The surge in AI applications in recent years has also led to a bigger demand for more powerful silicon that can train and run future AI models. Even as we continue to build smaller and faster chips, the limitations of electronics-based computing technology are coming to the fore.

In electronic chips, electron flux passes through components such as resistors, capacitors, transistors, etc., and produces heat as a byproduct. For AI-based applications and data centers, chips are assembled on a large scale, generating large amounts of heat. This requires more power on cooling infrastructure, making the entire setup highly inefficient.

What is a photonic chip?

A photonic chip is much like a silicon-based chip, but it uses photons or light instead of electrons to process information.

To facilitate the shift in processing information, a photonic chip uses optical components such as waveguides, lasers, and polarizers instead of electronic ones such as resistors or transistors to process information.

Such an approach allows chips to be further miniaturized while processing information faster and with low thermal effects. The technology can be easily scaled and integrated into existing data processing flows, making it an attractive proposition in the field of computation.

World’s first multifunctional photonic chip

Funded by the European Research Council, the researchers at UPV and iPRONICS have built a revolutionary chip that allows on-demand programming and interconnects the wireless and photonic segments seamlessly. The approach removes bottlenecks that can thus be generated and improves the capacity and bandwidth of the chip.

“It is the first chip in the world with these characteristics,” said José Capmany, a professor of optical communications at UPV. “It can implement the twelve basic functionalities required by these systems and can be programmed on demand, thus increasing the efficiency of the circuit.”

Applications such as 5G or autonomous vehicles require higher frequency and, therefore, need further to reduce the size of antennas and integrated circuits. Capmany’s team at the UPV has already been working on this and has successfully designed multiple components.

In a press release, the team said that it has developed an interface chip, a converter that goes behind the antenna, as tiny and compact as possible while it supports frequency bands that are currently used and those that will used in the future.

iPronics has integrated this chip into one of its products, the Smartlight processor, which Vodafone is currently testing.

“For us, the development of this chip is a crucial step because it has allowed the validation of our developments applied to a growing problem, the efficient management of data flows in data centers and networks for artificial intelligence computing systems,” said Daniel Pérez-López, co-founder and CTO of iPronics in the press release. “Our next goal is to scale the chip to meet the needs of this market segment.”

The research findings were published in the journal Nature Communications.