Data centers, responsible for powering our digital lives, consume a significant amount of energy, primarily due to chip-to-chip data transfer. Hyperlume, an innovative data transfer solution provider based in Ottawa, Canada, aims to tackle this challenge by offering microLEDs that enable faster and more energy-efficient data transfer.
Hyperlume’s microLEDs surpass the performance of copper-based connections prevalent in data centers. These microLEDs require less energy while delivering higher data transfer speeds. This breakthrough stems from the combined expertise of Hyperlume’s co-founders, Mohsen Asad and Hossein Fariborzi. Asad’s background in electrical engineering and Fariborzi’s expertise in low-power electrical circuit design led to the development of this game-changing technology.
Asad highlights that chip-to-chip communication faces significant latency issues, which are amplified by the rapid advancements in artificial intelligence (AI). Hyperlume’s solution not only reduces latency but also unlocks previously inaccessible chip capacity, enhancing overall performance.
Hyperlume’s approach is unique and cost-effective. Initially exploring silicon and lasers, the company settled on modifying microLEDs to facilitate chip-to-chip data transfer at speeds comparable to fiber optic connections. This “secret sauce” involves ultra-fast microLEDs and a low-power ASIC (Application-Specific Integrated Circuit) for efficient data communication.
Hyperlume is collaborating with select early customers, mainly in North America, to refine its product. The company has garnered significant interest from hyperscalers, cable manufacturers, and various industries seeking to leverage this technology.
Hyperlume recently closed a $12.5 million seed round to expand its engineering team and accelerate product development. This funding will pave the way for wider customer adoption. The company envisions expanding its bandwidth to meet the demands of future data centers.
In the future, Hyperlume aims to evolve into an AI connectivity solution provider, connecting chips, boards, and eventually providing comprehensive AI connectivity solutions.
Original source: Read the full article on TechCrunch
