Hyperlume wants to make chip-to-chip communication faster and more efficient | TechCrunch
Briefly

from TechCrunch2 weeks ago
Data centers in the U.S. consumed 4.4% of electricity in 2023 and could reach 12% by 2028. Hyperlume, based in Ottawa, has developed microLED technology for faster and more energy-efficient data transfer between chips. Co-founders Mohsen Asad and Hossein Fariborzi aim to address the issues of energy consumption and latency exacerbated by AI's demands. Their microLEDs are positioned as a solution to improve communication speeds between chips, addressing the critical challenges of data center efficiency and performance.
Data centers consumed 4.4% of U.S. electricity in 2023 and are estimated to use up to 12% by 2028.
TechCrunchhttps://techcrunch.com/2025/02/19/hyperlume-wants-to-make-chip-to-chip-communication-faster-and-more-efficient/
Hyperlume created a version of microLEDs that can transfer information faster and with less energy than copper-based connections.
TechCrunchhttps://techcrunch.com/2025/02/19/hyperlume-wants-to-make-chip-to-chip-communication-faster-and-more-efficient/
Energy consumption and latency have always been problems for chip-to-chip communication in data centers, exacerbated by rapid AI growth.
TechCrunchhttps://techcrunch.com/2025/02/19/hyperlume-wants-to-make-chip-to-chip-communication-faster-and-more-efficient/
Solving the latency issue will not only speed up existing links but also unlock previously inaccessible chip capacity.
TechCrunchhttps://techcrunch.com/2025/02/19/hyperlume-wants-to-make-chip-to-chip-communication-faster-and-more-efficient/
Asad noted, 'When you have large language models [...] you need the chips to communicate with almost zero latency.'
TechCrunchhttps://techcrunch.com/2025/02/19/hyperlume-wants-to-make-chip-to-chip-communication-faster-and-more-efficient/
Read at TechCrunch
#data-centers #energy-efficiency #microled #ai-impact #latency-solutions
[
|
]