In a significant statement, Nvidia's CEO Jensen Huang has confirmed that the company's staggering $500 billion artificial intelligence (AI) demand outlook will remain unchanged on a quarterly basis. This projection, which covers the years 2025 and 2026, is based on concrete orders already on the company's books.
Locked-In Demand and Growing Expectations
Speaking at the Consumer Electronics Show (CES) 2026 in Las Vegas, Huang clarified that the half-a-trillion-dollar figure is not a speculative forecast. It represents locked-in business for Nvidia's upcoming Blackwell GPUs, the next-generation Vera Rubin platform, and associated networking hardware. He emphasized this point to underscore the company's confidence in its product pipeline.
Interestingly, Huang also signaled that new developments are likely to push expectations even higher. "Many new developments should increase our expectation," he stated in an interview, pointing to potential upside beyond the already massive headline number.
CFO Reveals Figure Has Already Expanded
Nvidia's Chief Financial Officer, Colette Kress, reinforced the CEO's bullish stance with a crucial update. She revealed that the $500 billion visibility has already grown since the company's GTC conference in October 2025. "So yes, that $500 billion has definitely gotten larger," Kress confirmed.
This growth is driven by early orders for the yet-to-be-launched Rubin chips, as major customers plan their annual volumes. The customer base securing these future platforms includes:
- Major global cloud service providers
- Leading AI model developers
- Emerging neocloud companies
Open-Source AI: A Surprise Demand Driver
Huang highlighted an unexpected engine fueling the insatiable demand for Nvidia's computing power: the rapid rise of open-source AI models. He noted that models like DeepSeek, Qwen, and Meta's Llama are now responsible for generating approximately one out of every four tokens processed.
Contrary to some predictions, this proliferation has not reduced overall computing needs. Instead, it has dramatically expanded total usage, creating more demand for the high-performance chips that Nvidia specializes in. This trend is further solidified by major industry partnerships, such as the $15 billion strategic collaboration between Microsoft, Nvidia, and AI startup Anthropic announced in November 2025.
Inside Nvidia's Next-Gen Rubin Platform
At the heart of this future demand is Nvidia's announced Rubin platform, the successor to the current Blackwell architecture. CEO Jensen Huang claims the Rubin chips will deliver five times more performance.
The Vera Rubin represents Nvidia's first six-chip AI platform built using an "extreme codesign" methodology, where multiple components are developed in unison to optimize performance. The integrated platform includes:
- Rubin GPUs offering up to 50 petaflops of inference performance
- Vera CPUs designed for data movement and AI agent tasks
- NVLink 6 networking and Spectrum-X Ethernet Photonics
- ConnectX-9 networking cards and BlueField-4 data processing units
Nvidia states this holistic design aims to minimize bottlenecks and significantly improve efficiency for large-scale AI training and inference workloads, cementing its position at the forefront of the AI hardware race.
