Record demand for NVIDIA’s next generation chips is expected

NVIDIA’s CEO Jensen Huang shared remarkable financial predictions in his presentation at the company’s annual GTC conference in San Jose, California. In the speech where technical details were predominant, the figures given regarding the extent of demand for artificial intelligence hardware came to the fore. Huang’s demand forecast, especially for chips with Blackwell and Vera Rubin architectures, became the focus of investors. According to the statements, it was stated that the total order volume for these products could reach 1 trillion dollars in the coming years. This evaluation stands out as important data showing that the growth in the artificial intelligence market continues unabated.

In one part of his speech, Huang recalled previously shared data and stated that there is a demand of approximately 500 billion dollars for Blackwell and Rubin chips by 2026. Emphasizing that this is a very high figure, Huang stated that this expectation was seriously revised upwards in a short time. However, looking at the current picture, he said that total demand could reach at least 1 trillion dollars by 2027. Behind this increase is the acceleration of global investments, especially in data centers and productive artificial intelligence systems. In addition, the increase in infrastructure spending by major technology companies is among the factors that raise demand projections.

NVIDIA Rubin architecture raises the performance bar

NVIDIA’s new generation chip architecture, codenamed Rubin, was first introduced in 2024 and it was stated by the company that it would offer a performance level beyond Blackwell. According to the technical data shared by the company, Rubin architecture can run 3.5 times faster than Blackwell in artificial intelligence model training processes. However, it is stated that the performance increase in inference operations reaches up to 5 times. On the computing power side, it is stated that it can be reached up to 50 petaflops. These values ​​​​point to a significant development that makes it possible to train large language models and complex artificial intelligence systems in a shorter time.

On the other hand, NVIDIA announced that it started production of Rubin chips as of the beginning of 2025. Despite this, mass production capacity is planned to be increased significantly in the second half of the year. This shows that supply chain and production processes are being scaled in parallel with the increase in demand. In addition, the company’s collaborations with production partners such as TSMC play a critical role in achieving high-volume production targets.

Despite the increasing competition in the field of artificial intelligence, NVIDIA’s strong position on the hardware side draws attention. Despite this, it is known that competitors such as AMD and Intel have increased their investments in similar areas. Still, it seems that NVIDIA maintains its advantage in the market, especially thanks to its CUDA ecosystem and software optimizations. In addition to all these, the intense focus of cloud service providers on NVIDIA solutions stands out as another factor supporting the company’s revenue expectations.