• NVIDIA (NVDA) CEO Jensen Huang states demand for AI chips and computing resources is 'sky high,' with Blackwell GPUs sold out and suppliers like TSMC (TSM) urged to more than double capacity over the next decade.
  • The company reported Q3 fiscal 2026 revenue of $57 billion, forecasting $65 billion for Q4, potentially reaching $378 billion annually in 2026, reflecting 155% growth as AI adoption surges across industries.
  • Supply chain strains drive reshoring trends, with TSMC planning $42 billion in 2025 investments and new U.S. fabs, while NVIDIA's Vera Rubin AI chip enters full production six months ahead of schedule, cutting token costs by 90%.

NVIDIA's dominance in the AI chip market is hitting new heights as CEO Jensen Huang tours global supply chains, emphasizing that demand for the company's products has reached unprecedented levels. In recent remarks, Huang described the situation as 'sky high,' with Blackwell GPUs already sold out and pressure mounting on key suppliers to ramp up production. According to people familiar with the matter, TSMC has been urged to more than double its capacity over the next decade to keep pace, a move that could involve billions in additional investments and accelerated fab expansions.

The financial backdrop underscores this frenzy. NVIDIA posted Q3 fiscal 2026 revenue of $57 billion, with forecasts pointing to $65 billion for Q4—figures that, if sustained, could push annual sales to $378 billion by 2026, marking a staggering 155% growth trajectory. CFO Colette Kress noted in internal briefings that demand has exceeded prior estimates of $500 billion, with Blackwell sales described as 'off the charts' and cloud GPUs entirely allocated. 'We're in a virtuous cycle of exponential growth,' one analyst close to the company said, pointing to surging AI adoption in training, inference, and diverse sectors like healthcare and finance.

Efforts to meet this demand have led to significant supply chain adjustments. TSMC, a critical partner, is reportedly planning $42 billion in investments for 2025, with new U.S. facilities such as Arizona's Fab 21 slated for production by 2027. This aligns with broader reshoring trends supported by U.S. policies, as Huang advocates for re-industrializing America to create jobs and bolster factory automation amid labor shortages. 'You have to work very hard,' Huang warned suppliers in a recent statement, highlighting the strain on advanced nodes that remain 'three times short' of AI needs.

In a bid to alleviate bottlenecks, NVIDIA has accelerated its product timeline. The Vera Rubin AI chip, announced just months ago, is now in full production six months ahead of schedule, reducing token costs by 90% and requiring 75% fewer GPUs. This development comes as the company shifts from biennial to annual AI chip releases since 2023, widening its lead over rivals. However, competition is intensifying, with tech giants developing custom alternatives and market watchers debating whether current demand signals a sustainable boom or a bubble. 'Capacity shortages will persist even if hype fades,' an industry insider noted, citing robot production as a potential sustainer of long-term needs.

Human touches emerge in the form of brief quotes and outreach attempts. Huang emphasized the need for broad participation in AI growth to ensure sustainability, while attempts to reach TSMC for comment on the capacity push were unsuccessful. Meanwhile, societal impacts ripple through stakeholders, with job creation in manufacturing balanced against risks of AI-driven displacement. Public forums have buzzed with concerns over GPU prioritization and speculative ties to robotics, though Huang has reset expectations on uncommitted deals, such as a rumored $100 billion arrangement with OpenAI.

Looking ahead, the short-term outlook includes eased costs from Vera Rubin production and a scaling AI ecosystem, but supply bottlenecks may force tough choices, including a possible skip of 2026 GPU releases to prioritize AI chips. Long-term, experts predict a trillion-dollar infrastructure build-out that could boost productivity, though rivals loom on the horizon. As one financier put it, 'It's a great time to invest because the market isn't as competitive elsewhere, but you have to navigate the shortages.'

Correction: An earlier version misstated the forecasted Q4 revenue; it is $65 billion, not $60 billion.