Optical Interconnect Market to Hit $144B by 2030

The Photonics Buildout Just Got a $144 Billion Price Tag

For months we've been writing about the same migration: data moving on light instead of copper inside the AI data center. The argument was always physics and power. Now it has a market forecast attached. A new study from China Insights Consultancy projects the data center optical interconnect market growing from $13.7 billion in 2024 to $144.4 billion by 2030. That's a 48.1% compound annual growth rate. More than tenfold in six years. And nearly two-thirds of that future revenue, some $91.9 billion, is expected to land in one technology: silicon photonics.

Where the number comes from

The report was commissioned by Yuanjie Semiconductors, a Chinese laser-chip maker, as part of its Hong Kong IPO filing. Keep that in mind. An IPO sponsor benefits from a big market slide. But the forecast draws on LightCounting data and industry interviews, and LightCounting is the same research base the optics industry itself budgets against. We'd discount the precision, not the direction.

Three findings matter more than the headline number:

1. The curve steepens after 2027. The forecast is back-loaded on purpose. That's when co-packaged optics is expected to move from pilot programs to volume deployment.

2. Silicon photonics goes from niche to majority. Its share climbs from 16.6% in 2020 to a projected 63.7% by 2030. That isn't just a growing market. It's a share shift inside a growing market, and share shifts are where the violent repricings happen.

3. The speed ladder keeps climbing. Demand for 400G parts is expected to flatline as 1.6T becomes the standard. Whoever ships at the top of the ladder keeps the margin.

The forecast isn't waiting for 2030

Money is already moving. The industry has put more than $15 billion into co-packaged optics, photonic chips, higher-speed transceivers, and fiber over the past year. Buying startups. Forming alliances. Locking up capacity. Nvidia $NVDA partnered with Corning this spring to expand US manufacturing of optical connectivity, and analysts covering the deal said the quiet part out loud: the scarcity of advanced photonics now rivals power availability as the gating factor for AI infrastructure.

We made the demand-side case after Nvidia's $96 billion quarter. This report is the supply-side answer. The bottleneck has moved from GPUs to power to the optical layer itself, and the companies that own optical capacity own the choke point.

Reading it as a trader

The merchant optics names, Coherent $COHR and Lumentum $LITE, already trade on this thesis. They're the consensus expression, and they've been paid. The more interesting question is who owns the technologies driving the share shift. If silicon photonics really takes 64% of a $144 billion market, the winners aren't just the companies selling today's transceivers. They're the ones whose platforms were built for the co-packaged, chip-level future the forecast describes.

Where POET sits on the curve

POET Technologies $POET is one of the few small caps positioned directly inside that share shift. Its Optical Interposer platform targets the integration layer the CIC report says drives the market, and the commercial side has started to show. A $50 million initial purchase order from Lumilens under a framework worth up to $500 million. Six consecutive quarters of sequential revenue growth, with Q2 up 112% year over year. $796.3 million in cash to fund the ramp. Management's stated capacity target is up to 1 million units per month by the end of 2027, which lands right where CIC's curve turns vertical. The company has committed to a September update on the ramp. That's the near-term test. The 2030 number is the backdrop.