Nothing stops Nvidia anymore. Driven by financial results that are still exceptional – the green giant has just completed a quarter with 14 billion dollars in profits —, the firm has the firm intention of pressing on the mushroom to further accelerate its pace of innovation. Based on comments from CEO Jensen Huang, who recently spoke during a conference call with shareholders, Nvidia now appears to want to produce a new generation of chips every year. « I can announce that after Blackwell there will be a new chip. We are on a one-year pace », cite The Verge.
A not innocuous little sentence. Because until now, the company had a tendency tounveil new architectures approximately every two yearswithout interruption since the generation Fermi (GeForce 400 / 500) in 2010. They were followed by cards built on the architecture Kepler (GeForce 600, 700 & 800M) in 2012, then Maxwell (GeForce 900) and 2014, Pascal (GTX 10X0, Quadro P) en 2016, Turing (RTX 20X0) en 2018, Ampere (RTX 30X0) in 2020, and There’s Lovelace (RTX 40X0) as well asHopper (H100/H200) and 2022.
With the deployment of Blackwell architecture, which is expected within a few months, this tradition which dates back more than a decade should therefore be honored once again. But from there, you should expect to see brand new hardware arrive even more frequently. Earlier this month, Ming-Chi Kuo, one of the most renowned analysts in all of tech, had announced that a new Rubin architecture would arrive as early as 2025. And Huang’s new statements indeed seem to go in this direction.
Nvidia remains focused on AI
Be careful, though; This does not mean that gamers will be able to benefit from a new generation of graphics cards specialized in gaming every year. Because if the next RTX 50X0 should indeed be based on the new Blackwell architecture, which made a very strong impression when it was presented last March, this will not necessarily be the case for the following ones. With this boost, Nvidia seeks above all to continue its crazy race in the field of AI hardware.
The Rubin architecture, for example, should mark the arrival of a new AI R100 GPU. An initiative that is surprising. After all, at present, H100 (Hopper architecture) remain the absolute references in this field and still sell like hot cakes. Tesla, for example, recently purchased no less than 35,000 to drive the autopilot system of its vehicles. We can also mention Meta. In January, Mark Zuckerberg announced his intention to acquire more than 100,000 H100 GPUs by the end of the year, to reach a total of more than 350,000 units.
![Nvidia Gpu H100](https://www.journaldugeek.com/app/uploads/2023/12/nvidia-gpu-h100.jpg)
A titan that never stops growing
Moreover, Nvidia is banking heavily on this sector to continue its incredible growth (it became the third richest company in the world last March, surpassing the symbolic threshold of $2,000 billion in market capitalization).
According to the financial director of the firm cited by The Verge, automotive will be Nvidia’s primary source of growth this year. The other goose that lays the golden eggs is web giants like Meta. There is little doubt that these titans of industry will also rush to the next models. The road seems clear for the B100/B200 GPUs which will arrive with the deployment of Blackwell. And the trend will probably be the same with the R100s from next year.
With this strategy, Nvidia will be able to consolidate its position as undisputed leader. As a reminder, today it provides more than 80% of chips used by other tech giants in the field of AI. Enough to strengthen a quasi-monopoly which does not seem about to collapse. At this rate, the company could even end up overtaking Apple, still firmly in 2nd position with 2.7 trillion in market capitalization, within a few years.
🟣 To not miss any news on the WorldOfSoftware, subscribe on Google News. And if you love us, .