DOE puts quantum on a 2028 clock and changes what counts as progress

The Department of Energy has rewritten the federal yardstick for quantum computing. The Office of Science Advisory Committee's Quantum Subcommittee released a national roadmap on Tuesday that asks not how many physical qubits a system has, but whether it can solve a real scientific problem. By 2028, the agency wants a "scientifically relevant, error-corrected quantum computing capability": a machine that does real work in chemistry, materials, fusion plasma, or high-energy physics.

The report, formally titled "Path to an Integrated Quantum Future," is the work of a subcommittee chaired by Anna Grassellino at Fermilab, and vice-chaired by Supratik Guha of Argonne National Laboratory and the University of Chicago. It is the federal government's most explicit statement that counting qubits, the basic units of quantum information, is no longer a useful measure of progress when the field is racing toward fault-tolerant machines, systems that can run long computations by continuously correcting the errors that accumulate in their quantum bits.

Phase I runs from 2026 to 2028 as a set of "Quantum Grand Challenges" built on multidisciplinary co-design. The committee wants researchers in algorithms, the control electronics that drive each qubit, and domain science to build toward shared milestones in quantum chemistry, drug discovery, catalytic material design, fusion plasma simulation, and fundamental high-energy physics. The plan targets problems the rest of science would recognize, not a specification sheet.

Phase II is the more structural move. The committee recommends a national Quantum Computing User Facility, or QCUF, run as an open-access scientific user facility. In practice, that means the federal government would buy or build machines, then let university and lab researchers apply for time the way they apply for beamlines at national synchrotrons, the large accelerator facilities that already host thousands of outside experiments a year. The committee's language is pointed: the QCUF is meant to replace the "black-box" cloud quantum-computing services that have dominated access so far, where users submit jobs to a vendor's hardware without seeing the control stack, the cryogenics, or the underlying architecture. The new model would expose all of that, and co-develop it with the user community.

Phase III, beginning around 2030, embeds quantum processors inside the country's existing leadership-class supercomputers: the Oak Ridge National Laboratory's Frontier and the Argonne National Laboratory's Aurora, two of the world's fastest publicly documented machines. Quantum becomes a co-processor inside an HPC-plus-AI stack, the supercomputing and artificial intelligence infrastructure that already runs the country's largest simulations, not a standalone moonshot that has to justify its own infrastructure.

The roadmap does not pick a hardware approach. It is deliberately modality-neutral, which means superconducting circuits, trapped ions, neutral atoms, and photonic systems, the four leading physical ways of building a qubit, all stay in the running. Co-design investment continues across hardware, algorithms, and software. The committee puts its weight behind the integration layer. Control electronics, compilers, and hybrid quantum-classical algorithms are where the next decade of federal money will do the most work, the committee writes. Holding the hardware line open lets the strongest architecture win on the science.

The report is also written to align with the Trump Administration's Quantum Genesis Initiative and the existing DOE Q Competition, the agency's prize-style program for utility-scale quantum prototypes. Together those programs give the 2028 milestone an operational home.

The committee's report is a recommendation, not an executed budget, and Congress will decide how much of the user-facility and integration work actually gets funded. The 2028 target is a science gate, not a procurement deadline. The QCUF's governance and how much time goes to outside academic users versus the labs that built the machines are open questions the report does not resolve.

The federal government has stopped asking quantum to win a horse race on qubit counts. It is asking the field to show its work, on a problem a chemist or a fusion researcher would recognize, by 2028.