NTT and OptQC signed a capital and business alliance on August 3, 2026, aimed at a fault-tolerant, million-qubit-class optical quantum computer by fiscal year 2030. What makes this worth reading past the headline number is the roadmap attached to it: a named sequence of milestones with dates, not a single target four years out with nothing in between.
The actual roadmap
The plan runs in phases. Fiscal year 2026 starts co-creation work with prospective users, industry partners, and research institutions to define use cases. Fiscal year 2027 focuses on building a supply chain toward practical application. Fiscal year 2028 is where the roadmap gets a checkable number: proof-of-concept projects with users based on a 10,000-qubit-class optical system. Fiscal year 2029 expands application development and verification, pushing toward real-world implementation. The million-qubit target sits at the end of that sequence, in fiscal year 2030.
That 10,000-qubit PoC in 2028 is the milestone worth watching. It's roughly two years out, specific, and produces a real system other people get to evaluate directly, rather than a press-release qubit count with no interim checkpoint attached to it.
Why photonic, and why this matters next to PsiQuantum and Xanadu
Photonic quantum computing uses photons as qubits, encoded in properties like path, polarization, or time-bin. The appeal, and the reason PsiQuantum and Xanadu also work this modality, is that photons don't need the same cryogenic cooling as superconducting qubits and reuse parts of the existing telecom manufacturing base. Our modality comparison tracks all three companies. A fourth serious, well-capitalized entrant with a dated roadmap is a real data point on whether the photonic bet is paying off, rather than a repeat of a claim two other companies have already made.
NTT brings something OptQC doesn't have on its own: an existing telecom-scale optical manufacturing and supply chain relationship, the same category of capacity that underlies the GlobalFoundries silicon photonics CHIPS Act award covered here recently. That overlap is not a coincidence. Photonic quantum hardware and next-generation optical telecom infrastructure draw on the same fabrication base, which is part of why government and corporate investment in one keeps showing up adjacent to the other.
What the announcement does not commit to
A "capital and business alliance" is a real financial commitment, but the release does not disclose the size of NTT's investment, nor does it specify a fidelity, error rate, or gate count target for the 2028 proof of concept, only a qubit count. Qubit count alone tells you little about whether a system is useful, a point this site makes about every hardware headline: see our coverage of logical qubits and fault tolerance and of Quantinuum's Helios encoding ratio for why the ratio between physical and logical, or the error rate per operation, decides whether a number is a real result or a marketing figure.
What to watch next
Fiscal year 2028 is the checkpoint. If the 10,000-qubit proof of concept ships with disclosed error rates and real use-case results from named partners, that is meaningfully different from a slide reused at each fiscal year-end. Until then, treat this as a well-structured, credible roadmap from a company with real telecom manufacturing backing, not yet a hardware result.