Like many in the quantum computing world, particularly quantum algorithm/software developers, QC Ware is focusing near-term on classical and hybrid classical-quantum offerings. Today, QC Ware introduced Promethium, a software-as-service quantum chemistry platform intended to run on Nvidia GPU-accelerated systems. The company reports the new software is dramatically faster than other current platforms and can handle ab initio problems scaling up to 2,000 atoms. The new platform will become available on the AWS Marketplace tomorrow.

“Fundamentally, we remain a quantum computing software company,” said Christoph Siegert, SVP, product, in a briefing with HPCwire. “So, in five to 10 years, we’re hoping you will see us as the leading quantum and quantum-inspired software provider across many different fields.” In the meantime, he said, the vast majority of quantum hardware isn’t ready for production-quality use.

“We’re taking the ideas that we have for quantum computing and building them on GPUs, because we know by taking our ideas and some of our code [and porting them] on GPUs, we can [surpass] existing ab initio code available these days.”

That shift in thinking is shared by many in the quantum software start-up community. It is a recognition that scaling up practical quantum computing is perhaps proving slower than expected, and that by porting quantum-inspired approaches to classical systems, these young companies can capture revenue and deliver a form of quantum-inspired advantage now. Swiss startup Terra Quantum is another example of a company emphasizing the quantum-inspired-software-on-classical-systems approach.

In developing Promethium, QC Ware worked closely with Nvidia and highlighted that effort in its official announcement: “The core technology is an advanced quantum chemistry toolset that has been specifically designed to make use of the power of flagship Nvidia GPUs such as the H100 or A100 Tensor Core. Built from the ground up using proprietary algorithms, Promethium provides significantly more accurate predictions compared to simpler force field or coarse-grained models.”

The company claims Promethium is “10 to 100 times faster than anything else in the ab initio space.”  The new SaaS platform has been built for a wide range of chemistry and materials science applications, according to the company. Drug discovery is an early target. QC Ware also says Promethium is intended “to cater to both chemists without significant computational chemistry training as well as advanced computational chemists.”

Initially, the new platform is being offered only through AWS but other cloud providers are likely to be added. “We want the enterprise scalability that AWS or GCP or Azure offer, like scaling up to hundreds or thousands of GPUs. We also need all the other services that major cloud providers offer, such as proper SaaS capabilities and graphical interfaces,” said Siegert.

“A customer can go on AWS [and] submit one molecule or 10,000. Let’s say they drop in 10,000 into our tool. We run it through a geometry optimization and spin up a fleet of GPUs on Amazon, and then parallelize basically this job, run all of them, and then spin all the GPUs down again, and deliver [the results] back to the customer,” said Siegert. He suggested QC Ware might buy its own hardware “in the very long future” but emphasized QC Ware’s preference for the SaaS model.

QC Ware is an interesting company. Broadly, it has three businesses: its Q2B quantum business conference; its consulting and algorithm development services; and its quantum software. The company likes the SaaS model (flexibility, scale, control) for its software business and currently specializes in quantum chemistry and quantum machine learning; it is aggressively developing quantum-inspired versions in both domains to leverage existing GPU-accelerated systems.

“We hired a lot of GPU coders from the Stanford group over the last year, and it is all quantum chemistry, but it’s not all a quantum computing. That’s the nuance. People need to understand it’s important for QC Ware that we remain a quantum computing company – that is our vision, our brand – but [today] it’s really running on purely on classical systems, all running on GPUs basically. When quantum hardware is ready, we will bring it there seamlessly,” said Siegert.

It will be interesting to watch how Promethium and similar quantum-inspired offerings are received by the market. Easy-to-use interfaces, a SaaS delivery model, and alliance with cloud providers is currently the preferred approach by those offering these kinds of services. Most companies in this camp are, of course, still working with the developing hardware platforms. For example, Siegert cited QC Ware development work on IonQ’s trapped ion quantum computer.

For now, the only path to substantial revenue for the nascent quantum software community seems to be though offering quantum-inspired products running on classical systems. It’s not yet clear how big this market can become.

Stay tuned.

Header image: Nvidia H 100 GPU