The history of quantum computer hardware companies is well known. As technology giants Amazon and Microsoft are moving the conversation about quantum computers to the cloud, companies for quantum computer software are also emerging. One such company, Zapata, is building a business software platform for quantum computers.
Deep pocket companies are increasingly researching quantum computing, which relies on qubits to perform calculations that would be much more difficult or simply not feasible on classic computers. The quantum advantage, the turning point at which quantum computers begin to solve useful problems, is far (if it can be achieved), but its potential is enormous. Applications range from cryptography and optimization to machine learning and materials science.
We spoke to Christopher Savoie, CEO of Zapata, last month about what his company wanted to achieve. He mentioned several times that Zapata Fortune had 1
The business use case
Like the quantum computer startup IonQ, Zapata expects quantum computers to change the future of AI, especially when it comes to machine learning.
“AI itself, but more appropriate machine learning, already has a very horizontal applicability,” said Savoie. “But the place where quantum will really help, I think, is one of the main places, generative modeling. The GANs, time history data and the like. “
Savoie gave an example: “Suppose you have 100 patients with a very rare form of lung cancer. You can deeply falsify 1,000 of these MRI results. With the distributions that you can model with a quantum computer that you can’t do traditionally, you can not only recognize functions in data sets, but also reproduce and create artificial data sets that you can use to train machine learning models. A. much better and much more accurate with fewer samples. “
“So that will have a pretty big impact on all machine learning,” he continued. “The ability to determine from probability distributions that would take 10,000 years on a classic computer, even a powerful classic supercomputer, will really change the world of accuracy of our models and the time it takes to train how many samples needed to train them to the same accuracy? “
Savoie believes that the time component of ML training, and therefore the accuracy of training, will experience a change in step function as quality control becomes more efficient. In addition, some of these techniques will soon be applicable to classic systems in production, he claimed. A customer he did not want to name is currently developing a system for optimization work with machine learning and quantum-inspired algorithms. It is planned to bring them into production by the end of this year or early next year. This business system will then create “real business value,” Savoie said, before the company switched to Qubits.
Not yet in production
That’s Zapata’s pitch anyway – write algorithms for classic computing and then switch to the QC backend, “when the qubits are there to do more,” Savoie said. “So you can be forward and backward compatible with all your data analysis, all data preparation and stuff. You don’t have to repeat, you don’t have to tear it up and start over. It literally changes a few lines of code to pop out the backend. “
To be clear, Zapata has no customers who use ML algorithms on quantum computers in production. The company’s Orquestra platform is currently in beta. But its customers use it to build systems that will go into production in the near future, Savoie emphasizes. When then?
“Next year, these are likely to be quantum-inspired classic backends,” said Savoie. “In the next two to five years, I would not give you an exact time frame – the performance of these quantum computers if they stay on this path – we will replace this backend. The development of the algorithms, which are somewhat different for these backends, is currently being continued. Companies are investing in the development of these algorithms as this is imminent. Nobody will set a time frame. I wish I could for you. I can’t But in a way it doesn’t matter, does it? If it is two years, three years, five years – this disruption will occur in the medium-term business plan. “
The companies that can now invest in QC and ML because the potential is enormous.
“If it happens, it will be exponential,” said Savoie. “You add a qubit, you double the computing power and you double your options. It would be silly to wait five years and then think that you will develop a workforce, skill, and data infrastructure to use right away. That is probably very naive. It’s a when, not an if. I pinch myself because it’s really great for us that people are ready to make this investment because they see it now. It is very real. It will happen. It’s just a timing issue that no one knows. But it’s not 10 years. That’s for sure.”