In 2025, SAXON Q demonstrated a 4-qubit quantum computer running at room temperature in Germany. 2 years later, the startup opened orders for 128-qubit and 512-qubit systems that fit inside ordinary server racks

A quantum computer that sits at room temperature sounds almost wrong. For years, the field has been associated with giant refrigeration systems, elaborate laboratories and temperatures hovering just above absolute zero. Now a Leipzig startup says customers can order diamond-based machines that run without that infrastructure. But the most important number on these computers may not be the number printed on the front of the machine. SAXON Q’s new SXQ128 and SXQ512 are advertised as 128- and 512-qubit systems, yet their architecture reveals a more complicated story about what those qubits can actually do.

SAXON Q is selling a very different kind of quantum computer

SAXON Q, a spin-off from Leipzig University founded in 2021, has opened orders for two new systems: the SXQ128 and SXQ512. The company describes them as the first diamond-based nitrogen-vacancy, or NV-center, quantum computers to move beyond 10 qubits commercially. Both are designed to work at ordinary room temperature without cryogenic cooling, vacuum equipment or a specialized laboratory. SAXON Q says the SXQ128 is available now, while deliveries of the larger SXQ512 are scheduled to begin in the second quarter of 2027.

That changes the physical picture of quantum computing. Instead of placing a processor inside a dilution refrigerator operating near absolute zero, the machine can be deployed in a normal environment. SAXON Q lists an operating range of roughly 18 to 27 degrees Celsius and says its systems can run from a standard 230-volt outlet. Its newer platform is designed for rack and edge deployment, bringing quantum hardware much closer to the physical infrastructure already used for conventional computing.


There is an important caveat, though. Room temperature does not mean the machine is a quantum computer that suddenly behaves like a conventional server. The quantum processor still requires carefully engineered optical, microwave and electronic control systems. What disappears is the enormous cryogenic burden associated with several other leading quantum-computing architectures. That distinction is crucial because cooling is not merely an inconvenience. It is one of the major engineering obstacles to putting quantum processors into ordinary computing environments.

The secret is hidden inside the diamond

The heart of the system is a tiny diamond chip. SAXON Q creates nitrogen-vacancy centers inside the diamond lattice by introducing nitrogen atoms and vacancies into carefully controlled locations. A nitrogen-vacancy center consists of a nitrogen atom next to a missing carbon atom. The resulting defect has quantum-mechanical spin states that can be controlled and measured, turning the defect into a physical qubit.

This is very different from superconducting qubits, where the quantum information is encoded in electrical circuits that must be kept extremely cold. An NV center is an atomic-scale defect in a solid material. Its spin can remain useful at room temperature, which removes the need for the dilution refrigerators used by many superconducting quantum computers. SAXON Q combines the diamond with electrical wiring, microwave control and optical readout to manipulate those spins and extract their quantum states.

The architecture becomes even more interesting because the electron spin is not the only useful quantum degree of freedom. Nearby atomic nuclei can also act as qubits. SAXON Q's earlier systems used an NV center together with nuclear spins, creating a small quantum register around a single defect. The company's scaling strategy is therefore not simply to put more unrelated defects onto a chip. It is to create larger registers and then connect multiple quantum cores.

Why 128 qubits does not tell the whole story

This is where the headline numbers need some unpacking. SAXON Q says the SXQ128 contains 128 qubits and the SXQ512 contains 512. But those totals come from multicore architectures. The company says the SXQ128 provides eight fully entangled qubits per core, while the SXQ512 provides 16 fully entangled qubits per core.

That distinction matters because quantum algorithms depend heavily on how qubits interact. A system with many qubits that cannot all participate in the same quantum circuit is not equivalent to a single monolithic register containing the same number of fully connected qubits. In SAXON Q's design, several quantum processing cores work together, allowing the overall machine to perform parallel or distributed quantum computations. The total qubit count therefore describes the capacity of the complete multicore system, while the qubits per core describe the size of an individual fully entangled quantum register.

