 |
16 Oct 2024
Santa Clara firm co-founded by Columbia engineers targets AI data centers with multi-color photonics platform.
Xscape Photonics, a Silicon Valley startup aiming to reduce the enormous power consumption of AI-focused data centers with novel silicon photonics technology, has raised $44 million in a round of venture funding.
Supported by the likes of AI processor firm Nvidia and communications giant Cisco, the series A effort brings Xscape’s total fundraising to $57 million, and will be used to accelerate the development of its “ChromX” platform.
Nuclear option
According to the Santa Clara firm, ChromX can handle hundreds of wavelengths on a single optical fiber, providing flexibility and efficiency for AI workloads. Existing data center architectures have traditionally been constrained to four wavelengths.
But that traditional approach has limitations, and with AI-driven computing workloads expected to account for more than one-fifth of energy consumption in large-scale data centers by next year the likes of Google are even turning to nuclear power to provide the extraordinary levels of energy demanded by AI.
Xscape CEO Vivek Raghunathan, one of five company co-founders, explained: “Historically, performance and scalability challenges have been addressed by building bigger data centers to train large language models. This approach is not sustainable and unlocks a myriad of additional issues around energy consumption and cost.
“We are on a mission to help our customers completely re-imagine how they solve these challenges. This funding validates our mission and positions us for future growth to support next-generation AI data centers.”
Formerly the silicon photonics lead at Broadcom, Raghunathan established the startup alongside Columbia professors Michal Lipson and Keren Bergman - respective pioneers in silicon photonics and photonic interconnects.
Speaking at a plenary session during the SPIE Optics + Photonics event in San Diego in 2022, Bergman said that data centers were increasingly impacted by energy and bandwidth bottlenecks.
"Our recent work has shown how integrated silicon photonics with comb-driven dense wavelength-division multiplexing can scale to realize petabyte-per-second chip escape bandwidths with sub-picojoule-per-bit energy consumption," she reported at the time.
“We use this emerging interconnect technology to introduce the concept of embedded photonics for deeply disaggregated architectures. Beyond alleviating the bandwidth-energy bottlenecks, the new architectural approach enables flexible connectivity tailored for specific applications."
Frequency combs
Also hailing from Columbia, Xscape's two other co-founders are nonlinear photonics professor Alexander Gaeta, winner of the 2019 Townes Medal, and optical frequency comb specialist Yoshi Okawachi.
“We are excited about accelerating commercialization of our unique multi-color laser photonics technology, for different custom use cases in the AI and hyperscaler markets,” said Gaeta.
The latest financing effort was led by IAG Capital Partners, with Altair, Cisco Investments, Fathom Fund, Kyra Ventures, LifeX Ventures, Nvidia, and OUP also joining.
Xscape’s filings with the US Securities & Exchange Commission (SEC) indicate that the company raised $10 million in a seed funding round led by Altair in November 2022, and a further $3 million in debt financing last year ahead of the series A round.
IAG associate Alex Kash commented: “We see tremendous value in the technology that Xscape Photonics is developing to solve challenges customers are facing with AI data center energy usage and bandwidth performance.
“The future of the data center will be built around photonics and this capital will help to dramatically accelerate Xscape Photonics’ product development.”
Having recently attracted VP of engineering Ramesh Senthinathan from Lumentum, the startup is currently looking to hire for several senior photonics-related roles.
 |  |  |  |  |  |  |
| © 2024 SPIE Europe |
|
