In 2014, Nick Harris and Darius Bunandar were trying to combine optical technology with quantum computing at the Massachusetts Institute of Technology, where they were doing Ph.D. work in the same research group.
Unlike classical computing, which boils down to simple ones and zeros, quantum computing involves the creation of quantum bits, or qubits, which can be expressed as ones and zeros simultaneously. Quantum could potentially do certain kinds of computing jobs much faster than classical computers, and Google, IBM, Intel and Microsoft have all been exploring the area.
But in 2015, Harris and Bunandar began looking at fields beyond quantum computing, including AI. “Our feeling is that there are a huge number of challenges that remain to be solved” for their quantum approach, Harris said.
The pair were engineers. They didn’t know much about business. So they decided to take a class on entrepreneurship at MIT’s Sloan School of Management. That’s where they met Graham, who had worked as an analyst at Morgan Stanley. Students had to come up with business ideas and make videos about them. Graham saw Harris’ video on using light for computing and thought it sounded great.
The three decided to enroll in the MIT $100K Entrepreneurship Competition. “It was kind of like a bonus thing for the class,” Harris said. They took it seriously and put together a business plan. And they became friends.
They ended up winning the 2017 contest, beating out dozens of other teams with their pitch. They got a $100,000 prize. One of the judges happened to be Erik Nordlander, a general partner at GV, and the team shook his hand after the victory. The team used their winnings to take trips to Silicon Valley so they could meet with more investors. Meanwhile, a paper on their unconventional approach was published in the journal Nature Photonics.
In Silicon Valley the three bonded even more, eating together and staying in the same hotel rooms together. And they brought actual chips they had developed at MIT into their meetings with venture capitalists — an audience that not much earlier had not been especially receptive to the idea of pouring money into semiconductor companies.
“There was skepticism from some people about funding a chip start-up, because it’s so capital-intensive,” Harris said.
Even so, Graham said investors they met with were aware of challenges facing Moore’s Law — the notion that engineers can double the number of transistors onto a single chip every two years.
The pitch also pointed to the rise of companies that tap optical technology to deliver fast networking in data centers. In earlier years investors pumped money into these companies, such as Applied Optoelectronics.
Investors signed on, and the Lightmatter team got to work. In one year it produced not one but two early chips. The most recent one contains more than a billion transistors.
Now GV has signed on alongside Spark Capital and Matrix Partners, giving the start-up $22 million in new funding to work with. The founders are excited about working with GV when it comes to recruiting people who can work on software for its silicon.
“There is a lot of effort that goes into making this kind of device plug and play and making it look a lot like the experience of an Nvidia GPU,” Harris said. The team wants to ensure the chips work with popular AI software such as the Google-backed open-source project TensorFlow.
Initially the focus is selling chips to organizations that run big cloud-computing data centers and high-performance computing clusters. Other start-ups, like Cerebras and Graphcore, are also working on chips that could be used for AI models in such places, and big companies, including Intel, also want a piece of the pie.
But Harris and his group believe they’ll have an edge. He said that in data centers, two things are critical: throughput, or how many operations can be performed per second, and efficiency, or how many operations can be performed per second per watt of power.
“Our systems will be capable of greater than 10x existing solutions,” he said.