A new quantum computer startup is born. Her name is Siquance and she is French. Technological choice, continuous development, planning… we tell you everything!
It’s a new player in the world of quantum computing. Launched on November 29 during a press event at CEA in Grenoble, French start-up Siquance aims to develop and commercialize a quantum computer based on silicon semiconductors. “We started from an FD-SOI transistor, the basic unit of classical computing, to make a quantum bit, the basic unit of quantum computing.explains Maud Vinet, co-founder and director of CEA’s Siquance. So, we use microelectric technology, that is, we manufacture an integrated circuit that we encapsulate and place on a motherboard measuring about ten square centimeters. »
Then head to the cryostat to place the transistors in extreme cold conditions to reveal their quantum properties. “We’ve been cooling devices and measuring at low temperatures for over 20 years, adds Maud Vinet. We have a solid understanding of what happens at low temperatures when we cool these objects. » The new start-up is entering the supercomputing market and will ultimately provide end users such as large manufacturers with increased performance. Compute time will be provided by three types of vendors that Siquance will serve: shared computing centers, cloud access providers and systems providers.
Towards perfect quantum bits
Launching a startup in 2022 is the result of many previous steps, including two important ones. The first dates from 2016: CEA managed to turn an FD-SOI transistor into a quantum bit and registered a patent the following year. The second dates from 2019: the first quantum integrated circuit. Thanks to these two innovations, “we got a research grant at the ERC 14 million euros to remove the technological obstacles to the silicon-based quantum computer, explains Maud Vinet. We therefore decided to accelerate to increase the technology’s maturity, implementing tools and methodologies, such as statistical characterization, sample design and use of simulation, to prepare its industrialization. » The team will also work on software close to the integrated circuit that will allow for its control and correction of quantum errors.
The challenge is to obtain quantum bits with perfect fidelity, that is, 1. Today Siquance uses bits with a fidelity of 0.996 or even a little more. “They’re not particularly noisy, but we use an error correction code, which is a number of physical qubits to create a logical code with fidelity of 1, explains Maud Vinet. We demonstrate that corrective codes work. Our whole bet is to implement it so that our qubits, which are very good, are perfect. » The number of qubits needed around a qubit to fix it will depend on its quality. To give you an idea: for 100,000 physical qubits, it is possible to obtain a thousand logical qubits.
A prototype in two years
A computer should see the light of day in a few years. “Our goal is to be able to build a quantum computer in 10 years” says Stéphane Siebert, director of technological research at CEA. But within two to three years, the start-up aims to offer a working prototype accessible from the cloud. “We still don’t know how many qubits it will have, but we are working on it, explains Maud Vinet. We have 16-qubit circuits and we’re going up. » And you shouldn’t imagine gigantic centers of computing machines, their footprint will be only 2 m². The advantage of technology: “When we increased the power, we changed the motherboard, explains Maud Vinet. sIn a square millimeter of circuit, we put a few hundred thousand qubits. For 200,000 qubits, or 2 square millimeters, it still fits in the cryostat. Once we have an infrastructure, we are not going to change it. »
a robust ecosystem
For the start-up, the use of the FD-SOI technology that was born at CEA is a differentiating element to advance more quickly to the quantum computer. “The semiconductor industry makes it possible to manufacture billions of transistors, knows how to drive them and implement algorithms, adds Maud Vinet. And the advantage is that we will be fables: we will not create lines that do not exist in the industry to manufacture ourand quantum computer, but the existing one to create added value. » The start-up will also work with industrial partners such as OVH (quantum cloud), Atos or even Air Liquide (cryostat).
“In the quantum computer race, depending on the choice of technologies, there are faster phases and slower phases, emphasizes Stéphane Siebert. For us it is the beginning that is slow, because we take a long time to make qubits and architectures, but then we will go much faster to produce. »
 European Research Council or European Research Council