Lasers & precision photonics

The choke, physically. A trapped-ion or neutral-atom quantum computer is mostly a laser system with a vacuum chamber attached. Each machine needs 5–10 distinct narrow-linewidth laser systems spanning UV to near-IR — cooling, state prep, gate drive, readout — plus dozens to hundreds of acousto-optic modulators (AOMs) doing the fast switching and per-qubit addressing (supply-chain map). The merchant suppliers are countable on one hand. TOPTICA (private, Munich, family-owned) is the default vendor for tunable diode lasers and Hz-linewidth clock lasers across both modalities; its clock-laser system alone has sold roughly 20–30 units into quantum computer makers (optics.org). Gooch & Housego (OTC: GCHHF) is the dominant merchant AOM house and built the first visible-wavelength (397–780 nm) AOM line specifically for quantum and atomic sensing (trapped-ion supply-chain teardown). Around them sit the system players: Coherent's cold-atom and quantum-optics portfolio, IPG's purpose-built single-frequency quantum lasers, and MKS's Newport/Spectra-Physics quantum line.

Why the moat is real. Hz-level linewidth at oddball atomic wavelengths (369 nm for Yb⁺, 422/674 nm for Sr⁺, UV Rydberg lines) is artisanal physics, not catalog optics. Qualification is sticky: gate fidelities are tuned against a specific laser's noise profile, so swapping vendors means re-validating the machine. And the served market is small enough — tens of millions of dollars — that no scaled competitor bothers to attack it. That cuts both ways: the moat is partly that the prize isn't yet worth fighting for.

Honesty about size. The chokepoint is real; the revenue is tiny. M Squared Lasers — a leading supplier of CW Ti:sapphire Rydberg-excitation lasers for neutral atoms — collapsed into administration in August 2025 with a £64m creditor deficiency, its assets bought back by founder-led Novacene Photonics (The Quantum Insider). Owning a chokepoint in a pre-revenue industry does not pay the bills. For Coherent (repriced on AI optics and a $2B NVIDIA investment), quantum is a rounding error; same for IPG and MKS.

What breaks it. Photonic integration: waveguide light delivery on ion-trap chips (MIT Lincoln Laboratory) and directly modulated lasers — Coherent's new Sapphire XT explicitly markets eliminating external AOMs, though it launched at 488/532/561 nm for life-science and semiconductor work, not the atomic-transition wavelengths this chokepoint trades on — would collapse per-qubit component counts. In-housing by scaled QC makers is the other vector. Both take years, and both still buy from the same handful of fabs in the interim.

The repricing event. DARPA's Quantum Benchmarking Initiative moved 11 companies into Stage B in November 2025 — four of them (IonQ, Quantinuum, Atom Computing, QuEra) trapped-ion or neutral-atom, the laser-hungriest architectures in the cohort. Stage C decisions, expected late 2026, would fund utility-scale builds: any of those four surviving means the first real volume purchase orders this supplier base has ever seen. Layer on the NQI Reauthorization, which cleared Senate Commerce unanimously in April 2026, and the demand signal turns federal. The tell to watch: G&H's record £167m order book at H1 FY26 — when AOM lead times stretch, this chokepoint reprices first.