PsiQuantum, whose labs are pictured here, announced in April that it had raised $215 million.
In the last decade, quantum computing has moved from a science project in academia to a technology on the verge of changing businesses across industries. Naturally, startups and investor dollars have followed.
Investment in U.S. quantum computing startups has grown from $4 million and two deals in 2015 to well over $300 million and seven deals so far in 2020, according to PitchBook. A bulk of the investment is being poured into hardware, but there are also deals for software and all manner of niche companies cropping up around both fundamental underlying technologies and future applications.
We asked five investors to break down the opportunity they see in quantum computing, how startups might challenge the big tech companies in the space, and whether quantum is headed for an era of hype.
What are the different areas of investment?
11.2 Capital's Shelley Zhuang broke down the landscape of startups with a handy chart organizing her research into quantum technologies.
Chart: Shelley Zhuang/11.2 Capital
In hardware, startups like Rigetti Computing and PsiQuantum are competing against tech giants like Google and IBM. Rigetti has
raised $190 million
from blue-chip firms like Andreessen Horowitz to build a quantum computer. However, its latest round was reportedly at a "significantly" lower valuation than before,
according to TechCrunch
. Momentum isn't dead, though: PsiQuantum, which was founded by a former University of Bristol professor in 2016, announced in April that it had
raised $215 million
.
Then there's a layer of software companies that are tackling specific problems like error correction, including Australia's Q-CTRL, which raised $15 million from investors in 2019. On top of it all, there is a layer of companies trying to build applications using quantum computing, from machine learning optimization or cybersecurity companies and even companies thinking about chemistry.
Why are venture capitalists interested in quantum computing?
Quantum computing won't be just an incremental 5% or 10% bump to existing systems, scientists predict. It has the potential to be an exponential improvement that allows people to do things previously thought impossible, says Aaron VanDevender, the chief scientist and a principal investor at Founders Fund, which backed PsiQuantum, Rigetti Computing and Zapata Computing . "That means whole new applications, new industries, new markets — and that kind of new prospecting is interesting to venture capitalists."
The harder part is timing the market right to deliver venture-sized returns. "Every different group goes on their own journey to get conviction on it," says Brendan Dickinson of Canaan Ventures, which invested in Quantum Circuits Inc., or QCI. He visited more than a dozen research institutions before concluding the time is right for private companies to pick up the research from academics.
"At a high level, we know how to do it, but it needs to now be put into a commercial entity where somebody can spend all of their time iterating and perfecting specific parts of the device," he says. "Whereas an academic lab, no one's gonna write a Ph.D. dissertation because they figured out a better way to boost the signal between two different parts of the quantum computer. But that's what needs to happen to go from academia to commercial."
DCVC's Matt Ocko says it's natural for there to be haters from the science community who say it's a bad idea or too early for private companies. He saw it personally when he invested in D-Wave, one of the earliest quantum computing startups that was founded in 1999. His firm has since gone on to invest in Rigetti Computing, Horizon Quantum Computing, Q-CTRL, Atom Computing and several other stealth projects.
"The curse of, or the burden of deep tech investing, is to distinguish when that is a genuine signal of caution and prudence and maybe the cranky jerks are right and the VCs are just breathing helium — and when they're just defending the status quo," he says.
But like any venture investment, the high risk can translate to huge rewards. "If a quantum computer company is successful, and I think there will be multiple quantum computer companies that are successful, the winner looks like IBM at the height of their power and value," Ocko says. Even the runner-ups are still potentially multibillion market opportunities.
Has the pandemic changed the quantum computing market or the sector's investment potential?
Investors agree that it may be harder for quantum startups to raise money given the current level of development and the general cooling of the funding market.
"It's already a high risk, high return investment category, and that may cause investors to more carefully consider potential investments than they had in the past few years of a booming economy," Ocko says.
But the potential huge gains in computing power still makes it enticing for investors willing to make the bet, Dickinson says.
"Given COVID, the fundraising environment is more difficult for everyone — particularly those who don't yet have a clear exponential growth chart for revenue [and gross margins]," he says. "But again, quantum computing offers the opportunity to invest in a fundamentally new computation platform; platform shifts come around infrequently and are hugely valuable when they do."
If anything, investors also view the pandemic as validation of how powerful computing would be useful in a moment like this.
