion optical laser chips shipped in 2028, 3.2 times 2026, about 79% a year by our arithmetic. 200G EMLs grow fastest, 5.4 times, and CW lasers add the most units, 402 million.
A note on sourcing. Forecast figures are Nomura estimates, read from the chart as posted on X on 1 November 2026 and from 404K Research’s summary; we could not read the Nomura note itself. Growth multiples, shares and the 79% annual rate are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
