Highlights:
- By addressing one of the primary bottlenecks in AI, specifically the data transfer between the potent accelerator chips used for AI training and inference, Ayar Labs hopes to achieve this.
- According to a recent Goldman Sachs estimate, spending on AI infrastructure is expected to reach over USD one trillion over the next ten years, so it’s possible that all of those startups may have the opportunity to have an impact in the years to come.
Ayar Labs Inc. raised USD 155 million in the later-stage funding round spearheaded by Light Street Capital and Advent Global Opportunities. The startup develops optical interconnect systems to transmit data across computer chips through light.Leading chipmakers like Nvidia Corp., Advanced Micro Devices Inc., Intel Corp., and GlobalFoundries Inc. participated in the Series D round as well, contributing through their respective investment arms. Participating in the round was VentureTech Alliance, which has a strategic alliance with Taiwan Semiconductor Manufacturing Co.
At a time when the semiconductor industry is working to provide more effective chip-to-chip connections to serve artificial intelligence workloads, the investment raises Ayar Lab’s valuation above USD one billion and takes its total capital to USD 370 million.
By addressing one of the primary bottlenecks in AI, specifically the data transfer between the potent accelerator chips used for AI training and inference, Ayar Labs hopes to achieve this.
Traditional electrical interconnects are still used for the majority of data center operations today, but the industry has had difficulty increasing their bandwidth. Ayar Labs states that its light-based interconnects can transmit more data way more quickly than its optical interconnect systems.
According to Ayar Labs, their TeraPHY optical connectivity module can assist clients optimize the performance and computation efficiency of AI infrastructure while reducing expenses because it consumes less power than conventional interconnects. Performance as well as the “profitability metrics of AI applications” will be enhanced in this way.
Using their SuperNova remote light technology, the business claims that TeraPHY is the first in-package I/O chiplet in the industry that can be connected with system-on-chip packages to enable quicker chip-to-chip communications.
It claims that it is prepared to begin scaling up the TeraPHY chip technology with the funding infusion from the funding round.
According to Jordan Katz, a partner at Advent Global Opportunities, Ayar Labs is the most ideally positioned firm to spearhead the revolution in AI infrastructure that optical interconnect technology is expected to bring about. “We believe that supporting Ayar Labs and its groundbreaking technology will offer transformative benefits to AI systems,” he added.
Although the support of so many significant semiconductor industry firms seem to indicate that Ayar Labs is about to do something significant, there is no assurance that Nvidia AMD and Intel will quickly embrace the TerayPHY technology.
They can also gain a lot better understanding of the startup’s operations by investing in it, which could help their own optical interconnect research. Obviously, if they choose to adopt Ayar Labs’ technology, the money should also position them at the line’s head.
However, there is a lot of rivalry for Ayar Labs in this field. One of its most formidable competitors is Lightmatter Inc., a startup that has drawn a lot more capital from investors after closing a USD 400 million round in October, bringing its total fundraising to USD 850 million. Additionally, Lightmatter is valued at approximately USD 4.4 billion, which is far greater.
Other rivals include Lightium AG, a Swiss business that closed a USD seven million venture funding round in September to try to commercialize its technology-based Thin-Film Lithium Niobate optical interconnects. Additionally, Xscape Photonics, a competitor, reported raising USD 44 million in October, increasing its total investment to USD 57 million.
According to a recent Goldman Sachs estimate, spending on AI infrastructure is expected to reach over USD one trillion over the next ten years, so it’s possible that all of those startups may have the opportunity to have an impact in the years to come. The research also emphasizes how urgently technology that might remove bottlenecks in AI workflows are needed.