"Generally we don't think COVID impacts quantum computing investing very much," Founders Fund's VanDevender says. "Everything we think about the potential benefits of the technology are at least as true today as they were before, and actually in the race to develop antivirals and vaccines against the disease, a scalable fault tolerant quantum computer would really come in handy right now."
Can these investments really pay off anytime soon?
The question with quantum computing companies is whether they can deliver an investment return on the timeline a venture capitalist needs.
"This is not a two-year-and-IPO consumer software company, but I think the seven- to 10-year range is not crazy here," says Bill Coughran, an investor for Sequoia who also invested in QCI.
VanDevender points to the pharmaceutical industry as one model: "It might take 15 years to develop a drug, but there are plenty of exit ramps between the target discovery and the fully realized market value or peak sales of the drug 15 years later that that can still provide great returns for the company." Still, unlike pharma, there's no proven model, and so far, the big companies like Google or IBM have not been acquiring startups, one of the most typical exit paths in medicine.
What's the better investment: hardware or software?
Venture capitalists love to invest in software since it's a more capital-efficient business with better margins. But when it comes to quantum, that sentiment is reversed.
"I think it's crazy to be making software investments in the world of quantum computing," says Canaan's Dickinson. The hardware isn't there yet, and if quantum computing machines never get developed, they could be worth nothing, he warns. Plus, "the hardware companies aren't stupid," he says. They will probably build a lot of their own software, too.
"Investing in Slack in the '80s wouldn't be a very great investment," he says. "You don't have any of the underlying infrastructure to power that software layer."
Coughran sees the appeal of software to venture capitalists but echos Dickinson about the risks. "I think the challenge for them is some of the abstractions and the way that actual hardware work hasn't been completely sorted out, so there's some risk that they'll go write a lot of software that doesn't necessarily map onto the hardware that gets developed," he says.
Ocko's firm DCVC has invested in both hardware and software, but he thinks the hardware side might have a more likely path to outsized returns if the market plays out as expected and the winner becomes like IBM in the 1960s.
As for the firms looking even deeper into applications, 11.2 Capital's Zhuang thinks it's a little premature. "Based on what we have looked at and startups we've talked to in this field, it still seems a little limited by the capacity and capability of the software and hardware," she says.
Are we about to enter an era of AI-style hype?
Though there's considerable hype around quantum computing, it's much harder to claim to be a quantum company than it is to sell simpler technologies like AI.
"The con men at work in that market view AI as just another billy club to hit unsuspecting Fortune 2000 CEOs over the head with to sell more of the same crap," Ocko says. Because quantum computing is dealing with much more complex science and very specialized hardware, it's going to be harder to pass it off. "There's a robust scientific community who could totally destroy them," he says.
It's also simply not as popular as AI startups have become today. "Every third or fourth company I meet with as a potential new prospect for Sequoia to invest in says they're doing something with machine learning or AI. That's not the case with quantum computing; I think it's pretty rare," Coughran says.
Much of the hype doesn't come from the startup community at all, but from tech giants bragging about their research and showing off new devices, he says. "IBM, Microsoft, Amazon, Google are all publishing things to kind of assert that they're doing something interesting, and they're also trying to use it as a way to talk about adding quantum computing to their public cloud offerings and so forth," Coughran says.
What would be the best way to sniff out a good quantum startup from one riding the gravy train?
Right now, most quantum startups are in what's called the NISQ era , short for noisy intermediate-scale quantum computing, says VanDevender. What that means is that today's quantum computing machines are still small and error-prone. (That doesn't mean they're not useful. Some scientists argue that the noisiness is a feature , rather than a bug.)
The grandest visions of what quantum computing can do will require developing fault-tolerant machines that don't exist today. To vet a company, VanDevender always tries to understand the path it's developed to reach a point where it can apply enough fault-tolerance.
Dickinson agrees. "The fundamental challenge in building a scalable machine is this issue of error correction, which is not nearly as fun or sexy to talk about, as how many qubits you have," he says. "But at the end of the day, that's the thing that's important to solve if you want to build a machine that solves real-world problems."
So far, Zhuang has held off from investing because she's typically looking for a company to be starting commercialization in a year or two from investment. She's met a lot of companies that have plans that would fit on her timeline, but there's another challenge: They aren't solving problems that quantum can uniquely do better. (There's similar hype in blockchain, she points out.)
"There's often existing approaches that are taken by the practitioner in that field and it works just fine," she says. "There's just no quantum advantage."