Nvidia's Kyber NVL144 Delayed to 2028, Asian Suppliers Face Market Shake-Up
Nvidia’s highly anticipated Kyber NVL144 AI server rack, designed to power next-gen data centers, has been pushed back to 2028, marking a significant setback for the chipmaker. Analysts at SemiAnalysis report that manufacturing challenges with circuit boards have caused the delay, which now spans over a year. This postponement has ripple effects across the semiconductor industry, particularly for Asian suppliers, who have seen market valuations dip by double-digit percentages. The delay also signals the cancellation of the more advanced Rubin Ultra variant, further complicating Nvidia’s roadmap.
Manufacturing Hurdles Delay Nvidia's AI Server Launch
The Kyber NVL144 was initially slated for a 2026 release, positioning Nvidia to dominate the AI server market with its next-generation architecture. However, technical snags in producing the custom circuit boards have stalled progress. While Nvidia has not disclosed specifics, industry insiders suggest issues with scaling advanced packaging technologies are at the core of the delay. This bottleneck underscores the complexities of manufacturing high-density AI hardware, where even minor design flaws can lead to costly delays.
Asian Suppliers Face Market Reckoning
The delay has disproportionately affected Asian semiconductor suppliers, many of whom had ramped up production to meet Nvidia’s initial timeline. Companies like TSMC and Samsung Semiconductor, which supply critical components for the Kyber NVL144, have seen their stock prices fluctuate amid uncertainty. Analysts warn that without a clear recovery plan, these firms could lose market share to competitors capitalizing on Nvidia’s missteps. The financial impact highlights the fragile interdependence between chipmakers and their foundry partners in the AI hardware race.
Competitors Seize the Opportunity
With Nvidia’s delay, rivals like AMD and Google are positioning themselves to capture market momentum. AMD’s Instinct MI300A accelerator and Google’s Tensor Processing Units (TPUs) are already gaining traction in data centers, offering comparable performance at lower costs. The cancellation of the Rubin Ultra variant further weakens Nvidia’s ability to counter these moves. While the delay is a blow to Nvidia’s strategic plans, it also opens a window for competitors to solidify their footholds in the AI infrastructure space.
As the industry watches Nvidia navigate its challenges, the broader implications for supply chain dynamics and competitive innovation remain uncertain.
Source: The Decoder. AI-assisted editorial synthesis — TechnoExpress.