DigestAI news desk

Cut through the AI noise.

Listen
Hardware & Compute1 min read

Innoscience shows GaN power solutions for AI servers at OCP Summit

Innoscience presented its gallium nitride power semiconductors at the OCP Global Summit in late October as part of the NVIDIA MGX ecosystem. The Chinese manufacturer says its GaN-on-silicon devices cut conversion losses, raise power density and improve efficiency for AI servers and 800 VDC data-center architectures.

1 source

Key points

  • Innoscience demoed GaN power solutions at OCP Summit as NVIDIA MGX partner
  • Company claims largest vertically integrated GaN platform with 8-inch volume production
  • Yole Group projects power GaN market at US$3.5B by 2031, 35% CAGR

Innoscience claims to run the world's largest vertically integrated GaN platform with high-volume 8-inch wafer production, giving hyperscalers supply-chain resilience and cost competitiveness. Yole Group projects the power GaN market will reach about US$3.5 billion by 2031 at a 35% CAGR, with AI data-center and telecom applications growing at 45% CAGR. Grand View Research sees the broader GaN device market hitting roughly US$28.9 billion by 2033.

Full story from trendforce.com · by TrendForce · via Search: NVIDIAOpen source ↗

[News] Innoscience Advances AI Infrastructure Power as Part of the NVIDIA MGX Ecosystem

trendforce.com · 2 November 2026

Loading the full article…

This text was published by trendforce.com and written by TrendForce. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page
InnoscienceNVIDIAYole GroupGrand View Research

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Hardware & Compute

All →

Related stories

hot MIT researchers report GPT-6 Astra natively reads and writes neural network weights, raising Qwen3.5-0.8B MATH-500 from 7.9% to 48.1% by editing 0.016% of weigh new Google launches Dreambeans AI app in India hot OpenAI seeks $30 billion funding round at $1.25 trillion valuation new Innoscience shows GaN power solutions for AI servers at OCP Summit hot Oracle files to sell $10B in stock and $10B in bonds to fund AI buildout new Google’s AI Mode cuts search ad revenue by $1.6B in Q3 hot Anthropic reports $17B Q3 revenue, $895.4B compute commitments in amended IPO filing new Huawei reportedly plans to use Samsung, SK Hynix HBM in Ascend 960DT