Christopher Savoie says he was skeptical when former Harvard researcher Alán Aspuru-Guzik pitched Savoie last year on joining the quantum computing software venture he was planning to form.
Quantum computing technology is at least a decade away from commercial feasibility, responded Savoie, a tech developer and attorney who previously founded startups in machine learning, pharmaceuticals, and energy tech. Aspuru-Guzik—a researcher focused on the intersection of computer science and chemistry—had also been skeptical, but the technology seemed to have hit a turning point. “It’s getting close,” he told Savoie.
He insisted Savoie come talk to Aspuru-Guzik’s colleagues about the progress they—and companies such as IBM (NYSE: IBM) and Google (NASDAQ: GOOGL)—have been making. The next morning, Savoie was sitting in a Harvard building with Aspuru-Guzik and the other co-founders of what would soon become Zapata Computing. They all quickly became convinced they had a business opportunity on their hands, and began forming the initial plans for the company.
Savoie and the others finished the discussion over lunch—burritos at a restaurant in Harvard Square, he recalls in a phone interview with Xconomy. “The classic way you start a company,” he adds.
“Within two hours of the conversation, [Aspuru-Guzik] was introducing me to IBM as the CEO of Zapata Computing; Zapata Computing didn’t even exist yet,” Savoie says. “Alán doesn’t take no for an answer, and he’s very, very quick.”
Zapata’s founders (pictured above) soon licensed technology from Harvard and officially incorporated the business, Savoie says. He did indeed take on the role of chief executive. Aspuru-Guzik, who left Harvard for the University of Toronto this year, is the company’s chief scientific officer.
A few months ago, Zapata closed a $5.4 million seed funding round from Pillar Companies, FF Science, Prelude Ventures, and The Engine, a venture fund and complex technology incubator created by MIT in 2016. Zapata, which is based at The Engine’s offices in Cambridge, MA, is developing software for quantum computers, as well as quantum computing algorithms designed to solve business problems for customers in a variety of industries. Savoie says the startup has partnerships with IBM, Google, and Rigetti Computing, which are all developing quantum computing hardware.
Zapata is jumping into the sector at a critical time. Scientists have built quantum computers—machines that process information using principles of quantum mechanics—but they aren’t powerful enough yet to outperform classical computers. Nevertheless, progress is being made, and the technology is moving closer to delivering practical, real-world applications, industry leaders say. Zapata says it’s already developing algorithms for large clients in sectors such as materials science, pharmaceuticals, and finance. It’s part of a young crop of quantum-focused startups that also includes Quantum Benchmark in Canada, Q-CTRL in Australia, and Palo Alto, CA-based QC Ware, which is also trying to develop quantum computing software for large enterprise clients.
Quantum computers aim to greatly improve upon the data-crunching capabilities of today’s most formidable machines. Traditional computers process information as a series of bits—ones and zeroes. Quantum computers use quantum bits, or qubits, that can exist in multiple states simultaneously and work together in powerful ways—advantages that advocates say could enable quantum computers to perform tasks that are difficult or impossible for standard computers to achieve. (The quantum mechanics being harnessed are superposition and entanglement of tiny particles.)
One of the biggest problems: it’s difficult to control qubits. Current quantum computers are hindered by various kinds of “noise”—interference from both the surrounding environment and the qubits themselves—that destabilizes them and makes it challenging to run calculations without errors occurring, Savoie says.
But the field seems to be in a transitional stage on the path toward more robust and reliable quantum computers. John Preskill, a California Institute of Technology professor, recently declared this the era of “Noisy Intermediate-Scale Quantum (NISQ) technology.” The eventual goal is to develop quantum computers that can reliably correct for errors and noise as they compute.
During my recent conversation with Savoie, I was struck by how manual the process of programming quantum computers is today. Zapata’s engineers use tablets and software design tools for some of their work, Savoie says. But often they go old-school, using a chalkboard to write out the linear algebra behind an algorithm and determine the number of qubits they think will be required to perform the calculations—and then drawing on paper the layout of the proposed configuration of a quantum circuit. Then they send a PDF of the design to a Zapata partner that operates the quantum computing hardware, Savoie says. The two sides will go back and forth fine-tuning the design, Zapata will finalize the algorithm’s code, and then the partner—say, IBM—will run it on their hardware (either a quantum computer or a classical machine simulating one).
Zapata is developing software to automate more of the process, but those efforts are a “work in progress,” Savoie says.
Even if quantum computing technologies continue to advance, Savoie says it’ll take at least a few years to reach a point where self-service business software tools for quantum computing are feasible. He’s talking about the quantum industry’s equivalent of “big data” application programming interfaces (APIs) and open-source software programs. Such tools have enabled more business professionals—not just engineers who know how to code—to perform data analytics. Zapata has “a very conservative view” of the prospects for developing a platform that similarly “democratizes” quantum computing, Savoie says.
“This stuff is really hard,” he says. “This is not going to be something where you can get 50 people off the street and put a team together to do this anytime in the near future.” (You can click here to read more about Savoie’s thoughts on the field’s progress and the road ahead.)
The war for quantum computing talent is already on, Savoie says. Zapata has lost some potential recruits to larger tech companies, but it has also lured some away from big companies and startups, he says. He expresses confidence that Zapata has assembled one of the best teams in the nascent industry, with several PhD-holding scientists who have co-written papers on quantum technologies.
“There are only a limited number of people who have written novel algorithms for NISQ devices,” Savoie says. “They do not grow on trees.”
Savoie wouldn’t go into detail about the projects Zapata is working on for its clients, citing non-disclosure agreements. But, generally speaking, he says the coming generation of quantum computers will be able, for example, to help businesses improve their operational efficiency (think supply-chain scheduling) and enable machine learning networks to classify data better.
“Really what we’re talking about is accessing probability distributions that just don’t exist on classical computers,” Savoie says, referring to an advantage of quantum tech. It’s notable that he says the quantum hardware will be used as a “co-processor” in tandem with classical computers, at least for the foreseeable future.
Some of Zapata’s prospective clients have concluded they want to wait on the sidelines until quantum computing technology has matured more, and then they’ll spend heavily to catch up. Other companies are investing in the technology now because they think it has the potential to transform their sector—and they want to get ahead of that potential disruption, Savoie says.
Their thinking, he says, is “Let’s not have big data happen to us again and have us be the last ones in the room with it.”