NVIDIA Is Ready To Redefine 'Chip Power' Next Year With Its Upcoming RTX 50 Series And B300 Chip

On the cusp of 2025, the global chip industry is eagerly awaiting the start of a new round of "arms race"—industry leader NVIDIA is set to rewrite the upper limits of chip computing power in the coming months.

For NVIDIA's stock price, which has been volatile since the end of October, the importance of the new flagship product launch is also self-evident.

On January 7th next year, Jensen Huang will unveil the RTX 50 series graphics cards based on the BlackWell architecture at the CES opening ceremony in Las Vegas. The first graphics cards to be released include the 5090 and 5080, with the 5070 Ti and 5070 also possibly making an appearance, although their release and market launch will be later.

According to previous leaks, the GB202 chip used by the RTX 5090 graphics card will have an area of 744 square millimeters, an increase of 22% compared to the AD102 of the 4090. The 16 solder pads around the GPU also correspond to the 32GB of memory mentioned in the leaks.

Considering the bandwidth improvement brought by the new generation of GDDR7 memory, the industry is paying attention to the performance of this graphics card under high load conditions, such as 4K, 8K gaming, artificial intelligence applications, professional content creation, etc.

Looking at the trend over the past two years, NVIDIA is more inclined to pile up materials on the 90 series to challenge the limits of performance. Leaks show that the RTX 5090 graphics card will have 21,760 CUDA cores, while the 5080 graphics card will have a configuration of 10,752 CUDA cores and 16GB GDDR7 memory.

The PCB design also shows that the new graphics card uses a 12V-2x6 power connector that supports up to 600W of power, replacing the previous generation's 12VHPWR interface. Previously, after the release of the 4090, some users reported that the power interface was burned or melted.

While AI giants are still waiting for the delivery of B200 servers, NVIDIA's next-generation AI flagship chip is also quietly approaching.

According to the latest leaks, at the GTC 2025 in late March next year, NVIDIA will release the B300 chip and the corresponding GB300 server platform. The B300 is the upgraded version previously known as Blackwell Ultra.

Compared with the previous generation B200 chip, the design power consumption (TDP) of the B300 will increase from 1000W to 1400W. This is accompanied by a comprehensive upgrade in architecture, configuration, and performance.

The most obvious parameter upgrade of the GB300 is the memory. Compared to the 8-layer stacked 192GB configuration used by the GB200, the new generation of AI servers will use a 12-layer stacked 288GB of HBM3e.

In contrast, the main competitor AMD recently released the MI325X with 256GB of memory, and the expected MI350X to be released in the second half of next year can provide parameters similar to the GB300.

Other upgrades to the GB300 include the following: the network card will use ConnectX 8, the optical module will be upgraded from 800G to 1.6T, and the cooling system will also be redesigned to include more advanced water cooling plates. The cabinet will come standard with a capacitive tray and an optional battery backup unit system. Initial leaks show that the upgrade of the Ultra architecture will bring a 1.5 times FP4 performance improvement per card. Additionally, the server computation board of the new platform will use LPCAMM memory modules for the first time.