Lightmatter: This Company Is Developing The Future Of Semiconductor Chip Architecture

By Amit Chowdhry ● Aug 8, 2023

Lightmatter is a company delivering a new paradigm in semiconductor chip architecture and the next transition for large-scale computing. The company has developed photonic processors that are faster, more efficient, and cooler than any conventional processors today and is answering the call for increased compute speed, low energy density, and reduced chip heating. Pulse 2.0 interviewed Lightmatter CEO Nick Harris to learn more.

Nick Harris’ Background

Nick Harris

Harris is the founder and CEO of Lightmatter, where his team is leading the evolution of computing through photonic technology.

“Prior to founding Lightmatter, I received a Ph.D. in electrical engineering and computer science at MIT, where a lot of my work focused on quantum computing. I’ve authored over 80 academic articles in journals (including Nature, Nature Photonics, and Nature Physics) and 100 patents,” said Harris.

Formation Of Lightmatter

How did the idea for Lightmatter come together? “During my time at MIT as a Ph.D. student and postdoc, I invented and led the development of a new kind of integrated processor that uses light to calculate. This paved the way for my team at Lightmatter—my co-founders and I entered and won the MIT $100K and Harvard President’s Challenge with our business idea to power the rapid growth of artificial intelligence,” Harris reflected. “We’re excited about the rapid innovation and progress happening in artificial intelligence. It’s forcing computing infrastructure to improve at an unprecedented rate. We believe generative AI and supercomputing will be transformed by photonic technologies in the coming years and we’re thrilled to bring on new investors as well as partners and customers who are aligned with Lightmatter’s mission of enabling the future of computing infrastructure with photonics.”

Favorite Memory

What has been Harris’ favorite memory working for Lightmatter so far? “Our Series C is definitely an exciting milestone and has given the team time to reflect on all that we have accomplished. I love building, and every time a new piece of hardware comes back from fabrication, it’s like Christmas for me,” Harris shared.

Challenges Faced

What are some of the challenges Harris faced in building the company? “One of the greatest challenges of building a company is selecting and building the right team. I’m so proud of the team we’ve built, which has collected the brightest and kindest minds in the industry. The other piece is developing a deep understanding of the value of program management and the alignment it brings. When you’re very early, you just want to move fast, but when it’s time to deliver a product you’ve got to look at things a bit differently,” Harris acknowledged.

Core Products

What are Lightmatter’s core products and features? Harris told me that Lightmatter’s core products are Envise, Passage, and Idiom, which provide a full stack of hardware and software solutions to realize the benefits of photonic computing and interconnect technologies.

– Envise – The world’s first general-purpose photonic AI accelerator that enables ultra-high performance inference on the world’s most advanced AI models (GPT-3, Megatron) and staple neural networks (BERT-Large, DLRM, ResNet-50).

– Passage – A wafer-scale programmable photonic interconnect that enables arrays of all kinds of chips, as well as Lightmatter’s photonic chips, to communicate with unprecedented bandwidth and energy efficiency,

– Idiom – Lightmatter’s software stack enables frictionless interfacing with standard deep learning frameworks and model exchange formats while providing the transformations and tools required by deep learning model authors and deployers. Idiom also provides the required compiler, runtime, and tools support to achieve optimal inference speeds and accuracy on Envise.

Evolution Of Lightmatter’s Technology

How has Lightmatter’s technology evolved since launching? “We’ve come a long way since we launched the company in 2017. We unveiled the world’s first photonic computer in 2020 and introduced Passage, Envise, and Idiom in the year following. Now, we have several customer pilots with our Envise product and are working with our data center and semiconductor partners toward mass production for Passage in 2024,” Harris noted.

Significant Milestones

What have been some of Lightmatter’s most significant milestones? “The introduction of our three core products was a major milestone for us, but there’s been a lot along the way. We’ve expanded our team significantly, most recently last year with the additions of Richard Ho, who led multiple generations of Google’s Tensor Processing Unit program, and Ritesh Jain, who led datacenter chip packaging at Intel, as VPs of Hardware Engineering,” Harris pointed out.

Funding

The company recently announced a $154 million Series C funding round from Fidelity Management & Research Company, SIP, Viking Global Investors, Google Ventures, and others.

Total Addressable Market

What total addressable market (TAM) size is Lightmatter pursuing? “The biggest companies in the world are hitting an energy power wall and experiencing massive challenges with AI scalability, especially when it comes to chip memory and performance. This has profound effects for everyone, with major implications for carbon emissions and rising financial costs. The energy and power required to host and train these algorithms is immense – there are estimates that 10-20 percent of the world’s total power will go to AI inference by the end of the decade unless new compute paradigms are created,” Harris assessed. “Lightmatter is providing fundamentally new interconnect and computing technologies that enable sustainability for these power-hungry workloads by leveraging photonics. This is an entirely new category of semiconductor chip architecture, and we’re excited to power the next wave of large-scale computing.”

Differentiation From The Competition

What differentiates Lightmatter from its competition? “At Lightmatter, we’re building the most advanced photonic systems on the planet targeted at breaking through performance barriers in both computing and interconnect. In a real sense, we’re reinventing the core of computing infrastructure,” Harris revealed. “This comes at a critical time when the amount of power consumed by data center-class CPUs and GPUs is doubling every year—an unsustainable trend that will soon halt progress. Our chips enable the continued rapid growth of artificial intelligence as generative AI propels use cases across virtually every industry while at the same time minimizing computing’s growing impact on the environment.”

Future Company Goals

What are some of Lightmatter’s future company goals? “We’re excited to expand our Bay Area team and are continuing to hire across hardware and software. We have more than 20 open roles across product, R&D, and engineering. In addition, most recently last year, we brought on Richard Ho, who led Google’s Tensor Processing Unit program, and Ritesh Jain, who led datacenter chip packaging at Intel, as VPs of Hardware Engineering,” Harris concluded. “In addition to company growth, we’re working on several pilots with customers now with our Envise product, targeting the biggest cloud providers and enterprises. Envise will allow our partners to run the largest neutral networks on a fundamentally new computing platform that is decoupled from Moore’s Law and Dennard Scaling—a new path forward for computers at a time when the power consumed by AI chips and GPUs is double every 18 months. There’s a lot going on—we can’t wait to get Passage deployed in partner data centers.”