NVIDIA's revenue grew by 262% in the last quarter, exceeding investors' high expectations and pushing the company's stock up in trading. NVIDIA also announced a 10-for-1 stock split effective June 7, and reported a 150% increase in quarterly dividends.
Revenue for the quarter reached $26 billion, higher than previous estimates of $24.7 billion, thanks to record sales of artificial intelligence chips. Revenue of around $28 billion is expected for the current quarter. NVIDIA's shares, which have continued to soar by over 90% since the beginning of the year, rose by 2% in a day.
There has been a rapid increase in demand for NVIDIA's data center graphics processing units with artificial intelligence over the past year, as major technology companies rush to develop the computational infrastructure needed to create powerful new AI products. Google, Microsoft, Meta, and Amazon have indicated that their spending on this will remain high until 2024.
- NVIDIA will now develop new chips annually, rather than every two years. This was announced by the company's CEO Jensen Huang. NVIDIA will also accelerate the development of any other chips it produces to match the annual pace—this could include consumer graphics cards as well
“I can announce there is one more chip after Blackwell. We will work on a one-year cadence,” Huang noted.
Prior to this, NVIDIA was creating a new architecture roughly every two years—e.g., Ampere was introduced in 2020, Hopper in 2022, and Blackwell in 2024. Recently, analyst Ming-Chi Kuo reported that the next Rubin architecture will appear in 2025 with the R100 graphics processor, Huang's comments indirectly confirming this possibility.
“New processors, new graphics processors, new network adapters, new switches… there will be a whole range of microchips”.
When the analyst asked Huang to explain how the Blackwell graphics processors would evolve while the Hopper graphics processors are still selling well, Huang explained that the new generations of NVIDIA chips are electrically and mechanically compatible with previous versions and work with the same software. According to him, customers "will easily transition from H100 to H200 and to B100" in existing data centers.
Sources: Financial Times, The Verge
Comments (0)
There are no comments for now