EE Times has been running a special project (More on Moore) dedicated to exploring options for increasing chip “power, performance, area, cost and speed to market” now that the scaling that Moore’s Law predicted (i.e., the number of transistors on an IC doubles every couple of years) is stalling out a bit. My most recent post focused on their articles on packaging approaches that support heterogeneous design and integration as one of the “new design and production paradigms” that are emerging. With this post, I’ll take a look at their second option: silicon photonics.
Silicon photonics is attractive because it can increase data transfer between and within chips without consuming as much power and generating as much heat as traditional all-electronic circuits. With the increasing demands placed on bandwidth, there’s a need for this sort of energy-efficient scaling. Silicon photonics doesn’t replace electronics but, since silicon is already used in IC’s, it can be used to create hybrid components that combine both optical and electronic approaches in a single chip. No new semi fabrication techniques are required. And Moore’s Law gets to keep on keeping on.
Silicon photonics has been around for a while – twenty years now – and is wending its way from being used in data centers and “industrial strength” applications into the consumer realm.
Yole is a research organization that has been keeping an eye on silicon photonics for the past decade. In their December 2021 EE Times article, Yole’s Eric Mounier and Alexis Debray provided a synopsis of where they see the market going for this technology. A decade ago, datacom was its primary market, and the overall market was small (estimated at only $65M). In 2020, they pegged the market at $87M.
Today’s landscape has shifted.
Today, dynamized by cloud applications for home office and personal use, video on demand, and 5G expansion, the primary silicon photonics application is still optical communication, with the technology integrated into 25% to 30% of optical transceivers. Some applications, such as immunoassays (Genalyte) and fiber-optic gyroscopes (KVH), will continue to grow, while LiDAR and photonic computing applications are emerging markets.
Consumer health applications are gaining in importance with the release of smartwatches that include an expanding complement of sensors. Silicon photonics is also expected to be integrated into other wearables, such as earphones. As in LiDAR, silicon photonics will enable compact and affordable optical modules.
In addition to the companies mentioned, the authors devote considerable ink to Rockley Photonics’ as being at the center of the use of silicon photonics in consumer apps, much of it in the medical technology arena. (Among other medical device providers, Rockley is partnering with industry giant Medtronic, and also collaborates with Apple on its forays into the fitness market.)
If the landscape has shifted, so have the revenue forecasts. Yole predicts a robust market for silicon photonics of $1.1B by 2026.
Further out, they see silicon photonics finding a home in quantum computing.
While Yole predicts that consumer health will displace data centers as the principal market for silicon photonics, data centers remain a major factor in market growth. As part of the Special Project on Moore’s Law, EE Times included an article by Stefani Munoz on Marvell’s work targeting cloud data centers.
“What silicon allows you to do is leverage the toolset within the standard silicon industry, so we can build these things out of 200-millimeter wafers, which give you scale, which is important,” said Radha Nagarajan, senior VP and CTO of the optical and copper connectivity group at Marvell. “Secondly, it gives you speed. You can build both photodetectors and modulators at speeds higher than what you can do with standalone lasers. And it also gives a path to a lower overall cost, but that depends on performance. Cost is all relative to performance.”
The article provides a good insight into what Marvell is doing with silicon photonics, and why.
And if you want to drill down on silicon power MOSFET technology, here’s a link to an article from Power Electronics News by Maurizio DiPaolo Emilio that was included in the EE Times Special Project.
Moore’s Law has been around a long time. Even longer than I have been. Glad to see that, with a lift from silicon photonics (and new packaging approaches) it may well last a while longer.