Nvidia projects $1 trillion in chip orders through 2027

During a keynote address on Monday to kick off Nvidia’s annual GTC Conference, CEO Jensen Huang announced a projection of at least $1 trillion in orders for the company’s Blackwell and Vera Rubin chip architectures through 2027. This projection highlights the scale of demand for the company’s artificial intelligence hardware. The announcement, delivered during the keynote on Monday, sets a benchmark for the company’s long-term sales expectations as it prepares to transition to its upcoming chip architectures.

The new figure follows a previous demand projection of about $500 billion for the same chip architectures through 2026. Huang emphasized the scale of these numbers during his address, noting that about $500 billion represents an enormous amount of revenue. He contrasted that previous outlook with the company’s updated expectations. “Well, I’m here to tell you that right now where I stand — a few short months after GTC DC, one year after last GTC — right here where I stand, I see through 2027, at least $1 trillion,” Huang said. Huang delivered the projection a few short months after GTC DC and one year after the previous GTC event, emphasizing that the demand outlook has shifted within that timeframe.

The upcoming Rubin computing chip architecture, which was first announced in 2024, is designed to outperform its Blackwell predecessor. Following that announcement, Nvidia officially started production of the Rubin architecture in January, and the company expects to ramp up production in the second half of the year.

According to Nvidia, the Rubin architecture outperforms Blackwell, particularly in key artificial intelligence workloads. The company’s performance metrics indicate that the architecture operates faster across different stages of AI model development and execution:

• Model-training tasks: Rubin operates 3.5x faster than Blackwell during this phase, where an AI model learns from data.
• Inference tasks: Rubin operates 5x faster than Blackwell during this phase, where an AI model executes tasks and generates outputs.
• Computing performance: The architecture’s performance capability reaches as high as 50 petaflops.

Why it matters

The massive revenue projection underscores the sustained, high-velocity demand for AI infrastructure, positioning Nvidia’s upcoming Rubin architecture as a critical driver for the company’s growth through 2027.