HQS Quantum Simulations: How to survive a Quantum winter
How does a team of quantum researchers make the jump from academia to industry? In the case of HQS and its CEO and co-founder, Michael Marthaler, it began with a round of cold calls - ringing up companies and asking, if, on off chance, they’d be interested in doing some quantum simulation.
“Around 2012 it became clear to me, after a couple of breakthroughs had happened in the field, that this whole quantum computing business could really work out,” Marthaler says, with a degree of understatement. “I started to think it would be really cool to form a company and take this technology the last mile and make it useful. At the same time, I had several students working for me, and they became my co-founders. Then, around the end of 2016 and the beginning of 2017, we started to actually call various companies and ask them, ‘If we had a quantum computer that could simulate materials, would you find that interesting?".
The answer, once Marthaler and his team had explained what it was they were offering, was a steady trickle of yeses, culminating in partnerships with German chemicals company BASF and engineering and technology firm Bosch. Though its software is currently built for classical computers (like all other quantum software developers, HQS still has to wait for a quantum hardware that can operate reliably at scale) Marthaler sees material design simulations on quantum computers as quantum’s most promising discipline - with the earliest effects being felt in the fields of catalysis, energy storage and medicine.
Medicine is a particularly interesting area for quantum simulation. Taking a drug from a candidate molecule through to approval by medicines agencies is notoriously time-consuming and expensive. The pharmaceutical industry has a chequered reputation of fudging or suppressing studies, as well as charging high prices for drugs for which it holds a patent. The latter practice is what, a drug company will argue, makes the risks and R&D costs worth the gamble. Quantum simulations could upend this model.
“There is actually a super-concrete number that exists in the literature [for developing a drug today],” says Marthaler. “It’s about a billion dollars. Roughly half of that is spent in the pre-clinical stage and roughly half of that goes into the clinical stage. But the key point here is that if you improve your choice of possible molecule that you want to use as a drug at the pre-clinical stage, then you have a huge accumulation of value down the road. Basically you save a lot of money down the road in the clinical stage in your drug trials. Spending money in the early stages when you make your first choice about what type of molecule you want to consider further can be really valuable.”
The majority of potential drugs tested at the pre-clinical stage turn out to be duds - either ineffective or with the potential for complications, however small, somewhere down the line. What HQS offers now - on classical computers - and will do in the future on quantum hardware, is a reduction in the time a pharmaceutical company needs to run simulations by a factor of ten, or even one hundred. HQS would act as a sort of quantum sieve: discarding the unpromising candidate molecules and retaining the most viable, so R&D funds can be spent where they are most likely to do good.
“You would probably say, ‘I have these 10,000 candidate molecules, I need to run 1000 lab tests on each,” says Marthaler. "'But I would really rather have simulation software that can reduce that from 1000 lab tests to 10, or to 100.’ Basically, using data science you make your first choice of a couple of molecules and then very quickly you go and test those explicitly using small-scale experiments... But these experiments are very expensive. What we want to do is put a level between choosing your molecule and the experiment where you do really microscopic simulations of the effect of the drug, where you try to achieve really high precision by using the laws of quantum mechanics which, in principle, give you all of the interactions and processes that can happen on the microscopic level.”
This level of control, simulation and design, Marthaler is quick to stress, is still a ways off. Like other quantum software companies, HQS is still waiting for the hardware to catch up to the software. Without reliable QCs, HQS currently offers simulation packages for classical computers. When stable QCs become available, that’s when the company will switch focus, move its software to the new hardware and reap the rewards.
“When we talk about the readiness of our software, we have to distinguish between two things that we do,” Marthaler says. “One is to develop the software for quantum computers, the other thing is individual simulation software packages that run on classical computers. And a key point where this meets is that in this kind of individualised simulation software we make what are called multi-scaled simulations, where you start with a rather large-scale system with many thousands of atoms and then cut it down into smaller pieces. Then at some point you identify the part that you would like to calculate again with high precision using, for example, quantum mechanical calculation. Currently at this stage we still use powerful classical computers that do the best they can - but in the future we want to put in the quantum computer. So, we’ve created these software packages for customers, and they create value right now. But once we can put in a quantum computer at this stage they will create even more value in the future.”
This flexibility - designing software for classical computers now while planning for quantum down the line - is important for avoiding any potential collapse of the quantum hype bubble. Speaking to The Economist, Marthaler warned that all the promises and expectations of quantum computing left unchecked could lead to quantum’s own ‘AI Winter’ - a prolonged and disastrous collapse in interest and funding when a technology (like AI in the late 1980’s) doesn’t deliver on investor expectations and media hype.
“People are still fascinated because it is a really important long-term technology; there are a lot of opportunities there. So people are still fascinated. To some extent we always need this kind of hype around technologies to get them off the ground. But look at the current work in the last couple of years. There are really strong questions about the applications of quantum computers that are pretty fundamental… Quantum computers will make specific applications much faster, but certainly not all of them.
“Every technology goes through a hype cycle. You can’t really avoid it. If you look at the Wikipedia page for the so-called AI Winter, the first AI Winter was something like 20 years ago, with an insane funding cutback over a long period of time. I don’t expect a Quantum Winter to be that significant, but a certain kind of cutback of expectations is going to happen at some point. It could take time: maybe five years, maybe another ten. And who knows how big the hype around quantum is going to be then - maybe the hype in ten years will be way, way, way bigger than it is even now - and then a correction is necessary simply to get back to productive work.”
For HQS’ part, Marthaler is quick to put his own goal for the company - running successful simulations on a quantum computer within 12 months – into realistic perspective. The modelling that the company does is filling a need that Marthaler describes as “super-niche” and, even if the company is successful, “it won’t change the fact that we are on this wave of hype that will cool down at some point.” 2020 for HQS will still be a year of classical computing simulations, beginning with the launch of a materials simulation platform, mostly for academics, in March. That will begin as an online tool before being packaged as an API for customers to use anywhere. The company will also continue to develop simulation tools and expand its user base.
“But in the next twelve months, we’d really like to run one of these material problem simulations using an actual quantum processor with a large number of qubits - to do something really at the edge of what current, classical computers can do.”
For more information, email Amit Das directly on firstname.lastname@example.org
To visit the HQS Quantum Simulations website: https://quantumsimulations.de/