Microsoft Patents a Method to Run Quantum Computers Using Fewer Supporting Components
Quantum computers are notoriously wasteful, burning through extra helper components just to keep calculations stable. Microsoft thinks it has found a way to do more with much less.
What Microsoft's two-ancilla-qubit method actually does
You're building a quantum computer and every component you add makes the whole machine harder to control and more likely to fail. Extra helper units, called ancilla qubits, are especially costly because they pile up fast with complex math tasks.
Microsoft's patent describes a method that cuts the number of those helper qubits down to just two, no matter how complicated the calculation. It does this by breaking a big math problem into smaller pieces, running those pieces through a specific sequence of steps on the quantum circuit, and then recombining the results in a way that avoids needing extra hardware overhead.
For anyone using quantum computing to model molecules, run financial simulations, or crack optimization problems, fewer helper components means machines that are smaller, more practical, and closer to working reliably outside a research lab.
performing a Hamiltonian block encoding with unitary evolution on a first multiplicand and a second multiplicand resulting in a first intermediate operator and a second intermediate operator, respectively; …
Translation: The system starts by translating math problems into quantum states.
How the circuit encodes and combines quantum operators
The patent describes a circuit-level procedure for computing the commutator of two quantum operators (a commutator is a way of measuring how two mathematical operations interact with each other, a calculation that shows up constantly in quantum chemistry and physics simulations).
The method works in several stages:
- Hamiltonian block encoding with unitary evolution: each of the two inputs (called multiplicands) is encoded into the quantum circuit in a reversible way, producing two intermediate operators.
- Gate sequences: specific quantum gate operations are applied to each intermediate operator to prepare them for combination.
- Commutator product formula: the two modified operators are merged using a mathematical recipe that captures how they interact.
- Conjugate transpose gates: a final set of gates (which are the mathematical mirror image of earlier gates) is applied to the combined result to clean up the output.
The key claim is that only two ancilla qubits are needed throughout the entire procedure. Ancilla qubits are scratch-pad qubits the computer uses to manage intermediate results, and typical approaches to this kind of calculation require many more. Reducing that number directly reduces the physical complexity of the quantum hardware required.
The another second gates include a conjugate transpose of the another first gates. Two ancilla qubits are utilized on the quantum circuit.
Translation: By using a mirrored pair of gates and just two helper qubits, the machine runs efficiently.
What fewer qubits means for real quantum hardware
Quantum computers today are limited by how many stable, high-quality qubits can be packed into one device. Every extra qubit you need for bookkeeping is one fewer you have for doing useful work, and errors compound with scale. A method that keeps ancilla overhead at a fixed, tiny number (just two) could make certain classes of quantum algorithms runnable on hardware that currently isn't big enough to support them.
Microsoft's interest in fault-tolerant quantum computing means this kind of efficiency work is central to its roadmap. For researchers and eventually for businesses running quantum workloads, a lower qubit count per calculation could be the difference between a proof-of-concept and something that actually runs.
Microsoft's fifth patent in the quantum computing work we've tracked since June builds on earlier applications like one on atom traffic planning and one on hybrid physics solvers.
Almost nobody using a computer today would notice this patent, because the machines it applies to don't yet exist in a form you can buy or rent at useful scale. The payoff is real but deferred.
The specific problem it solves is a hardware constraint that kills quantum calculations before they finish: running certain math operations requires borrowing extra processing units, and the more you borrow, the more the whole system breaks down. This patent cuts that borrowing from many units to two, which is a concrete, measurable reduction rather than a vague promise of improvement.
When quantum hardware does mature, the people building software on top of it will care deeply about work like this, because smaller resource requirements translate directly into calculations that actually complete versus ones that collapse mid-run. The user never sees the mechanism, but they absolutely feel the difference between a result and a failure.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings
12 drawing sheets from US 2026/0278432 A1 · click any drawing to enlarge
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →