Boston-based startup, Lightmatter, which uses light for AI processing, has announced that it has raised $154 million in funding and increased its valuation as a result of clients purchasing its systems for quicker and more energy-efficient computing.

Revolutionizing AI Processing with Light-Based Hardware and Software

LightMatter is a tech company that produces chips that use the flow of light to perform computational processes such as matrix-vector products which are the core of many AI computations.

Currently, the AI industry employs GPUs and TPUs that use conventional silicon gates and transistors for such computations. However, these traditional chips are nearing their speed and density limits for a certain wattage or size.

As such, the supercomputers that enable training models like GPT-4 are large, use a tremendous amount of electricity, and generate a great deal of waste heat, a problem Lightmatter hopes to solve using light.

The telecommunications sector has used light for decades through fiber optic cables that carry data around the globe. Due to its efficiency, large computer firms are currently looking to deploy light-based technology in their data centers to cut back on power expenses.

For this reason, in 2017, Nick Harris founded the company based on optical computing work he and his colleagues completed at MIT, which is granting them the necessary patent licenses. A year later, he was able to get a $11 million seed investment that enabled the company to take the technology from prototype to product.

However, in 2021, investors began to worry about when the company would begin production with one saying at the time that “this isn’t a science project.” Harris acknowledged that while they understood “in principle” the technology should work, there was a tremendous amount of work to be done to make it operational.

Fortunately, the company was able to raise an additional $80 million from investors at around that time and the company began working on scaling the technology to production grade.

The company just held its Series C round where it has secured $154 million from SIP Global, Fidelity Management & Research Company, Viking Global Investors, HPE Pathfinder, and previous investors.

Lighting the Way to the Future

According to the CEO, Harris, Lightmatter is getting ready to make its public debut and is now running a number of test projects with its full stack of Envise – the computing hardware, Passage – the interconnect, essential for massive computing operations, and Idiom – a software platform that should allow machine learning developers to adapt quickly.

“We’ve built a software stack that integrates seamlessly with PyTorch and TensorFlow. The workflow for machine learning developers is the same from there — we take the neural networks built in these industry-standard applications and import our libraries, so all the code runs on Envise,” he explained.

With its hardware-software stack, the photonic computing firm hopes to disrupt the quickly expanding AI computation sector. The company asserts that its technology will advance the sector and significantly reduce electricity use.

“We use light to link computer chips together and we also use light to do calculations for deep learning,” Harris said to TechCrunch. He add3d:

“The reason that we’re getting these customers and data center scale deployments with our interconnect is that the generative AI boom is driving high-end chips like crazy.”

Lightmatter’s solution uses arrays of microscopic optical waveguides to enable the light to practically complete logic operations just by passing through them, making it at least theoretically faster and more effective. Since the waveguides are passive, producing the light itself, followed by reading and managing the output, uses the most power.

This type of optical computing has a particularly intriguing feature: you can boost chip power by employing multiple colors at once. Red does one action, while blue performs another, albeit in reality, 800 nanometers wavelength does one action and 820 for the other.

And while it is not easy to actualize this, these “virtual chips” can significantly boost the amount of processing done on an array. Double the colors, double the power.

What's the Best Crypto to Buy Now?

  • B2C Listed the Top Rated Cryptocurrencies for 2023
  • Get Early Access to Presales & Private Sales
  • KYC Verified & Audited, Public Teams
  • Most Voted for Tokens on CoinSniper
  • Upcoming Listings on Exchanges, NFT Drops
See the 15+ Coins