Xanadu and Mitsubishi Chemical Take EUV Photoresist Simulation Into Phase Two With Canadian and Japanese State Backing

by Karl Lindström · August 31, 2026

Semiconductors & Hardware Reporter, Frontier Tech Wire

Xanadu Quantum Technologies (Nasdaq and TSX: XNDU) said last week that its collaboration with Mitsubishi Chemical on semiconductor manufacturing chemistry is moving into a second phase with support from government innovation programmes in both Canada and Japan. The announcement went out over GlobeNewswire on August 25, 2026, with an updated version of the release issued on August 27.

The subject is narrower and more interesting than the usual quantum-for-chemistry press release. The two companies are working on extreme ultraviolet lithography photoresists - the light-sensitive polymer films that turn an EUV exposure into a pattern on a wafer - and specifically on the radiation-induced blur that limits how finely those films can resolve a pattern. Quantum Zeitgeist's account of the announcement describes that blur as a major obstacle to simulating EUV lithography, and says chip production currently relies on increasingly complex and computationally expensive classical simulations to mitigate it. That is the stated justification for reaching for a quantum simulator at all; the release itself offers no physical account of the blur mechanism, and none is asserted here.

What phase one actually established

According to the release, phase one demonstrated that quantum algorithms "can accurately model key aspects of EUV lithography, specifically the optical properties of the photoresists used to etch lithographic patterns." Quantum Zeitgeist's account describes the same result. That is a modelling validation, not a manufacturing result: it establishes that the method reproduces known optical behaviour, which is the precondition for trusting it on the harder problem. The release attaches no accuracy metric, benchmark or comparison to a classical method to that claim.

Phase two is where the ambition sits. The release says the partners aim to develop an industry workflow that integrates parameters derived from Xanadu's quantum simulations into Mitsubishi's multi-scale models, to predict blur and to identify materials that prevent it for next-generation semiconductor manufacturing. In practice that means the quantum calculation would supply parameters to a classical multi-scale simulation stack that resist chemists already use - a hybrid arrangement, not a replacement of existing tooling.

It is worth being precise about what has and has not been shown. The companies are describing a workflow they intend to build. There is no claim in the release of a quantum advantage over classical methods for this problem, no reported speed-up, and no statement that any resist formulation has been changed as a result. The strongest language in the announcement is the chief executive's - a marketing characterisation, not a measured outcome, and treated as such below. This is early-stage industrial R&D with a government cost-share, and should be read as such.

The state money, and its size

The Canadian side is the National Research Council of Canada's Industrial Research Assistance Program, NRC IRAP. The Japanese side is the Strategic Innovation Promotion Program, or SIP, led through the National Institute of Advanced Industrial Science and Technology (AIST) and its Global Research and Development Center for Business by Quantum-AI Technology, known as G-QuAT.

The only funding figure in the release is historical: it says Xanadu has received more than $800,000 in funding from NRC IRAP to date, without specifying the currency or the period over which it was provided. The release does not put a number on the new phase from either country, does not say whether the new support is awarded, contracted or announced in principle, and does not describe any cost-share split. The scale of the commitment behind this announcement is therefore not public, and that absence is the most important thing about the funding disclosure.

Xanadu's founder and chief executive Dr Christian Weedbrook said in the release that the collaboration "bridges quantum simulation and semiconductor fabrication, eliminating a critical manufacturing bottleneck" - a claim about a bottleneck being eliminated that the announcement supports with no measurement, and which should be read as company positioning rather than a demonstrated result. Dr Qi Gao, Distinguished Scientist in the Analysis Technology Laboratory at Mitsubishi Chemical, described the partnership as expanding on the companies' previously successful collaboration. Dr Masahiro Horibe, Sub-Program Director of the cross-ministerial SIP, said the participants expect the Japan-Canada collaboration to accelerate the real-world implementation of quantum technology. Those last two characterisations are the participants' own, reported here in indirect speech.

The balance sheet behind the announcement

Xanadu is a pre-commercial-scale company and its filings say so in the numbers. In results for the quarter ended June 30, 2026, released on August 5 and prepared under US GAAP in US dollars, the company reported revenue of $1.5m, down from $2.8m in the first quarter of 2026 though up from $1.1m in the same quarter of 2025. Net loss for the quarter, on a GAAP basis, was $42.1m, roughly double the $20.6m reported for the first quarter. Adjusted EBITDA loss, which the release labels a non-GAAP measure, was $21.3m against $13.9m in the prior quarter.

Spending is rising across the board. Research and development was $19.7m, which the release states was an increase of $2.4m from the first quarter; general and administrative expenses were approximately $11.1m against $9.8m in the prior quarter; capital expenditure increased to approximately $6.4m from $0.3m in the first quarter. Cash and cash equivalents stood at $312.8m as of June 30, 2026. Part of that balance was raised in the quarter through the company's at-the-market facility, which generated $67.2m from the sale of 5.5 million shares at an average net price of $12.28 - continuing issuance that dilutes existing holders, and a funding channel investors should watch alongside the burn rate. The release contains no going-concern statement and no dated runway estimate; its only runway language is a forward-looking remark that the balance sheet lets the company keep investing in engineering talent and wafer capacity, which is not a quantified management assertion and is not treated as one here.

On the technical side, the same results release reported an average edge-coupling loss of 0.085 dB per facet in chip packaging, credited to the company's internal packaging facility and collaborations with Corning and DISCO, and an algorithmic result on quantum read-only memory that the company says cuts the required Toffoli gate operations by roughly half. Both are company-reported figures at component and algorithm level; neither is a system benchmark. The release states no wavelength, chip type or measurement method for the coupling-loss figure and no baseline for the gate-count halving, and it discloses no qubit counts or error-correction thresholds.

Why the EUV angle matters

Photoresist blur is a genuine constraint on how far EUV lithography can be pushed. Every nanometre of stochastic edge roughness that resist chemistry contributes eats into the resolution budget that the scanner optics and the fab's process control are fighting for, and resist development is largely empirical. If first-principles simulation could reduce the number of formulations that have to be synthesised and exposed, that would be a real commercial lever - which is presumably why a chemicals company and two national programmes are willing to fund a second phase.

The gap between that prospect and today's demonstrated capability remains wide, and Xanadu's collaboration is one of several efforts pointing at it. For now the verifiable facts are these: phase one validated an optical model with no accuracy metric published, phase two is a stated intention, the new funding is undisclosed in size and undefined in status, and the company burning the cash reported a $42.1m GAAP net loss in its most recent quarter with $312.8m of cash and equivalents as of June 30. Nothing here constitutes investment advice.