Upgrade to SI Premium - Free Trial

Nvidia develops 6G radio chip as data center shifts stay on track - Jefferies

June 12, 2026 1:03 PM

Investing.com -- Jefferies said delays in 800V HVDC adoption for AI data centers and co-packaged optics deployment do not represent material changes to existing timelines for European semiconductor suppliers.

SemiAnalysis reported that hyperscaler adoption of 800V HVDC architecture has been pushed to 2028. Jefferies said this does not constitute a significant delay, as most semiconductor companies already anticipated small volume shipments of 800V HVDC products from late 2027, with momentum building in 2028.

The firm said it continues to expect Infineon's AI power revenues to reach at least €1.5 billion this year and grow to around €3 billion in fiscal year 2027. This growth is driven by the conversion of GPU and ASIC customers from lateral to vertical power delivery on voltage regulator modules and expansion in gigawatt installations.

Jefferies said it expects Aixtron to receive GaN-related orders for 800V HVDC by the end of 2026 or in the first half of 2027.

Reports also indicated that co-packaged optics shipments in scale-out applications have been revised down for 2026-2027, with scale-up shipments starting in 2029. Jefferies said the timeline for CPO introduction in scale-up matches its existing expectations.

The firm views any CPO adoption delays as neutral or slightly positive for STMicroelectronics. While STM provides silicon photonics and BiCMOS components for scale-out pluggable transceivers, TSMC will likely handle silicon photonics fabrication for CPO in both scale-out and scale-up applications.

LightReading reported that Nvidia is developing a GPU for 6G radio units to address beamforming and other applications in future massive MIMO radios. Jefferies views this as neutral for Nokia and Ericsson, given Nvidia already announced its entry into 6G base station hardware through a partnership with Nokia in October 2025.

Nokia has stated it plans to reduce RAN chip development while focusing on software that runs on Nvidia GPUs.

Categories

AI General News

Next Articles