ReportGem ReportGem 中文

REAL-TIME GLOBAL RESEARCH

AI Networking: Scale-up Technology Enables China‘s AI Supernodes

Published: 2026-07-27Institution: Morgan StanleyCompany / ticker: 688256.SS,9903.HK,688041.SS,688802.SSPages: 18Original language: EnglishEvidence page: 1

Research evidence excerpt

AI Networking: Scale-up Technology Enables China‘s AI Supernodes

Idea

July 27, 2026 09:00 PM GMT

Morgan Stanley Asia Limited+MGreater China Semiconductors | Asia Pacific Ethan Jia

Research Associate

AI Networking: Scale-up Ethan.Jia@morganstanley.comMorgan Stanley Taiwan Limited+ +852 3963-2287

Charlie Chan

Equity AnalystTechnology Enables China's AI Charlie.Chan@morganstanley.com +886 2 2730-1725

Daniel Yen, CFA

Equity AnalystSupernodes

Daniel.Yen@morganstanley.com +886 2 2730-2863

Following our WAIC takeaways, we further examine China's Morgan Stanley Asia Limited+

Daisy Dai, CFA

Equity Analystemerging scale-up landscape, the role of optical interconnect,

Daisy.Dai@morganstanley.com +852 2848-7310

and implications for the domestic interconnect supply chain.

Henry Zhao

AI Supernodes enabled by scale-up are becoming the focus of China's AI Henry.Zhao@morganstanley.com +852 2239-7731

computing landscape. Alongside our China WAIC takeaways, we provide an in-

Morgan Stanley Taiwan Limited+

depth follow-up study of China’s scale-up technology development. As we wrote in Tiffany Yeh

China's AI Accelerators – Who's Poised to Win?, Chinese AI GPUs are strong in Equity Analyst

Tiffany.Yeh@morganstanley.com +886 2 7712-3032

optical networking and server rack design at the system level, although they remain

Lucas Wang

constrained by wafer process technology at the chip level ( Exhibit 6 ). These chip- Research Associate

level constraints are shifting competition in China's AI computing market from Lucas.Wang@morganstanley.com +886 2 2730-2875

standalone chip specs to system solutions. WAIC 2026 featured fewer chip launches

and more supernode solutions. Leading domestic accelerator vendors displayed

rack-scale or multi-rack supernodes with scale-up domains of 64 accelerators or

The English excerpt is extracted automatically from the cited source page and may contain layout or recognition errors. It is never batch translated.

Open report viewer