Qruise wants to build AI to automate quantum device development | TechCrunch

3 min read
DynamoFL raises $15.1M to help enterprises adopt 'compliant' LLMs

Can software work alongside human physicists and engineers in R&D labs to develop new quantum computers? That’s the question Qruise, a startup in TechCrunch’s Battlefield 200 competition at Disrupt 2023, is attempting to answer.

Qruise, a spinoff of Forschungszentrum Jülich, a national research institute based in Germany, is building what its founders describe as a “machine learning physicist” — AI-powered software that can do much of the work currently handled by junior physicists in quantum laboratories.

“At the heart of the difficulty developing quantum hardware is the lack of predictive power of simulations of the device at the design stage,” Nicu Becherescu, the business development manager at Qruise, told TechCrunch in an email interview. “When dealing with bleeding-edge R&D at the interface of physics and engineering such as quantum computing, there’s a gap between the performance predicted by the simulations at the design stage and the actual performance. This is due to the complicated intertwining of multiple physical phenomena, and significant repercussions of even minute effects, such environmental noises and tiny imperfections in manufacturing.”

Qruise tries to close this gap by combining simulation technology with the results of experiments performed with quantum hardware. Scientists using Qurise’s platform can select a set of experimental data, then have Qruise create a “digital twin” — i.e. a copy of the experiment in simulation — to slowly optimize the “parameters” and controls of the hardware by performing the experiment virtually over and over.

The goal is to create a system that can essentially read a scientific paper, recreate the results in simulation and, if relevant, implement them in a real-world experiment with a quantum device, Becherescu said.

“The result of a collaboration between the human domain-experts in the lab and the Qruise platform is a highly detailed predictive simulation of the system — a simulation from which we can extract what we must improve in the next iteration of the device under development to improve overall performance,” Becherescu said.


Image Credits: Qruise

That might sound ambitious. But Qruise, which was founded in late 2021, already has a grant under its belt from the Helmholtz Validation Fund and a go-to-market plan to tackle device design challenges pertaining to quantum control, or manipulating the properties of quantum devices to accomplish specific tasks.

Qruise is pre-revenue, with a team of about 20 people and beta tests ongoing at unnamed labs. But despite having raised no funds, Qruise has three years of runway at its current burn rate of about $2.25 million per year, Becherescu claims.

It helps that quantum interest is high.

According to 2022 survey commissioned by quantum hardware startup Classiq, 83% of companies have already invested in some form of quantum research or technology. Ninety-four percent of respondents said that they see quantum being beneficial to their industry.

“As Qruise delves deeper and deeper into the machine learning physicist, and the software is becoming more and more capable, we are quickly eclipsing the capabilities of both internal solutions and our direct competitors,” Becherescu said. “Qruise’s platform can put a thousand junior physicists in a lab, working 24/7, assisting the existing team to achieve better insights into their novel devices, and make progress faster.”

Source link