In today’s digital era, the demand for ever-increasing computing power has been overwhelmingly huge, driven primarily by advancements in artificial intelligence. However, the constant innovation in computing technology is facing obstacles, primarily due to the limitations in the shrinking size of transistors used in chips. This imposes a strict limit on Moore’s Law and Dennard’s Scaling which have so far governed the advancements in computing capabilities.
In this regard, Lightmatter, a startup established by three MIT alumni, has taken a revolutionary approach to progress in computing technology. The company’s innovation centers around using light, instead of relying solely on electricity, for data processing and transportation in chips. The first two products from Lightmatter, an AI chip and an interconnect for data transfer between chips, respectively utilize both light and electricity to enhance their efficiency.
The company has gained solid financial backing to the tune of more than $300 million with a valuation of $1.2 billion, demonstrating the market’s high expectations for the broad application of its technology. It is now working with some of the most prominent technology companies globally to potentially meet the exponentially growing energy need of data centers and AI models.
Born out of the work done by Nicholas Harris, one of the co-founders, for his PhD project on photonic quantum computing at MIT, the company’s core technology utilizes the benefits of photonics for AI operations. With this technology, multiple calculations can be performed simultaneously depending on the different colors of light used which results in driving up operational efficiency and hardware reuse, and thus, energy efficiency.
The other product, Passage, harnesses the latency and bandwidth advantages of light to link processors in a similar way to how fiber optic cables use light for transmitting data over long distances. This could enable larger chips to operate as single processors, significantly optimizing data communication for large AI models or cloud computing servers.
Increasing energy efficiency in computing is an urgent need today in response to an anticipated rise in power consumption for data centers and computing, climbing to 80% of all energy usage on the planet by 2040. Large AI model deployments are contributing to the power usage scale of entire cities.
The long term aim of Lightmatter is to continue using light to enhance the speed, efficiency and capabilities of computers, potentially leading to a revolutionary change in the computing landscape. It is currently engaged with chipmakers and cloud service providers for mass deployment of its game-changing technology which uses silicon and therefore can be produced without significant changes in existing semiconductor fabrication processes. This could lead to a major breakthrough in the computing industry with significant environmental and economic implications.
