Nvidia has reportedly delayed the launch of its next-generation Kyber NVL144 AI server rack by over a year, pushing it back to 2028 due to severe issues in the circuit board manufacturing process. According to market analysis firm SemiAnalysis, this technical setback not only slows down the US chip giant's hardware roadmap but has also indirectly wiped out double-digit valuations for several major Asian component suppliers. This marks a significant hurdle for Nvidia amidst an intensifying AI hardware race.
Detailed Developments
According to sources from SemiAnalysis, the development roadmap for Nvidia's new ultra-server lineup hit a major roadblock linked to the printed circuit board (PCB) supply chain. This delay forced the company to reschedule the Kyber NVL144 release from its original timeline to 2028. Furthermore, Nvidia has reportedly decided to completely cancel the Rubin Ultra variant, which was intended to be the higher-performance configuration in this product line. The news immediately sent negative shockwaves through Asian stock markets, leading to sell-offs and sharp declines in the share prices of Nvidia's manufacturing partners.
Technical & Technology Analysis
The root cause of the Kyber NVL144 delay lies in the complex technical challenges of manufacturing large-scale integrated circuit boards. The NVL144 server rack requires extremely high connection density, superior heat resistance, and high-signal transmission efficiency to operate multiple high-power AI GPUs simultaneously. Failing to secure acceptable production yields for boards meeting these standards created a bottleneck in the assembly line. The cancellation of the Rubin Ultra variant also indicates that Nvidia is restructured its entire board architecture to optimize feasibility for mass production.
Expert Opinions & Insights
Analysts at SemiAnalysis suggest that this setback is a non-trivial blow to Nvidia's near-monopoly. Although Nvidia still dominates the AI chip market, a delay of over a year for a flagship product line creates a substantial market gap. Direct competitors like AMD, with its Instinct chip family, and Google, with its custom Tensor Processing Units (TPUs), now have a golden opportunity to narrow the gap and capture market share from major data center clients.
Impact & Future
Nvidia's setback demonstrates that even leading tech giants are bound by physical limitations and supply chain risks when racing to upgrade AI hardware. For technology enterprises in Vietnam and globally that depend on Nvidia's infrastructure roadmap, this delay may force them to recalculate their data center investment plans and consider alternative solutions from AMD or custom cloud services to optimize operational costs in the coming years.