Nvidia is going to dominate with the RTX 50 Blackwell GPUs

Nvidia has finally unveiled the new GeForce 50 series, along with DLSS 4 technology. Four high-end GPUs have been announced: the RTX 5080 and RTX 5090 will be launched first from January 30th.

---

After months of leaks and rumors, the next-gen Blackwell RTX GPUs are now official, and four models are on the way. Nvidia CEO Jensen Huang unveiled the RTX 50 series GPUs during his keynote at CES 2025.

The RTX 50 series GPUs feature a new design for the Founders Edition, with just two dual-stream fans, a 3D vapor chamber, and GDDR7 memory. All RTX 50 series cards are PCIe Gen 5 compatible and include DisplayPort 2.1b connectors to support up to 8K and 165Hz displays.

RTX 5090

The new RTX 5090 offers significantly improved hardware compared to its predecessor. Not only does this GPU use Nvidia's new Blackwell architecture, but it also integrates a significantly larger number of CUDA cores, higher memory bandwidth, and increased VRAM capacity.

The number of Streaming Multiprocessors (SMs) increased from 128 with the RTX 4090 to a whopping 170 with the RTX 5090, marking a 33% increase in core size. However, GPU boost frequencies have decreased slightly, from 2.5 GHz to 2.4 GHz.

The memory subsystem has been completely revamped, now equipped with GDDR7 technology on a 512-bit bus. With GDDR7 memory set to 28 Gbps, memory bandwidth reaches 1,792 GB/s, which is an almost 80% increase over the RTX 4090. The GPU also includes 32 GB of VRAM.

The RTX 5090 requires 575W of power, which is a significant increase of 125W over the RTX 4090. NVIDIA has priced this SKU at $1999.

The RTX 5090 Founders Edition will be a two-slot GPU and can be installed in PCs with compact form factors, thanks to the new 3D vapor chamber design and compact PCB.

RTX 5080

The RTX 5080 is equipped with the GB203 GPU with 10752 CUDA cores. NVIDIA has decided that this SKU will launch with 16 GB of GDDR7 memory on a 256-bit memory bus.

This is the only model in the RTX 50 series to integrate faster 30Gbps memory. The RTX 5080 has a TDP of 360W, which is 40W more than its predecessor. According to NVIDIA, this model will retail for $999.

RTX 5070 Ti and RTX 5070

The NVIDIA RTX 5070 Ti will feature a scaled-down GB203 GPU with 8960 CUDA cores. Memory-wise, it will have the same 16 GB of VRAM as the RTX 4070 Ti SUPER, but upgraded with new GDDR7 memory chips, offering a 78% bandwidth increase over the 4070 Ti, reaching 896 GB/s.

According to the specs page, the GeForce RTX 5070 Ti will operate at a boost frequency of 2.45 GHz.

NVIDIA also announced the RTX 5070 graphics card, which the company described as “twice as fast as the RTX 4070”. It is confirmed that this card will have 12 GB of memory and a bandwidth of 572 GB/s. This model will be equipped with 6144 CUDA cores and will be sold at a price of $549.

The RTX 5070 Ti will not have a Founders Edition design; it will be launched exclusively as a model by NVIDIA partners, while it would appear that the RTX 5070 will feature a Founders Edition design.

NVIDIA confirms that the RTX 5070 series will launch in February. Both the RTX 5080 and RTX 5090, on the other hand, will be launched on January 30.

DLSS 4, the next-generation upscaling technology

Nvidia also announced DLSS 4, the next generation of their upscaling technology. The main feature is the new frame generation, which allows you to generate 3 frames for every 1 real frame, instead of generating 1 frame for every 1 real frame.

In addition, DLSS 4 represents what the company calls the “biggest update to its AI models” since the release of DLSS 2. DLSS Ray Reconstruction, DLSS Super Resolution, and DLAA will now be powered by the same advanced architecture that powers AI models like ChatGPT and Google's Gemini.

The company says this results in greater time stability, less ghosting, and higher detail of moving objects.

A total of 75 games and apps will support DLSS 4 from day one. When the new RTX 50 cards come out, games like Alan Wake and Cyberpunk 2077 will be updated with the ability to take advantage of the Multi Frame Generation feature.

GeForce RTX 40 series GPUs will not receive Multi Frame Generation, but they will get DLSS 4's advanced frame generation, advanced reflection reconstruction, super resolution, and deep learning antialiasing capabilities.