Sivers to provide lasers for GlobalFoundries; Nvidia ramps optical AI 'engine'

Sivers Semiconductors, the Swedish device foundry with a photonics division in Scotland, has agreed a strategic collaboration with semiconductor giant GlobalFoundries (GF) that will see its lasers deployed in cutting-edge AI connectivity applications.

Sivers says that its laser arrays will be integrated into reference designs built on GF’s silicon photonics platform, for technologies including co-packaged optics (CPO) and linear pluggable optics (LPO).

The laser emitters will also feature in GF’s recently announced silicon photonics co-packaged advanced light engine (SCALE) optical module for CPO, which is expected to accelerate the adoption of optical scale-up interconnects.

Scalable solutions

SCALE combines integrated photonic devices, both coarse and dense wavelength-division multiplexing (CWDM and DWDM), and advanced packaging to improve bandwidth density and system scalability.

Vikas Gupta, a GF senior fellow responsible for the foundry giant’s silicon photonics product line, commented: “GlobalFoundries continues to see strong momentum for silicon photonics solutions as AI data center architectures evolve toward higher bandwidth density and improved power efficiency.

“Pairing Sivers Semiconductors' laser array technology with our silicon photonics and SCALE CPO platforms provides our customers with advanced, scalable optical engine solutions for high-bandwidth co-packaged optics and optical interconnects.”

Sivers’ chief revenue officer Raymond Biagan added: “The rapid expansion of AI workloads and hyperscale data center architectures demand advanced photonics technologies that deliver higher bandwidth, improved energy efficiency, and scalable optical connectivity.

“Our collaboration with GlobalFoundries positions both companies at the leading edge of silicon photonics innovation.”

Nvidia ramps CPO with Vera Rubin AI 'engine'

Meanwhile Nvidia, whose cutting-edge chips underpin the AI infrastructure build-out, announced that it was now ramping production of the Vera Rubin platform, which incorporates the firm’s “Spectrum-X Ethernet Photonics” CPO-based switches.

The CPO architecture, developed in collaboration with Lumentum, Coherent, Corning, and others, is said to deliver five times better power efficiency than traditional transceivers.

“By simplifying design and freeing more power for compute, Nvidia CPO networking provides the foundational fabric for million-GPU AI factories, with CoreWeave, Lambda, and Oracle Cloud Infrastructure among the first ecosystem partners and adopters,” said the chip firm. 

Nvidia also says that the likes of Dell Technologies, HPE, Lenovo, Supermicro, ASUS, Foxconn, and others are among those adopting the firm’s “DSX” platform designed to work in tandem with Vera Rubin-powered AI compute, with production shipments slated to begin this fall.

The AI chip giant’s CEO and founder Jensen Huang added: “Agentic AI is a new kind of workload. One prompt can launch a thousand-step journey of reasoning, retrieval, tool use and response generation.

“Vera Rubin was built for this moment - an AI factory engine that delivers intelligence at scale, with the performance, efficiency, and security needed to power the next industrial revolution.”