Nvidia's Kyber NVL144 reportedly pushed back more than a year, Asian suppliers drop

A report from analyst firm SemiAnalysis says Nvidia's next-generation AI server rack has been delayed by more than a year. Suppliers across Japan, Taiwan, South Korea, and Hong Kong saw their stock prices drop sharply in response.

Asian tech stocks slid after analyst firm SemiAnalysis reported on X that Nvidia's next AI server rack system, Kyber NVL144, has been pushed back more than twelve months to 2028 because of manufacturing problems.

The issue, according to SemiAnalysis, is the circuit boards. Specifically, the PCB midplane, a central board that connects all the individual components, has proven extremely difficult to produce without defects. Nvidia CEO Jensen Huang had shown off Kyber NVL144 just three months earlier at the company's GTC conference.

The report hit an already jittery investor base. After years of AI-fueled gains, even small setbacks trigger sharp sell-offs. Japanese PCB maker Ibiden, which counts Nvidia as its largest customer, dropped as much as ten percent. Kingboard Laminates fell 18 percent in Hong Kong, Elite Material lost ten percent in Taiwan, and Samsung Electro-Mechanics slid eleven percent in South Korea. The sell-off follows massive run-ups. Samsung Electro-Mechanics had gained more than 600 percent this year, and Kingboard Laminates more than 470 percent.

Scrapped designs give competitors an opening

SemiAnalysis lists more setbacks. A planned alternative design called NVL72x2, which would have placed two Oberon racks back to back, has been scrapped entirely. Cloud providers and large data center operators pushed back against the unusual form factor and the high operational overhead. The more powerful version of the upcoming Rubin Ultra chip with four compute dies has also been canceled. Only the smaller two-die version remains, delivering roughly half the real-world compute performance.

A key interconnect technology called CPO-NVSwitch, which links many chips into a single large system, won't arrive until the generation after next, called Feynman. That means Nvidia lacks a proven way to scale Rubin Ultra to very large systems for now. The gap could give competitors like AMD's MI500X or Google's TPUv8i Broadfly room to move in. Nvidia plans to make up for the shortfall by selling more Oberon-Rubin racks in the existing form factor.

A Kyber delay doesn't necessarily mean overall AI spending will shrink, said Gary Tan of Allspring Global Investments, according to Bloomberg. It shows that Nvidia's most ambitious system is taking longer than expected. The current stock weakness is mostly driven by profit-taking. Shawn Oh of NH Investment & Securities pointed to growing uncertainty around Nvidia's expansion plans, which gives alternative AI platforms more room to compete.