Zapata Computing CEO on A.I., Automation & Breakthroughs in Quantum

Quantum computing is one of the most intriguing—and complex—areas of technology right now.

For years, it seemed like the field wouldn’t deliver practical applications any time soon. But that thinking has changed, and efforts at big companies, startups, and university research labs seem to be picking up steam—with use cases in finance, artificial intelligence, and other areas, though the exact timeline isn’t clear yet.

I recently caught up with Christopher Savoie (pictured above), the CEO of one of those quantum computing startups, Zapata Computing, based in Cambridge, MA. You can click here to read about the main takeaways from our conversation.

We also dove deeper on the technology and Zapata’s business strategy. Below are some additional highlights from our discussion, in which Savoie touches on things like qubits (the building blocks of quantum computing), probability distributions (mathematical functions that can help provide an advantage over classical computers), and the quantum talent pool.

Xconomy: How close are we to having quantum computers that outperform classical computers?

Christopher Savoie: This generation of NISQ [Noisy Intermediate-Scale Quantum] devices is very close to achieving a major breakthrough in the field, known as quantum supremacy—or, in layman terms, the demonstrated ability that quantum computers can outperform classical computers in some specific tasks.

The coming generation of small-scale NISQ computers … will comprise around 100-250 qubits, and—used as co-processors of classical computers—will have compelling, practical use cases. As for the timeline, to give you an example, just recently Rigetti Computing announced their intention to deploy a 128 qubit-chip device within the next year. [Savoie says current quantum computers usually have around 50 qubits.—Eds.]

The key to understanding this quantum advantage is understanding that in machine learning or optimization use cases, quantum hardware (again, as a co-processor) would allow one to sample data over probability distributions that are not tractable on classical hardware. This translates to better classification of data, better generative models, or better scheduling in supply chain problems. Unfortunately, I can’t get very much more specific than that using specific use cases without violating [Zapata’s non-disclosure agreements with clients]. The optimizations are very much use case-specific and proprietary because they are problem- and data-specific.

X: So, it sounds like you’re working on enhancing artificial intelligence technologies using quantum computing?

CS: Yes, we are. This is an exciting area. Quantum A.I. will allow us in machine learning applications to sample from probability distributions that are not tractable classically.

X: What will it take to develop self-service software tools that make it easier to design and run quantum computing calculations?

CS: We are already developing our “Z-machine” platform, which is intended to automate much of the work that is now manual. But stabilization of the hardware standards and platforms would be necessary to make this tenable for delivery as a self-service enterprise software-level service.

The other gating item is people. Without people who know very well the usage and limitations of quantum circuits, giving someone a software toolset to do quantum state preparation, quantum gates, and measurements would be pretty useless. It would be like giving Cloudera/Hortonworks and Tableau to a bunch of people who do not understand big data statistics, and expecting a fantastic analytic result. Therefore, there has to be an effort in educating a larger scientific community (like engineers, computer scientists, software engineers, etc.) in the field [of] quantum technologies.

X: If so much of Zapata’s work is manual and customized to each client, how do you scale this business in the next couple of years?

CS: In the process of doing these client-specific engagements, we are building out our Z-machine infrastructure, which will allow us to automate and scale over the medium term, as the platforms mature. The advantage to this approach is that we are developing our tools and enterprise integrations based on actual use cases with real Fortune 100 partners.

Jeff Engel is a senior editor at Xconomy. Email: jengel@xconomy.com Follow @JeffEngelXcon

Trending on Xconomy