Whether you were ready for it or not, the new year has arrived, which also means it’s time for the annual Consumer Electronics Show.
While there will be a plethora of announcements for all sorts of new technology that we just can’t wait to get our hands on, arguably, the biggest announcement is the one from Nvidia showing off the latest 50 series GPUs, which we can expect to hit the market towards the end of January.
The announcement from Nvidia showed off four new cards in the 50-series range: the RTX 5070, the RTX 5070 Ti, the RTX 5080 and the RTX 5090.
The complete 50-series range comes with Nvidia’s new microchip architecture, Blackwell, which is said to boast up to 125 Shader Teraflops, 380 RT (Real-time Ray Tracing) Teraflops and 4,000 AI TOPS, all backed by its 92 billion transistors. The most powerful card announced, the RTX 5090, is said only to utilise 3,400 AI TOPS, perhaps hinting at the possibility of another card down the line that could squeeze those extra TOPS out of the Blackwell chips.
The 50-series also sports the latest GDDR7 memory, an upgrade over last year’s GDDR6X, which promises more performance while increasing power efficiency.
All this new AI performance means that Nvidia can release the 50-series GPUs with its latest frame generation technology, DLSS4. During the live stream at CES, Jensen Huang said that this technology, alongside the GDDR7 memory, allows the RTX 5090 to double the performance of the RTX 4090 and makes the RTX 5070, an entry-level card, just as powerful as the RTX 4090.
Nvidia also released a few graphs showing off the 50-series cards, with the new DLSS4 technology enabled against the 40-series cards.
However, the caveat is that DLSS4 performance will only be seen in games that support it. So, while you may get double the frames in games like Cyberpunk 2077, Black Myth: Wukong and Alan Wake 2, other older games may only see a slight performance improvement.
Below is a table showcasing all the relevant specs of the 50-series cards, as well as when they should hit the market and how much each will cost.
| RTX 5070 | RTX 5070 TI | RTX 5080 | RTX 5090 | |
| Price ($) | 549 | 749 | 999 | 1,999 |
| Expected Release Date | Early February | Early February | End of January | End of January |
| Memory | 12 GB GDDR7 | 16 GB GDDR7 | 16 GB GDDR7 | 32 GB GDDR7 |
| Memory Bus | 192-bit | 256-bit | 256-bit | 512-bit |
| CUDA Cores | 6144 | 8960 | 10752 | 21760 |
| Boost Clock (GHz) | 2.51 | 2.45 | 2.62 | 2.41 |
| Base Clock (GHz) | 2.16 | 2.3 | 2.3 | 2.01 |
| Tensor Core (TOPS) | 988 | 1406 | 1801 | 3352 |
| RT Core (TFLOPS) | 94 | 133 | 171 | 318 |
| Power Draw (Watts) | 250 | 300 | 360 | 575 |
While the RTX 5090 will be released at a whopping $500 more than the RTX 4090, the rest of the series will be released at a lower cost compared to its previous generation counterpart on release day.
Will you be lucky enough to pick up one of Nvidia’s latest cards when they hit the market later this month? Or will you be on the hunt to snatch up someone’s second-hand RTX 40-series card? Let us know in the comments below.
