Lumentum – the optical and photonic developer subject to a recent $2 billion cash injection from Nvidia – unveiled an interconnect solution aimed at AI scale-up networks.

Showcased at the OFC event in Los Angeles, the platform uses 1060 nanometer (nm) vertical-cavity surface-emitting lasers (VCSELs) to connect AI chips via light while overcoming bandwidth and heat limitations of traditional connections.

Using high-density VCSEL arrays similar to those found in 3D sensing, Lumentum’s offering comes co-packaged with a host application-specific integrated circuit (ASIC) that supports scale-up protocols like universal chiplet interconnect express (UCIe) and peripheral component interconnect express (PCIe).

“AI infrastructure is pushing the limits of conventional electrical interconnects,” Lumentum CTO Matt Sysak explained. “Our 1060nm VCSEL scale-up platform optically enables ASICs with high shoreline bandwidth density, strong signal integrity, and thermal resilience, while providing a differentiated and highly reliable alternative to traditional silicon photonics-based approaches.”

Lumentum’s OFC-debuted offering looks to provide a scalable optical platform for next-generation rack-level architectures.

The San Jose, California-based firm contends that traditional electrical connections in AI scale-up infrastructure need to be capable of supporting intense bandwidth in addition to heat.

Its new offering is designed to support operations exceeding 302°F (150°C). Meanwhile, lasers in silicon photonic interconnects typically need to be kept away from heat, with high temperatures impacting reliability.

Lumentum’s VCSEL arrays use less power to turn electricity into light, while being able to be co-packaged directly onto the aforementioned ASIC to provide a more practical hardware offering.

The 1060nm VCSEL platform is built on the vendor’s ultra-high-volume 3D sensing manufacturing base, of which it claimed to have shipped more than 10 billion emitters to date.

Nvidia's strategic supply chain investments

Lumentum said the light-based interconnect “also provides an independent supply chain alternative to silicon photonics and indium phosphide (InP) laser-based architectures.”

Nvidia’s mammoth investment in the firm was part of the tech giant’s ongoing effort to shore up technologies and components it deems vital to keeping up with demand for its stack. Having previously supplied InP laser solutions for Nvidia’s Spectrum-X and Quantum-X networking switches, the capital injection will see Lumentum build out its manufacturing capabilities with a new fabrication site.

During its recent GTC event, Nvidia CEO Jensen Huang touched on the Lumentum deal and other recent investments in Ayar Labs and Coherent, saying he was “constantly just looking across the entire life cycle of the technology and manufacturing across the supply chain” to “prepare our company for growth.”

“When you look upstream, you'll come to the conclusion that we're scaling,” the CEO explained during a press conference attended by SDxCentral. “We're starting to scale our silicon photonics technology with Spectrum-6, and we're starting to add it to our scale-up technology with NVLink in the coming years, which means the amount of silicon photonics technology capacity that we need is substantially higher than the world has today.

“So we work with the supply chain to make sure that we can help them build up the capacity in advance," Huang continued. "Sometimes we give them [a] prepayment. Sometimes we just give them a forecast. Sometimes, if the technology is in capacity or is quite light but needs to be much, much greater, we might decide to invest in the company and provide all of the other things as well. We did that with Coherent, Lumentum, etc."