Nvidia’s Rubin Road Gets Bumpy: Ceramic Capacitor Squeeze and Kyber Delay Test Pricing Might

Nvidia's Rubin platform faces ceramic capacitor shortages and a delay in its high-end Kyber rack, raising supply chain risks despite strong financials and a low forward P/E of 18.

Nvidia’s dominance in artificial-intelligence chips has long been built on a simple formula: unrivaled performance, soaring prices, and a captive customer base. As the company prepares to roll out its next-generation Rubin platform, that formula is facing its most tangible stress test yet — not from rival chipmakers, but from the mundane physics of ceramic capacitors and the sliding timeline for its most powerful rack system.

The stock has been drifting. After touching a record high of €202.50 earlier this year, shares have slipped to around €178, lately oscillating between €176.54 and €178.68. The forward price-to-earnings ratio has fallen to an historic low of 18, a level that has some analysts arguing the stock is deeply undervalued even as supply-chain headaches mount.

The Component Crunch

Rubin’s launch is scheduled for the second half of next year, with the first systems destined for eight major cloud partners. But the sheer complexity of the hardware is creating a new bottleneck. Each Rubin rack requires nearly 600,000 multilayer ceramic capacitors — a staggering volume compared with standard servers. Spot prices for these specialized components have surged up to 60% since the winter, and availability remains tight.

The cost impact on Nvidia is modest in absolute terms. High-bandwidth memory adds only €200,000–€300,000 per rack, and the company’s pricing power is formidable: a Rubin rack is expected to command €6–€7 million, up from a few million for the previous Blackwell generation. That pricing leverage should allow Nvidia to hold gross margins near 70%. But the risk lies not in the bill of materials, but in the logistics. If capacitor shortages force production delays or lower volumes, the revenue implications could be severe.

A High-End Rack Slips

Adding to the uncertainty, the most powerful variant of the Rubin family — the Kyber NVL144 rack, part of the Rubin Ultra line — is reportedly facing a significant delay. According to one source, the Kyber system may not ship until 2028, a delay of more than 12 months. Nvidia’s management has denied that the regular Rubin platform is behind schedule, emphasizing that the overall AI roadmap remains on track. Yet the gap left by a missing high-end option opens a window for competitors.

AMD, with its MI500X, Google’s TPUv8i, and Huawei’s Atlas 350 are all targeting the inference market that Nvidia currently dominates with an 80% share. If cloud giants cannot access the Kyber racks, they may accelerate qualification of these alternatives.

Financial Firepower Versus Physical Limits

Nvidia’s financial position remains extraordinarily strong. Operating cash flow exceeded $50 billion in the fiscal first quarter, and the company has a three-digit-billion-dollar stock buyback authorization. Revenue in the current quarter is expected to nearly double year-over-year. Yet the company’s supply-chain commitments — $119 billion in outstanding obligations — underscore the risk that hardware built for promised orders could sit idle if data-center construction slows.

Power constraints are an additional wild card. In the first quarter of 2026, regulators and grid operators blocked AI projects worth over $130 billion globally. If cloud providers can’t install the hardware fast enough, Nvidia could face an inventory build-up.

Bullish Case: Undervalued Pricing Power

Optimists point to the historic P/E trough and the accelerating value of each new system. The Rubin GPU itself packs 336 billion transistors, up from 208 billion in Blackwell, promising sharply lower inference costs and faster training with fewer compute cores. Analysts at Bank of America emphasize Nvidia’s unmatched ability to raise prices, and the average analyst target of €263.69 implies nearly 50% upside from current levels.

A potential loosening of export restrictions on the H200 chip for China could also provide an incremental revenue boost, though that remains speculative.

Bearish Case: Execution Risk and Competition

The bears see a different picture. Beyond capacitor shortages, the Kyber delay hands rivals a rare multiyear window. AMD, Google, and even Huawei are pushing specialized chips for the inference segment, where Nvidia’s absolute performance advantage is less critical. Meanwhile, rising component costs could eat into margins, even with pricing power.

On the technical side, the stock has slipped below its 50-day moving average of roughly €181 and is testing support near the 200-day line at €164.65. A smooth Rubin rollout could lift the stock back toward the 50-day average, but any sign of shipment delays could trigger a retest of the 52-week low of €138.98.

What’s Next

All eyes are on Nvidia’s fiscal second-quarter earnings, scheduled for August 26. The market is expecting revenue of nearly $92 billion. That report will provide hard data on whether supply constraints are denting the growth trajectory. For now, Nvidia’s pricing power remains the strongest argument in its favor — but the physical and logistical hurdles surrounding Rubin are proving that even the most dominant chipmaker cannot simply buy its way around a shortage of ceramic capacitors.