Nvidia has introduced the RTX 50-series, codename Blackwell, at CES 2025. The new playing cards include large enhancements to AI acceleration, with CEO Jensen Huang claiming “AI is coming residence to GeForce”.
CEO Huang introduced all the lineup at CES 2025, from the RTX 5090 to the RTX 5070. Let’s get proper into costs:
- RTX 5090 — $1,999
- RTX 5080 — $999
- RTX 5070 Ti — $749
- RTX 5070 — $549
RTX 50-series specs
The Blackwell household will likely be led by the RTX 5090, which is a mighty beast. No surprises there. Nvidia cites as much as 125 shader TFLOPS, 380 RT TFLOPS, and 4,000 AI TOPS for the Blackwell household, presumably the highest GPU configuration with 92 billion transistors, although the RTX 5090 is famous with 3,400 AI TOPS—so somewhat under the utmost, suggesting it is not fairly utilizing the complete chip.
For comparability, the RTX 4090 has simply over 82 FP32 shader TFLOPS from simply over 76 billion transistors.
The RTX 50-series playing cards additionally function GDDR7 reminiscence, with a most of 1.8 TB/s of reminiscence bandwidth.
| Row 0 – Cell 0 | RTX 5090 | RTX 5080 | RTX 5070 Ti | RTX 5070 |
| CUDA cores | 21760 | 10752 | 8960 | 6144 |
| Boost clock (GHz) | 2.41 | 2.62 | 2.45 | 2.51 |
| Base clock (GHz) | 2.01 | 2.3 | 2.30 | 2.16 |
| Tensor core TOPS | 3352 | 1801 | 1406 | 988 |
| Ray tracing core TFLOPS | 318 | 171 | 133 | 94 |
| Memory | 32 GB GDDR7 | 16 GB GDDR7 | 16 GB GDDR7 | 12 GB GDDR7 |
| Memory bus width | 512-bit | 256-bit | 256-bit | 192-bit |
| Total Graphics Power (watts) | 575 | 360 | 300 | 250 |
| Required system energy (PSU wattage) | 1000 | 850 | 750 | 650 |
| Power connector | 1x 600 W PCIe Gen 5 OR 4x PCIe 8-pin adapter | 1x 450 W PCIe Gen 5 OR 3x PCIe 8-pin adapter | 1x 300 W PCIe Gen 5 OR 2x PCIe 8-pin adapter | 1x 300 W PCIe Gen 5 OR 2x PCIe 8-pin adapter |
| Price | $1,999 | $999 | $749 | $549 |
RTX 50-series Founders Edition design
Image 1 of 2
The Founders Edition card, which historical past tells us would be the one to purchase whereas it is obtainable, comes with two followers in a extra typical format to the previous two generations. Though that is most likely underselling the design of those playing cards considerably. It’s not your common dual-fan GPU, that is for certain.
The PCB sits centrally, with the GPU surrounded on nearly all sides by reminiscence and MOSFETs. This then has a vapor chamber hooked up with heatpipes extending in two instructions. The two followers blow air by means of these and the hooked up heatsinks.
Altogether, it seems to be to be a plucky PCB with a ton of cooling, which is likely to be helpful contemplating the RTX 5090 gobbles 575 watts.
Image 1 of 4
The RTX 5090 and RTX 5080 dimensions are 304 mm lengthy, 137 mm large, and a couple of slots. The RTX 5070 is 242 mm lengthy, 112 mm large, and a couple of slots.
There’s no Founders Edition of the 5070 Ti.
RTX 50-series Performance
As for efficiency comparisons, Nvidia did not give us a lot to go on through the stream, besides the promise that the RTX 5090 will supply ‘twice the efficiency of the 4090’ and the RTX 5070, a $549 card, will compete with the RTX 4090. Though DLSS 4 and new AI options are enjoying an enormous position right here, which additionally means the advance is likely to be extra restricted in video games that do not help these options.
In the video under, printed shortly after the announcement, Nvidia exhibits how DLSS 4 (a function referred to as Multi Frame Generation, actually) helps the RTX 5090 ship double the frames because the RTX 4090 in a Cyberpunk 2077 ray-traced benchmark.
At least Cyberpunk 2077, Alan Wake 2, and Black Myth: Wukong are confirmed to help DLSS 4 with Multi Frame Generation.
Watch On
Huang doubled down on this AI uplift for the RTX 5070: “4090 efficiency at $549,” proclaimed Huang through the present.
“Impossible with out synthetic intelligence. Impossible with out 4 teraops of AI Tensor cores. Impossible with out the G7 reminiscences.”
Here are some benchmark graphs, which Nvidia printed alongside the announcement. What’s necessary to notice is this isn’t strictly evaluating silicon right here, but in addition the brand new DLSS Multi Frame Generation function, which is enabled on the RTX 50-series.
Image 1 of 4
At the start of the stream, Nvidia confirmed off a ray-traced scene rendered (inferred?) in real-time on Blackwell, which included new AI options to massively speed up efficiency and save reminiscence.
The RTX 50-series options an AI Management Processor, new Blackwell Tensor Cores with FP4 help, DLSS 4, and Huang alluded to in-shader AI acceleration to assist unfold the workload of the brand new AI methods.
There are RTX Neural Shaders, which permit the GPU’s programmable shaders to run neural networks. Essentially, in-shader AI acceleration, alongside the Tensor Cores. These can be utilized to compress textures by as much as seven occasions. As proven within the screenshots under.
Image 1 of 2
There can also be the brand new DLSS Transformer mannequin, which can energy Ray Reconstruction, Super Resolution, and DLAA, and makes use of the identical basic structure design as AI fashions corresponding to ChatGPT or Google Gemini. Nvidia claims these can enhance temporal stability, cut back ghosting, and have increased element in movement.
Image 1 of 2
RTX 50-series availability
As for when you may get your fingers on these playing cards, here is the supply:
- RTX 5090 — obtainable January 30
- RTX 5080 — obtainable January 30
- RTX 5070 Ti — availability beginning in February
- RTX 5070 — availability beginning in February
