John Martinis, the Google computer scientist who helped the company achieve 'quantum supremacy' last year, has resigned: 'There was a lot of tension going on' (GOOG, GOOGL)
John Martinis joined Google's quantum computing division in 2014, where he led efforts to achieve 'quantum supremacy.' Last year, the team announced they had succeeded.
But Martinis was also pushed from his leadership role into an advisory position last spring, something he says led to tensions within the team. He resigned from Google earlier this month.
Hartmut Neven, who leads Google's Quantum Artificial Intelligence Lab, will continue to oversee the entire division.
"It's hard for me, I'm very sad." Martinis told Business Insider, but said he was glad the team were able to prove to Google's executive team that quantum computing wasn't a "pipe dream."
Visit Business Insider's homepage for more stories.
In October last year, Google made waves by announcing a major breakthrough: it had achieved 'quantum supremacy.' Or put more simply, it had made a processor that could perform a calculation in 200 seconds that would have taken the world's most powerful "classical" computer 10,000 years.
John Martinis led the hardware division inside Google's secretive quantum project, which was founded in 2012 and created the 'Sycamore' processor. But earlier this month, Martinis resigned from the company, a departure that has caused much head-scratching among the scientific community.
Martinis told Business Insider that he'd been reassigned to a leadership role last spring, with other members of the group taking the reins, and Martinis had been unhappy since.
"At the time I was really worried about it," he said, describing last year's reshuffle. "And then over the nine months I tried it out and it kind of wasn't working for me, and I told management it wasn't working for me. But the young group had done very well with quantum supremacy, and Hartmut [Neven] really wanted them to lead the team, and that didn't really work for me, and in the end I felt I had to quit."
Martinis said that three members of the team will be leading the hardware group going forward, though he wouldn't say who. Hartmut Neven – Martinis' now ex-boss – who leads Google's Quantum Artificial Intelligence Lab and has been a key player in various other Google projects, such as advancing its visual search technologies, will continue to oversee the whole team.
It was Neven who in 2012 pitched the idea of a quantum computing project to then-CEO Larry Page. Page agreed to build a department to research and advance quantum computing, and Martinis joined the team in 2014. The division's work has evolved through various stages, but achieving quantum supremacy was by far its crowning achievement.
"One of the reasons to do quantum supremacy was to show Google executives that this quantum computing was real and not just some physics pipe dream," said Martinis, who says management continued to give the team the resources and time they needed to make it happen. "Google just kept supporting us," he said. Martinis, who is still a professor at UC-Santa Barbara, also brought several of his students over to join Google's quantum supergroup.
"What I appreciated was that Google wasn't trying to get us to make a lot of money by selling quantum computer time or anything like that," he said. "We were doing it to develop the industry and develop the ecosystem, but they weren't pressuring us to do things in order to make a few million dollars."
Martinis described himself as the "chief realist" and Neven as the "chief optimist" of the group. "I always thought it was a good combination because they're almost two sides of how you want to think of things, and if you get together and talk about it you can come to good solutions," he said.
"And that happened for a long time, but I would say over the past year or year and a half with these other problems and other things it wasn't working out so well. And in the end, Hartman makes the decision, he's my boss."
According to Martinis, younger members of the team wanted to take leadership of the project. "That was the top criteria and we couldn't figure out a way to fit me into this in a way that everyone else was comfortable with and I was comfortable with. It just happens."
Martinis said that he would have remained at Google if he could have retained a leadership role, but discussions were management weren't fruitful – resulting in the decision to walk away this year. Google has a history of holding onto prized brains within the company to keep them away from rivals, sometimes paying them to effectively do very little. Google held discussions with Martinis to keep him at the company, but without the ability to steer the quantum ship he felt as though it was no longer a good fit.
"To build a quantum computer is really hard. And while I'm sure it's going to be fine for the next few years, when you think about on the 5- or 10-year scale, there are significant scientific and engineering challenges," said Martinis. "I felt that I needed to be in a leadership role and not in an advisory role to meet those challenges. And again, Hartman had a different opinion on how to do things, so I tried it out for nine months and then I made a decision to leave.
Although Larry Page signed off on Google's quantum computing group, it was Alphabet's current CEO Sundar Pichai who put his stamp on the project when the team announced its breakthrough last year. As far as Google executives went at the time, the low-key Sundar, who then only had the singular title of CEO of Google, was the face of Sycamore.
A little over a month later, he was appointed CEO of Alphabet, and Larry Page and Sergey Brin stepped down.
"Sundar was involved and very excited," said Martinis. "He was supporting it, and not just supporting it, but supporting it for the right reasons."
But here's something you might not know: Google has a second team working on quantum computing inside the even-more-secretive X lab – Google's "moonshot" division – from which some project have emerged (Wing) and other probably never will (space elevators).
Martinis says that while Hartman does collaborate with the X team, Martinis' own work didn't overlap. "Some of the theory directions they were going in, talking about quantum gravity and the like, and using a quantum computer to look at those problems... let's say they're a little more esoteric than what I'm trying to do to get this thing to work, so I let other people deal with that," he said.
The team's big breakthrough of 'supremacy,' which was detailed in the journal Nature last year, proved that superconducting circuits could be used to build qubits, rather than the bits that today's computers use to process information.
Unlike bits, which can only exist in either on or off states, qubits can represent both states simultaneously, allowing a quantum computer to carry out certain tasks millions of times faster than a regular computer.
But the breakthrough moment was not without controversy. IBM, a quantum computing rival of Google's, fired back at the announcement in a lengthy blog post where it disputed Google's core claim of "quantum supremacy." Though it conceded Google had hit an important milestone, it took umbrage with the idea that Sycamore was capable of doing something that classical computers could not.
"We argue that an ideal simulation of the same task can be performed on a classical system in 2.5 days and with far greater fidelity," wrote IBM. Essentially, it said that Google had underestimated IBM's own supercomputer, known as Summit.
Inside Google, the team were scratching their heads over IBM's response. "IBM proposed an algorithm and in computer science it's fine to propose something, but if you want to claim something you have to run the program," said Martinis. "So it was kind of weird that we did an experiment, but they could do a white paper and make this big statement of whatever they made about it."
The spat naturally attracted a lot of media attention. "In the end that controversy was good because we got more press," said Martinis.
Martinis will continue to work at the UC Santa Barbara, and although a lot of what he built under Google's roof belongs to the company, he says he plans to continue his work in this field.
"I knew there was some technology that needs to be improved, and I'm reading books and thinking about that, and inventing some things now that I think will be helpful for the field," he told Business Insider.
But Martinis said he was disappointed he couldn't continue to pursue quantum computing for Google, where he worked for almost six years.
"It's hard for me, I'm very sad. But I'm happy for people because they get to have their own career, run their own project, and do their own thing now, and that's good for them professionally," he said.
"In some sense leaving Google was something I thought could help everyone, because there was a lot of tension going on, and this way it's better for me and better for the people at Google that they get to do what they want. Maybe it's better for everyone."SEE ALSO: Google is cutting summer-intern pay by as much as 50% in countries like Canada by switching to local currencies
Join the conversation about this story » NOW WATCH: Inside London during COVID-19 lockdown