Throughout its CES 2025 keynote, Nvidia gave us our first trustworthy take a look at its upcoming GeForce RTX 50-series graphics playing cards. Although these new GPUs, based mostly on Nvidia’s cutting-edge “Blackwell” structure, had been extremely anticipated, the extent to which these new GPUs surpass their predecessors, at the least on paper, remains to be spectacular. For instance: The GeForce RTX 5070, at $549, will reportedly have the ability to match the efficiency of the corporate’s older GeForce RTX 4090, which launched at $1,599, marking an astonishing leap ahead in efficiency if it pans out in testing.
Nvidia’s GeForce RTX Blackwell Launch Lineup
For starters, Nvidia advised us that at the least 4 new graphics chips are due within the subsequent few months: the Nvidia GeForce RTX 5090, the GeForce RTX 5080, the RTX 5070 Ti, and the RTX 5070. At launch, Nvidia has not indicated that it’ll launch an Nvidia GeForce RTX 5060 or Nvidia GeForce RTX 5060 Ti, however these ranges of card are typical, too, so we’ll very possible see one or each of these graphics playing cards later.
(Credit score: Nvidia)
On the CES keynote, we didn’t get data on fastened shader counts, or any conventional efficiency numbers (like body charges), for any of those graphics playing cards. As a substitute, Nvidia targeted on AI efficiency and supplied some pace data for that metric. The GeForce RTX 5090, which is the quickest mannequin and can characteristic GDDR7 reminiscence, can reportedly course of 3,352 AI TOPS (trillion operations per second) and can comprise 92 billion transistors, making it presumably the biggest single chip by way of transistor rely ever produced. We don’t know the transistor rely for the Nvidia GeForce RTX 5080, however its AI processing energy is rated at 1,801 TOPS.
(Credit score: Nvidia)
This means that the RTX 5080 possible has roughly 50% of the sources of the RTX 5090, marking a steep step down. The worth of those two GPUs additionally helps this, because the RTX 5090 is priced at $1,999, whereas the RTX 5080 is priced at $999.
(Credit score: Nvidia)
The GeForce RTX 5070 Ti, which prices $749, reportedly can course of 1,406 AI TOPs. The Nvidia GeForce RTX 5070, at $549, has the facility to deal with 988 AI TOPS.
Blackwell Conjures Up Nvidia DLSS 4 & AI Processing
The costs Nvidia set on its new GPUs weren’t all that shocking, as they fall roughly consistent with these set by Nvidia on its present GeForce RTX 40-series GPUs. Much more startling was Nvidia’s declare that its new GeForce RTX 5070 would have the ability to match the RTX 4090 in gaming efficiency. That is surprising not solely as a result of the RTX 4090 is presently the world’s quickest single graphics chip, but additionally because of the vital worth distinction between the 2, with the RTX 4090 having an MSRP of $1,599 and ceaselessly promoting for greater than that.
A detailed take a look at the remainder of Nvidia’s keynote offers us some insights into how that is potential, and it’s all all the way down to AI. The AI {hardware} on these new GPUs is cutting-edge and pushed to offer far higher efficiency than earlier generations may deal with.
“Blackwell, the engine of AI, has arrived for PC players, builders, and creatives,” stated Jensen Huang, founder and CEO of Nvidia. “Fusing AI-driven neural rendering and ray tracing, Blackwell is probably the most vital laptop graphics innovation since we launched programmable shading 25 years in the past.”
(Credit score: Nvidia)
In the course of the keynote, Huang acknowledged that for considered one of its demonstrations, the GPU solely needed to render 2 million pixels in a scene that contained 33 million pixels, with the entire remaining pixels generated by AI. Two key applied sciences make this potential. One is Nvidia’s new DLSS 4 know-how, which takes body era to a brand new stage by producing extra frames based mostly on fewer pixels. It additionally makes use of ray reconstruction and super-resolution fashions to do that with higher picture constancy than earlier generations of DLSS.
Nvidia additionally launched a know-how known as RTX Neural Shaders, which equally helps generate higher-quality photographs akin to 3D character fashions. By making the most of these applied sciences and the highly effective AI {hardware}, that is possible how the RTX 5070 can match the RTX 4090 in video games. Nonetheless, on the similar time, it’s possible that the Nvidia GeForce RTX 5070 gained’t have the ability to match the RTX 4090 in video games that don’t help DLSS, which makes help for this characteristic critically essential.
Really helpful by Our Editors
The RTX 50 Sequence: Transport Quickly
In case you’re desperate to get your palms on considered one of Nvidia’s new RTX 50-series GPUs, you gained’t have to attend for much longer. The high-end RTX 5090 and RTX 5080 will likely be out there on January 30, with the GeForce RTX 5070 Ti and RTX 5070 to comply with in February.
(Credit score: Nvidia)
Nvidia may even have cellular variations of those graphics chips coming simply barely later in March. The cellular variations of those chips are listed as having much less total AI efficiency, possible as a result of their energy consumption and clocks had been lowered to work higher inside a cellular platform.
Laptops with an Nvidia GeForce RTX 5090 will begin at $2,899, whereas techniques with an RTX 5080 will begin at $2,199. Notebooks geared up with the RTX 5070 Ti are priced at $1,599, and PCs with the RTX 5070 are comparatively inexpensive, beginning at $1,299.
However, in fact, if you need the utmost energy, that may reside in Nvidia’s two-grand desktop GPU behemoth, the GeForce RTX 5090. Keep tuned for our full evaluation of the brand new lineup within the coming weeks.
Get Our Greatest Tales!
This text could comprise promoting, offers, or affiliate hyperlinks.
By clicking the button, you verify you’re 16+ and conform to our
Phrases of Use and
Privateness Coverage.
You could unsubscribe from the newsletters at any time.
About Michael Justin Allen Sexton
Analyst
Learn the newest from Michael Justin Allen Sexton