Scientists have announced that they have developed a breakthrough 3D interconnect solution that can increase the number of quantum bits (qubits) that a quantum computing chip can support by 100 times.
A typical quantum computing processor (QPU) is built with two-dimensional horizontal wiring, similar to the central processing unit (CPU) in a classical device. However, this traditional wiring limits the number of qubits that scientists can cram into a given processor. For example, the chips currently available from Google and IBM contain approximately 105 and 120 qubits, respectively.
you may like
This adds a single QPU that can support 10,000 qubits simultaneously on a smaller chip. This is 100 times more advanced than the current state-of-the-art technology in superconducting quantum computers. According to QuantWare, this is the first time such a number of qubits has been achieved in a single quantum processor.
“People have been hearing for years about the potential of quantum computing to transform fields from chemistry to materials to energy, but the industry is stuck with 100-qubit QPUs, forcing the field to theorize about an interesting but distant technology,” QuantWare CEO Matt Reilersdam said in a statement. “QuantWare’s VIO finally removes this scaling barrier and paves the way for economically viable quantum computers. VIO-40K gives the entire ecosystem access to the most powerful hyperscale quantum processor architecture to date.”
Vertical integration meets quantum democratization
QuantWare representatives said they plan to begin shipping the first VIO-40K units in 2028. To support this goal, the company says it is building an industrial-scale QPU manufacturing plant in Delft, Netherlands, scheduled to open in 2026. It will be “one of the world’s largest quantum fabs” and the first dedicated fab for quantum open architecture (QOA) devices.
To put this timeline in perspective, IBM’s current quantum computing development roadmap puts the arrival of 2,000 qubit QPUs in 2033 or later, with no deadline set for chips capable of supporting 10,000 qubits.
For most companies working on superconducting quantum computers, the bottleneck lies in how quantum processors are built. Manufacturers can only cram so many wires onto a single wafer, so physicists have to chain together multiple processors. Although the connections between the qubits on each chip are high-fidelity, the connections between the chips themselves are often low-fidelity, creating a data transmission bottleneck.
QuantWare’s VIO series uses vertical interconnects, which it says allows it to pack as many as 10,000 qubits on a chip smaller than today’s 100-qubit wafer-style chips. This is achieved through the use of “chiplet” technology, where individually manufactured modules are stitched together to form a complete chip.
Rather than relying on low-fidelity chip-to-chip connections like current quantum processors, the chiplets are manufactured individually and then encapsulated together to create a system-on-a-chip environment that can function as a single QPU.
you may like
quantum brain in a box
Although QuantWare’s schedule is relatively ambitious compared to its peers, representatives say one factor working in its favor is its adoption of QOA.
Unlike Google and IBM, QuantWare is not developing an end-to-end quantum computing solution. The company’s QPUs are built to work with components from other companies, including Qblox controllers and Nvidia software.
This means that the VIO-40K is essentially plug-and-play with Nvidia NVQLINK, an architecture designed to allow QPUs to connect to GPUs in hybrid classical quantum systems, allowing it to connect with existing supercomputers. This also enables connectivity with Nvidia CUDA, a parallel computing platform and programming model, allowing developers to seamlessly integrate entire quantum workloads into hybrid systems.
Ultimately, this puts QuantWare in the position of potentially acting as an Intel-like hardware provider to quantum computing systems, working with other quantum computing entities in the process.
Source link
