“The advantage that silicon photonics can bring is a small form factor solution, which can result in a compact size of the device in the car in the end,” says Kiyoul Yang, a postdoctoral researcher at Stanford University who focuses on hardware. photonic. Many companies today use a lidar system based on rotating mirrors, says Yang, which requires the manufacture of discrete and expensive components. “If everything can be integrated into a chip in a small form factor, then everything can be produced at a low cost,” he says.
Again, Mobileye is not the only company operating on FMCW or on the larger lidar faces. But it has a distinct advantage that Intel already has a silicon photonics plant in operation in New Mexico. “The ability to build an FMCW lidar requires know-how, but even if you don’t have the special products to create lidar on a chip, it becomes too expensive. It’s getting difficult, “says Shashua. He expects the cost of each SoC lidar to be hundreds of dollars each, orders of magnitude cheaper than today’s systems.
Even if Mobileye’s production roadmap remains constant, an uncertain regulatory outlook could slow the schedule. However, it is making progress in the short term as well, announcing today at CES that it will expand its autonomous vehicle testing in Detroit, Paris, Tokyo and Shanghai in 2020. (Locations are strategic; automatic driving for.) And has used millions of cars with Mobileye on board to provide a map of nearly 1 billion kilometers of world roads to date, processing 8 million kilometers each day. For all the attention Tesla receives, Mobileye is by far the leader in market share in the autonomous driving space.
This reputation and Intel’s deep pockets will help it against smaller competitors in the lidar SoC race. “I’m very confident that in the automotive industry, trust is a big differentiator,” says Mike Ramsey, a car analyst at Gartner. “Can I trust this supplier to deliver on time, to deliver in quality? And Intel has the very important feature of being a very big neck to suffocate if something goes wrong. Don’t underestimate the value of this. “
Mobileye accounts for a small percentage of Intel’s overall revenue. But along with the customer computing group – the chips that go into PCs and adjacent products – it’s the only segment that has grown in the company’s most recent quarter. It’s exactly the kind of new territory that Intel needs to occupy aggressively to avoid another smartphone-style desire.
“If you look long-term, a company like Intel needs to look for new areas of growth. It is not easy to find one. You want to look for a new market that has the size of hundreds of billions of dollars “, says Shashua, as well as one that capitalizes on Intel’s strengths. “These areas are rare. We are in this field. “
XPU marks the place
Mobileye’s lidar SoC is the clearest example of what Intel calls its “XPU” strategy – that is, looking beyond the computing CPU in all its forms. The company launched its first discrete graphics card last fall, has a dominant position in data center processors and, in 2019, acquired chip maker AI Habana Labs, which a few weeks ago won business from Amazon Web Services to use their accelerators to train deep learning models.
“We have a computing company at our center,” says Gregory Bryant, who leads the computing group for Intel customers. “We see this world where more and more things need computing, more and more things look like a computer, not only the server or the computer, but also the car, the house, the factory, the hospital. All these things need calculation and they need intelligence. “
This expansion comes at a time when Intel is facing more challenges than ever before on its traditional business lines. Manufacturing delays kept it stuck on a 10-nanometer process for making its chips, while competitors switched to smaller forms. The company’s chief engineer, Murthy Renduchintala, left last summer. And the Third Point hedge fund issued a burning public letter in late December, urging Intel to “retain a reputable investment adviser to assess strategic alternatives, including whether Intel should remain an integrated device manufacturer and potential divestment. failed purchases ”.