Daily Management Review

Chip Startups That Use Light Instead Of Cables Are Gaining Traction And Funding


Chip Startups That Use Light Instead Of Cables Are Gaining Traction And Funding
Computers that process data using light rather than electric currents, which were once considered research projects, are gaining traction, and businesses that have mastered the engineering difficulty of exploiting photons in chips are receiving significant funding.
In the most recent example, Ayar Labs, a startup creating silicon photonics, announced on Tuesday that it had raised $130 million from investors including chip giant Nvidia Corp.
While transistor-based silicon chips have enhanced processing power tremendously over the past decades as transistors have shrunk to the width of few atoms, further decreasing them is difficult. Not only is it difficult to create anything so little, but signals might flow between them as they get smaller.
As a result, Moore's law, which said that the density of transistors on a chip would double every two years, lowering costs, is decreasing, forcing the industry to seek new ways to address increasingly high artificial intelligence computing needs.
According to data provider PitchBook, silicon photonics startups raised more than $750 million last year, more than doubling from 2020. That amounted to nearly $18 million in 2016.
"A.I. is growing like crazy and taking over large parts of the data center," Ayar Labs CEO Charles Wuischpard told Reuters in an interview. "The data movement challenge and the energy consumption in that data movement is a big, big issue."
The problem is that many large machine-learning algorithms can utilise hundreds or thousands of processors to compute, yet current electrical technologies have a speed limit on data transmission between chips or servers.
Light has been used to transfer data over fiber-optic cables, including undersea cables, for decades, but getting it to the chip level has proved difficult due to the fact that components required to create or control light are not as easy to compress as transistors.
Brendan Burke, senior emerging technology analyst at PitchBook, predicts that silicon photonics will be mainstream hardware in data centres by 2025, with a market size of $3 billion by then, equivalent to the market size of the A.I. graphic chips market in 2020.
Startups leveraging silicon photonics to construct quantum computers, supercomputers, and processors for self-driving vehicles are also raising large sums of money.
PsiQuantum has raised around $665 million to far, despite the fact that the promise of quantum computers altering the world is still years away.
Lightmatter, which develops processors that use light to accelerate AI workloads in datacenters, has secured a total of $113 million. The company plans to release its chips later this year and begin testing with clients soon after.
Luminous Computing, a firm sponsored by Bill Gates that is developing an AI supercomputer utilising silicon photonics, has raised a total of $115 million.
This technology is being advanced by more than just startups. Semiconductor companies are likewise preparing to apply their silicon chip-making capabilities for photonics.
Collaboration with PsiQuantum, Ayar, and Lightmatter, according to GlobalFoundries Head of Computing and Wired Infrastructure Amir Faintuch, has helped establish a silicon photonics manufacturing platform for others to exploit. The platform went live in March.
Peter Barrett, the founder of venture capital firm Playground Global and an investor in Ayar Labs and PsiQuantum, believes in the long-term prospects for silicon photonics for speeding up computation, but warns that the road ahead is long.
"What the Ayar Labs guys do so well ... is they solved the data interconnect problem for traditional high-performance (computing)," he said. "But it's going to be a while before we have pure digital photonic compute for non-quantum systems."