It is a little like comparing a building containing many small computer rooms with one enormous computer room. The total number of machines may be identical, but the way work can move between them is different. Quantum computing makes that distinction even more important because entanglement and coherent control are fundamental to how quantum algorithms operate. The useful question is therefore not simply, “How many qubits are there?” It is also, “How many can be controlled together, how accurately can they be operated, and how are the separate registers connected?”

Entanglement is the real engineering challenge

A qubit count by itself says very little about practical quantum performance. Researchers care about gate fidelity, coherence time, connectivity, readout accuracy and error rates. A quantum processor has to manipulate fragile quantum states without destroying the information it is trying to calculate. Every additional qubit also creates more opportunities for unwanted interactions and errors.

SAXON Q's own development history illustrates that challenge. In 2025, the company supplied four-qubit room-temperature systems to organisations including Fraunhofer IWU and the German Aerospace Center's Quantum Computing Initiative. DLR's acceptance tests required more than 95 percent quality for single-qubit gates and more than 90 percent for two-qubit gates across the four-qubit demonstrators. DLR also stressed that moving from a handful of qubits to larger connected registers remained a major scaling challenge.

That earlier four-qubit machine was not powerful enough to solve major scientific problems. DLR explicitly noted that four qubits could not deliver an economic or scientific advantage over conventional computers. Their value was different: researchers could work with real quantum hardware, study its noise and behaviour, and develop software and algorithms before larger machines became available.

The SXQ128 therefore represents a much more ambitious step, but it should not be confused with the arrival of a fault-tolerant quantum computer. More physical qubits do not automatically mean more useful quantum computing. Error correction can require large numbers of physical qubits to create a smaller number of reliable logical qubits. The quality and connectivity of the hardware remain just as important as its headline capacity.

Room temperature could change where quantum computers live

The most consequential part of SAXON Q's approach may ultimately have little to do with the 128 or 512 figure. It is the attempt to change quantum computing from laboratory equipment into infrastructure. If quantum processors can operate without cryogenic systems, they can potentially be installed closer to conventional computers, industrial machinery and sensors.

SAXON Q is already positioning the technology for areas such as robotics, automotive systems, aerospace and edge computing. The company argues that a compact quantum processor could eventually work alongside classical processors instead of being located hundreds or thousands of kilometres away in a specialist quantum-computing centre. Its roadmap extends toward systems containing thousands and eventually more than 10,000 qubits. Those future numbers remain targets, not established capabilities.

There is also a manufacturing argument behind the technology. SAXON Q says its diamond processing uses techniques derived from semiconductor manufacturing, including ion implantation and nanowiring. If those processes can be made sufficiently precise and repeatable, the company hopes to scale quantum hardware using methods closer to established chip manufacturing than the highly specialised fabrication used by many quantum platforms. That could become just as important as the room-temperature operation itself.

The bigger breakthrough may be architectural

The SXQ128 and SXQ512 are significant because they bring together several ideas that have rarely appeared in one commercial machine: diamond-based NV-center qubits, room-temperature operation, multiple quantum cores and a path toward semiconductor-style manufacturing. SAXON Q says the SXQ128 is field-upgradeable and that the architecture can be expanded by changing the diamond chip or adding cores.

But the qubit labels deserve careful reading. A 512-qubit multicore machine is not the same thing as a single 512-qubit fully entangled register. The company itself distinguishes between total system qubits and qubits per core, because those numbers describe different aspects of the hardware. For anyone comparing quantum computers, that is a useful reminder that the race is no longer simply about putting the largest number on a specification sheet.

The real test will come from the workloads. Can these room-temperature processors maintain high-fidelity gates as the number of cores grows? Can separate registers communicate efficiently enough for useful algorithms? Can error rates remain manageable as the architecture expands? And can the hardware deliver an advantage over powerful classical systems on problems that matter outside demonstrations?

Those questions are still open. What has changed is the setting in which they can now be asked. A quantum computer no longer necessarily has to live inside a cryogenic laboratory. SAXON Q's machines suggest that the next stage of the race may take place in ordinary server environments, where the hardest problem is no longer simply keeping a quantum processor cold, but making a large collection of fragile quantum registers work together as one useful machine.