KISTA, Sweden, June 2, 2026, 13:03 CEST
- Sivers Semiconductors said its laser arrays are set to be used in GlobalFoundries’ silicon-photonics reference designs targeting AI data-center optical links.
- Sivers surged roughly 65% in Stockholm, while GlobalFoundries was last seen at $79.93 ahead of the U.S. open.
Sivers Semiconductors jumped Tuesday after the Swedish photonics company said it’s working with GlobalFoundries to deliver laser arrays for optical connections in AI data centers. That part of the chip market has been getting a closer look as copper lines run into limits in computing clusters.
Timing is a big deal here since AI systems now have more to deal with than just processor limits. These systems need to shuttle data between chips and racks with less heat and less power wasted. That’s driving chipmakers to look at optical interconnects — these are links that use light to move signals instead of electrical currents.
Sivers said its lasers will be part of reference designs on GlobalFoundries’ silicon-photonics platform. Silicon photonics uses light for high-speed data on chips or in chip packages. The companies said their work is aimed at a pluggable-optics market they estimate could hit $25 billion by 2030.
GlobalFoundries and Sivers are teaming up on co-packaged optics, or CPO—optics that go right next to processors or switches. The deal also includes linear pluggable optics, or LPO, which are swappable modules aimed at lowering power use and complexity in fast networks. Sivers’ lasers are set to be offered through GlobalFoundries’ SCALE optical-engine platform.
Raymond Biagan, chief revenue officer at Sivers, said AI workloads need “higher bandwidth, improved energy efficiency, and scalable optical connectivity.” GlobalFoundries senior fellow for silicon photonics Vikas Gupta said the company has “strong momentum” as AI data-center designs push to denser bandwidth and lower power. PR Newswire
GlobalFoundries put out SCALE in May, saying it’s the first platform to meet the Optical Compute Interconnect Multi-Source Agreement, which is an industry spec for AI scale-up interconnects. SCALE relies on coarse and dense wavelength-division multiplexing to combine several colors of light on a single fiber, lifting bandwidth density over copper.
GlobalFoundries’ chief business officer Mike Hogan said at the SCALE launch the company “stands ready to unlock” high-bandwidth, energy-efficient connectivity. GF at the time said it had shown 8-wavelength and 16-wavelength two-way dense wavelength-division multiplexing on its platform. GlobalFoundries Inc.
Chipmakers are piling in. STMicroelectronics said Tuesday it expects to decide by year-end whether to grow its Crolles fab in France, citing higher demand for silicon photonics as AI expands. Reuters reported last year Nvidia had teamed up with TSMC to package some networking chips with optical links.
GlobalFoundries has been moving to strengthen its business. The company acquired Advanced Micro Foundry, a Singapore silicon-photonics firm, in November, saying the buy would grow its photonics capacity and research presence in Singapore. Deal terms weren’t made public.
Motley Fool put GlobalFoundries’ latest message in simple terms: AI struggles aren’t just about compute, but about chip interconnects. That’s what the Sivers partnership is pitched to tackle. For now it’s still just a partnership announcement, not confirmation of meaningful orders.
There are risks. The companies have not shared financial terms, production timelines, customer details or any forecast for revenue from the partnership. Sivers said on May 29 that net sales for the first quarter dropped 22% to SEK 61.9 million, blaming U.S. defense budget delays and forex pressure.
There’s another issue in play. Rosen Law Firm said Monday it’s looking into possible securities claims after Ningi Research put out a short report accusing Sivers of accounting and customer-contract issues. Rosen said Sivers’ OTC shares dropped 9.2% on June 1. Sivers didn’t mention the report in its collaboration update Tuesday.
Related articles: Data Centers
