NLM Photonics is a Seattle-based deeptech developing silicon-organic hybrid (SOH) photonic integrated circuits that use a patented electro-optic polymer (Selerion™-HTX) to bring 1.6T and 3.2T modulation to standard silicon photonics foundry processes. This Series A extension is a $100K pro-rata follow-on from Fund I into a $13M round led by Pangaea, alongside strategics Emerald, Idemitsu, TOK, OVF, and Mitsubishi Diamond Edge.
The thesis: after eighteen months of technical execution, NLM has crossed from materials story to shipping-parts story — Tower Semiconductor validated the SOH modulator on its PH18M platform in March 2026, Cisco has taken delivery of 3.2T PICs, and Google, Foxconn, and Centera are all in active evaluation. Given the flat-round dynamics, our proprietary information advantage from having invested in the prior round, and the 25%+ CAGR industry backdrop where $15B+ has moved in silicon photonics M&A in the last twelve months, exercising pro rata is the low-cost way to protect a strategic-exit-shaped optionality without underwriting the full risk of a new check.
Photonics is one of the highest-conviction sub-sectors in semiconductors right now: the AI training footprint is the pull, and every hyperscaler and NIC/switch vendor is racing to solve the electrical-optical bottleneck at 1.6T and 3.2T lanes. NLM does not need to own the whole market — capturing even a mid-single-digit share of the electro-optic modulator layer inside the CPO stack would be a multi-hundred-million revenue business inside the SOM the Series B will need to underwrite. The relevant question is not TAM; it is which of the 3-4 organics/TFLN/plasmonics approaches becomes the reference architecture, and NLM is in the final heat.
| Company | Approach | Stage | Notes |
|---|---|---|---|
| NLM Photonics | Silicon-organic hybrid (SOH) with Selerion™-HTX polymer on standard silicon photonics Internal | Sampling 1.6T & 3.2T PICs; Tower PH18M PDK Web | Cisco PO in hand; Google, Foxconn, Centera in evaluation Internal |
| Polariton Technologies | Plasmonic modulators on silicon photonics Web | Acquired by Marvell April 2026 Web | Removed from independent market; validates 3.2T thesis |
| Lightwave Logic | Electro-optic polymer (competing chromophore chemistry) Internal | Public (LWLG); no PIC / no customer shipments per Brad Internal | Brad: “sharing 85% reliability data we achieved three years ago” |
| TFLN players (Liobate, QCi, others) | Thin-film lithium niobate modulators Web | Multiple entrants; TFLN is main 3.2T alternative Web | Yield challenges at 3.2T pushing customers to evaluate NLM Internal |
| Coherent / Lumentum (incumbents) | Legacy transceiver / laser vendors Web | Public, NVIDIA-backed ($2B each) Web | Buyers/partners for NLM's blueprint model, not head-to-head |
Moat: a proprietary organic electro-optic material (Selerion™-HTX) that has now been demonstrated in a commercial foundry PDK on Tower's PH18M platform. That combination — a patented material stack that a Tier-1 pure-play foundry has validated at process level — is not something a TFLN or plasmonic entrant can spin up in a quarter. Brad's claim of a one-year lead is consistent with what we can see externally: no competitor has publicly shipped a working 3.2T PIC to a named hyperscaler.
The traction shape is the classic deeptech J-curve: revenue looks tiny in absolute dollars but the composition changed — 2025 was material sales and small NREs, 2026 is paid PIC samples and cost-share partnerships with named strategics, and Centera alone is guiding to ~1,000 parts in 2026 with volume production 2028. Every sample is paid, per Brad, which is the discipline you want to see in a pre-revenue-scale deeptech.
The exit dynamic here is unusually favorable: Marvell just paid to acquire Polariton in April 2026 explicitly for 3.2T-and-beyond optical scaling; Marvell also acquired Celestial AI for $3.25B; NVIDIA has invested $2B into each of Coherent, Lumentum, and Marvell so far in 2026. Named acquirers who logically consolidate NLM's SOH stack: Marvell (already talking to NLM per notes, ex-Inphi relationships from Brad), Coherent, Lumentum, Cisco (Acacia parent, existing PO), Broadcom, Global Foundries. The scarcity of foundry-validated 3.2T modulator IP is the deal — most acquirers will pay for the material + PDK, not the P&L.