Key takeaways
- KeyBanc’s John Vinh expects Nvidia to be able to reduce 2026 Rubin GPU manufacturing to 1.5 million units from the initial target of 2 million units.
- High-bandwidth memory supply constraints from SK Hynix and Micron Technology result in lower expected production.
- KeyBanc maintains its Overweight stance with a $275 price target on NVDA shares.
- Nvidia CEO Jensen Huang confirms that Vera Rubin AI servers have entered “full production,” targeting a launch in the second half of 2026.
- Nvidia controls roughly 90% of AI accelerator spending and about 85% of the entire AI chip sector.
According to a new analysis from KeyBanc, it appears that Nvidia is ready to trim its manufacturing targets for its upcoming Rubin GPUs. The financial research firm expects actual production volumes to stabilize near 1.5 million chips in the current calendar year, representing a 25% decline from the originally planned target of 2 million units.
What’s driving this potential shortage? The answer lies in the limited availability of high-bandwidth memory components. Industry suppliers SK Hynix and Micron Technology have struggled to supply sufficient quantities of the advanced memory modules necessary for Rubin GPU functionality, resulting in the expected production gap.
John Vinh, the KeyBanc analyst behind the research note, highlighted these supply chain concerns while maintaining a constructive outlook. His Overweight recommendation for the stock remains unchanged, as does his $275 valuation target – which represents a significant upside from current trading levels.
The Rubin GPU serves as the basis for Nvidia The upcoming Vera Rubin AI server platform, which CEO Jensen Huang recently confirmed has reached “full production” status. The company expects to begin commercial shipments during the latter half of 2026.
These next generation servers represent a huge leap in performance. Vera Rubin systems are expected to deliver about 3.3 times faster performance than Blackwell Ultra – Nvidia’s current flagship offering. The architecture combines Rubin GPUs and complementary Vera CPUs.
Nvidia had not provided a statement in response to inquiries at the time of reporting.
The market continues to be led by supply headwinds
Despite these manufacturing challenges, Nvidia’s leadership position in AI semiconductors remains intact. The chipmaker accounts for nearly 90% of spending on AI accelerators and maintains control of nearly 85% of the total AI chip market.
Industry forecasts indicate that technology leaders will devote between $600 billion and $700 billion to AI data center infrastructure throughout 2026 — an investment wave that makes Nvidia the main beneficiary among semiconductor manufacturers.
During the most recent financial reporting period, Nvidia Achieved 75% year-on-year revenue expansion. The company’s first-quarter forecast exceeded analyst expectations by $5 billion, indicating continued growth momentum approaching 77%.
Hedge fund veteran Ken Griffin maintains a roughly $4 billion stake in Nvidia stock, representing his largest equity stake based on regulatory disclosures.
Expanding revenue sources and emerging opportunities
Beyond its core hardware operations, Nvidia has been systematically developing its software capabilities. The AI Enterprise platform is expected to achieve profit margins exceeding 80% and could generate $10 billion in annual revenue by 2027.
Physical AI applications — which include robotics, self-driving vehicles, and human manufacturing systems — constitute an additional growth cycle for devices that market watchers believe is still in its early stages.
Pre-market trading activity on Monday showed modest gains in line with broader market trends. S&P 500 futures rose 0.1%, while Dow Jones futures recorded minimal movement.
Broadcom (AVGO) shares rose 0.5% in pre-market trading, and Advanced Micro Devices (AMD) shares rose 0.7% during the same period.
