Home
  • Products
  • Solutions
  • Support
  • Company

This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

  • Products
  • Solutions
  • Support
  • Company
Community Blogs Corporate News > 10X Faster, 90% Less Energy – Lightmatter’s Photonic Supercomputer…
Tanushri Shah
Tanushri Shah

Community Member

Blog Activity
Options
  • Subscribe by email
  • More
  • Cancel
CDNS - RequestDemo

Discover what makes Cadence a Great Place to Work

Learn About
Virtuoso Studio
designed with cadence
Genus Synthesis Solution
Innovus Implementation
AI

10X Faster, 90% Less Energy – Lightmatter’s Photonic Supercomputer for AI

27 Nov 2024 • 2 minute read

The Designed with Cadence logo

With generative AI now used everywhere, the amount of power required to train and host these algorithms is immense. The number of data centers that are equipped to deal with AI workloads is rapidly increasing, but the demand for generative AI is skyrocketing. AI buildouts are driving semiconductor activity as data centers need to be outfitted with the right hardware and infrastructure.

At the same time, the countervailing limit of slower energy generation is creating demand and rewards for solutions that increase compute throughput while significantly increasing energy efficiency. Enter Lightmatter, the photonic supercomputer company that’s changing the game with the world’s fastest and first 3D-stacked photonic chips.

By using light as their primary source, photonic technologies consume a lot less power, making it a much more sustainable option for the most advanced AI and HPC workloads. Lightmatter is answering the call for a compute paradigm to power the next 50 years with its photonic technologies, and to build their future-proof photonic chips, they’re using Cadence tools.

When light transfers data, it does so at light speed, so it’s super fast yet doesn’t consume any power. Electronic circuits, on the other hand, consume energy and produce heat as they move data, which negatively affects large-scale processing, such as AI large language models (LLMs). According to Lightmatter, their chips provide 10X greater bandwidth for interconnect performance. Data centers and hyperscalars can meet the demands for LLMs while consuming 90% less power. Lightmatter is targeting the biggest cloud providers, semiconductor companies, and enterprises to help them reduce their power consumption and increase the performance of their silicon.

The chips that Lightmatter is building require a tight integration of silicon photonics as well as some electronic circuity in a single chip. For analog, they’re using Cadence Virtuoso Studio, which encapsulates all the schematics, simulations, and layout environment. They’re also using Cadence’s Genus Synthesis Solution and Innovus Implementation System for digital synthesis and placement and routing of the blocks, Voltus IC Power Integrity Solution for IR drop analysis and power consumption, and Tempus Timing Solution for timing signoff and digital circuitry. Learn more about how Lightmatter is making future-proof photonic chips for AI.

“Designed with Cadence” is a series of videos that showcases creative products and technologies that are accelerating industry innovation using Cadence tools and solutions. For more Designed with Cadence videos, check out the Cadence website and YouTube channel.


CDNS - RequestDemo

Have a question? Need more information?

Contact Us

© 2025 Cadence Design Systems, Inc. All Rights Reserved.

  • Terms of Use
  • Privacy
  • Cookie Policy
  • US Trademarks
  • Do Not Sell or Share My Personal Information