A non binding Atom Computing and Phasecraft agreement on batteries and solar cells is the latest in a 2026 wave of hardware software tie ups, a test of whether tailoring code to one machine can shorten the path to useful quantum computing.
The quantum computing industry's center of gravity has shifted from the chips to the software stack. In 2026, almost every serious hardware maker has announced some form of partnership with an algorithms specialist, a company whose core product is software written to run on a quantum machine, and the June 15 Atom Computing and Phasecraft memorandum of understanding is the most recent example of a pattern that now spans multiple pairings this year.
The two companies are tying algorithms to one specific machine. Phasecraft, a London-based algorithms shop, will adapt its software to Atom Computing's neutral-atom hardware, a quantum computing approach that uses arrays of individual trapped atoms as its computational bits, based in Boulder, Colorado. The target, per the press release announcing the deal, is materials development for batteries and photovoltaics, a class of problems that classical supercomputers struggle to simulate accurately because the underlying physics of electrons, atoms, and molecules is hard to model with conventional bits.
The deal is a memorandum of understanding, a non-binding agreement to explore working together, with no legal commitment to deliver a product, a result, or a timeline. The two CEOs framed the partnership in the standard press-release language of strategic collaboration. What makes the announcement worth parsing is the larger pattern it joins.
The strategic question underneath all of these pairings is whether tailoring algorithms to a specific hardware architecture actually accelerates the path to a useful quantum computer, or whether the industry has settled on co-design as its new way to talk about work that was already happening. The optimistic reading is that different quantum hardware approaches, including superconducting circuits, trapped ions, neutral atoms, and photonic systems, have such different strengths and weaknesses that a generic algorithm cannot extract the most from any of them. The skeptical reading is that the partnerships are real but the wins are incremental, and the announcements are running ahead of the science.
The companies themselves are not neutral observers. Atom Computing has spent the past year making a series of large claims: a toric-code quantum error correction demonstration, a benchmark test of how well a quantum computer can protect information from noise; a $100 million Letter of Intent with the U.S. Department of Commerce; what it calls the first commercial quantum computer with logical qubits, groupings of physical qubits that act as a single, more reliable unit of computation; and a seat in Stage B of DARPA's Quantum Benchmarking program, a U.S. defense-agency effort to independently test which quantum platforms are actually approaching utility-scale, the industry's term for a quantum computer that solves a useful, real-world problem faster or cheaper than a classical supercomputer. None of those are peer-reviewed results, and a corporate press release is not the same as a published demonstration.
The honest reading of the Atom-Phasecraft MOU is that it is a signal, not an outcome. It tells you that Atom Computing has an algorithms partner willing to invest engineering time on its hardware, which is useful context for anyone tracking who might credibly claim a useful quantum advantage in materials science first. It does not tell you that any specific battery or solar cell is about to get better, or that any timeline has moved.
What to watch next is whether the 2026 wave of co-design partnerships produces peer-reviewed benchmarks in 2026 or 2027, whether the industry's utility-scale claims survive contact with the same kind of independent testing that DARPA's program is designed to apply, and whether at least one of these pairings delivers a result on a real materials problem that a classical supercomputer cannot match. The MOU is the easy part. The hard part is the one no press release announces.