Researchers at Laval University have created an optical chip capable of transmitting data at 1,000 gigabits per second with minimal energy consumption, presenting a significant advancement for AI technology.
Researchers from the Centre for Optics, Photonics and Lasers (COPL) at Université Laval in Quebec have unveiled a breakthrough optical chip that can transfer massive amounts of data at ultra-high speeds while consuming minimal energy. This innovation, published in the latest edition of Nature Photonics, could revolutionize artificial intelligence systems known for their significant power demands.
AI systems such as ChatGPT are infamous for their extensive energy consumption. Addressing this issue head-on, the COPL team designed an optical chip, as thin as a strand of hair, that can transmit data at unprecedented speeds with superior energy efficiency.
By leveraging the power of light, this chip not only uses the light’s intensity but also its phase — essentially the shift within the light signal — to achieve optimal performance levels, all within an ultra-compact design.
“We’re jumping from 56 gigabits per second to 1,000 gigabits per second,” lead author Alireza Geravand,a doctoral student at Laval, said in a news release.
The Equivalent of 100 Million Books Transferred in 7 Minutes
Geravand highlights the technology’s potential to transform AI training processes.
“At 1,000 gigabits per second, you could transfer an entire training dataset — the equivalent of over 100 million books — in under seven minutes. That’s about the time it takes to brew a cup of coffee,” Geravand added.
Astonishingly, this feat would only consume 4 joules of energy, comparable to heating one millilitre of water by one degree Celsius.
The innovation relies on microring modulators, tiny ring-shaped silicon devices that manipulate light to encode information. The system employs two pairs of rings: one dedicated to intensity and the other to phase.
Today’s AI data centers rely on tens or even hundreds of thousands of processors working in tandem, mimicking neural communication in the brain. Each processor spans several millimeters, creating vast structures that consume substantial energy.
“You end up with a system that’s kilometers long,” added Geravand.
With their new technology, processors can communicate seamlessly and efficiently, as if they were just meters apart, presenting a significant technological leap as AI demand continues to surge.
Notably, companies like NVIDIA have started incorporating microring modulators into their systems, although ones limited to light intensity alone. This new development by COPL could inspire broader industry adoption in the near future.
“Ten years ago, our lab laid the groundwork for this technology. Today, we’re taking it to the next level. Maybe in a few years, the industry will catch up, and this innovation will make its way into the real world,” Geravand concluded.
Source: Université Laval