Key takeaways from Day 1 of GTC:
Jensen Huang presented an even more aggressive outlook: NVIDIA expects the Blackwell and Rubin architectures to generate at least US$1 trillion in cumulative revenue by the end of 2027.
Demand for AI compute continues to surge: major technology companies are still aggressively purchasing AI GPUs and servers, and total investment in AI infrastructure could reach hundreds of billions of dollars.
Blackwell is the core growth engine: designed specifically for large-model training and inference, it is expected to become the centerpiece of the AI server procurement cycle over the next few years.
Rubin represents the next generation of compute upgrades: expected to launch around 2026, it will continue to improve both performance and energy efficiency.
A new architecture every year: NVIDIA is using rapid iteration to maintain its technological lead against AMD, Intel, and cloud providers developing their own AI chips.
AI cluster sizes continue to expand: data centers are scaling from thousands of GPUs to hundreds of thousands, and eventually even millions of GPUs.
The AI compute cycle is still in its early stages: the trillion-dollar revenue target reflects the view that AI infrastructure remains in a long-term investment phase.
Comment:
At this stage, I still believe that the AI industry as a whole does not represent a true bubble.
Unlike many past technology bubbles, today’s AI boom is supported by real capital expenditure, genuine enterprise demand, and continuously expanding investment in compute infrastructure.
Within this supply chain, NVIDIA remains the dominant “arms dealer” of the AI era. As long as companies around the world continue building AI infrastructure, NVIDIA is likely to remain the most direct beneficiary. And judging from the most recent quarter’s earnings, the company’s revenue mix is also becoming more diversified. It is no longer driven solely by one-dimensional GPU demand; NVIDIA’s influence across the broader AI ecosystem is expanding.
That said, major technology conferences like GTC are, by nature, still a stage for presenting vision. Whether it is the trillion-dollar revenue target or the future product roadmap, much of what is discussed is a picture of the future rather than something that can be immediately verified.
So whether the market truly buys into the story will still depend on actual earnings performance in the next quarter and the quarters beyond. If capital expenditure and AI demand continue to show up in reported results, then those visions may gradually turn into market consensus.
Disclaimer:
The above content reflects personal views and market discussion only. It does not constitute any investment advice or recommendation to buy or sell. Investing involves risk, and readers should make their own assessments and bear responsibility for their own decisions.