Insider Brief
- Opticore, a photonic computing startup, has officially launched with funding from deep tech investors including Sagax Capital, Jetha Global, and Bioeconomy, aiming to revolutionize datacenters and AI through energy-efficient optical processing units (OPUs).
- The company’s technology uses light and waveguides instead of traditional electrical components, enabling computations with significantly lower energy consumption, higher speeds, and scalability for advanced AI and deep learning applications.
- Opticore’s innovative approach addresses computational bottlenecks in existing architectures and positions its photonic chips as a transformative solution for AI supercomputing, paving the way for larger machine-learning models and sustainable energy use.
PRESS RELEASE — Opticore, a photonic computing company, announces its official launch and funding today. Opticore aims to revolutionize datacenters and AI with its innovative approach to photonic computing.
Opticore is developing optical processing units (or OPUs) that can do the same computational tasks as GPUs but using light and waveguides instead of electrical components allowing computation at multiple orders of magnitude lower energy costs and higher clock speeds.
“The era of computing with light has arrived with the Opticore technology. Our architecture overcomes the memory bottlenecks in CMOS electronics and allows for processing billions of parameters in a single chip using 100-fold lower energy. Our chips are fabricated with standard foundry processes. This funding would allow us to build a much larger chip.” commented Zaijun Chen and Ryan Hamerly, co-founders of Opticore. As photonics has become an essential in datacenters for chip-to-chip communications, computing with light will fully unlock its potential for the next-generation energy efficient supercomputers.
“I had never seen such a smart computing architecture before. Opticore technology with co-packaging optics is the solution for scalable AI computing”, said Prof. Mengjie Yu, co-founder of Opticore and Professor at UC Berkeley. “AI models are limited in their size by the power of today’s supercomputers. It’s just not economically viable to train models that are much bigger. This technology could make it possible to leapfrog to machine-learning models that otherwise would not be reachable in the near future”, said Prof. Dirk Englund (an advisor to Opticore) in MIT News.
The round was funded by Alex Turnbull of Sagax Capital and Karan Danthi of Jetha Global, two renowned deep tech investors. It was also backed by Dorjee Sun of Bioeconomy, a climate-focused investor, among others.
“The Opticore team has lit a path to a state of the world in which high performance computing, climate targets and supply chain security are totally compatible. They have done this with a novel and creative approach on existing platforms that can be fabricated in the US, EU and Singapore” said Alex Turnbull at Sagax Capital.
“At Jetha Global, our extensive research shows that optics will play a crucial role in the evolution of data centers, initially transforming the interconnect layer and eventually extending to the xPUs,” said Karan Danthi, Chief Investment Officer at Jetha Global. “We’ve made substantial investments in the silicon photonics space, exploring a range of opportunities, and we are confident that Zaijun and his team at Opticore are uniquely positioned to achieve the coveted outcome: a chip that not only matches but surpasses the performance of leading xPUs, all while consuming a fraction of the power.”
Opticore’s technology allows parallelized computations and particularly matrix multiplication to be done using light. This de-bottlenecks a significant constraint in computational speed and efficiency. This can allow better progress in AI and deep learning at much lower energy and infrastructure costs all the while being fabbed on lagging edge nodes.
For more information about Opticore, please visit https://www.opticore.ai/ and for an outline of the company’s technology and vision please see our post “Hurdling the Power Wall With Photonic Computing”.
-End-
About Opticore:
Opticore is founded by a team of MIT scientists. Their mission is to provide scalable, energy-efficient computing solutions that address the global energy and environmental challenges posed by the growing demands of AI computing. They have previously demonstrated a computing system with 100x more energy efficiency and area density, in Nature Photonics 17, 723–730 (2023). Opticore will scale up this technology with silicon photonics and optoelectronic packaging.
For media inquiries, please contact Adam Cai adam@opticore.ai
