Starting last month, there have been successive reports that NVIDIA is preparing to launch a 24GB version of the RTX 5080, increasing the current 16GB of video memory to 24GB, and it may even directly replace the currently sold 16GB version. Many AI users and gamers have already begun looking forward to it going on sale, and the market has consistently said it could arrive as early as December this year or as late as January next year.
But this timeframe will likely not be met. According to the latest supply chain information obtained by BenchLife, as of the end of October, NVIDIA still had not provided the relevant design data to partner graphics card manufacturers; combined with PCB material preparation and shortages of other components, it will be extremely difficult for the RTX 5080 24GB to officially launch in January 2027.

Image source:Wccftech
RTX 5080 24GB probably won’t arrive by January next year! NVIDIA hasn’t submitted the design data yet, and PCB and components are also held up.
BenchLife An earlier report noted that NVIDIA had originally expected to provide the RTX 5080 24GB Design Kits to AIC partners in October, but as of late October they had still not been delivered. The report also mentioned that although word of a January 2027 launch had already been circulating within the industry since September, given current progress, the probability of a January launch is very low.
As for this card’s launch timeline, several different accounts have already emerged over the past month or so:
- Around October 10–11: Leaker hongxing2020 and others reported that NVIDIA will stop supplying the RTX 5090 and will simultaneously launch a 24GB version of the RTX 5080, replacing the existing 16GB version.
- October 12–13: VideoCardz reported that multiple AIC partners were verbally told by NVIDIA that the relevant information would be “provided soon,” but no exact date was given; the new card is expected to launch in Q1 2027.
- October 30: Chinese channel ChannelGate reports that NVIDIA will stop supplying the Chinese market with the RTX 5090D V2 and RTX 5080 16GB. Inventory of the 16GB version is enough for about two more months of sales, while the 24GB version is being pushed out faster, launching as early as December and no later than January; after that, it will be the only flagship graphics card in the Chinese market.
- November 2: The latest news from BenchLife indicates that the design data had still not been delivered by the end of October, making a January launch highly unlikely. As for the discontinuation of the RTX 5090 D V2 mentioned by ChannelGate, BenchLife added that this product had actually been out of stock for a long time.

Why would graphics cards be delayed without design data? Because Design Kits are like the architectural blueprints you get before building a house: they contain the specifications, reference designs, and related technical data for the new product. Graphics card manufacturers must first receive this data before they can start designing their own circuit boards, selecting components, and performing validation, and only then enter mass production.
Although the GPU itself hasn’t changed much, the memory capacity has changed, so PCB routing, power delivery design, cooling, memory compatibility, and BIOS firmware may all need to be revalidated.

Image source:Wccftech
In addition, BenchLife also explicitly mentioned that the current PCB stocking situation and shortages of other components are all obstacles to launching new cards. This means that even if NVIDIA hands over the design data quickly, partners still have to handle subsequent ordering, stocking, and production line scheduling; from receiving the data to getting the product onto shelves, there will still be a period of time in between.
As for how long we’ll have to wait? That’s still unknown—it looks like there are still several months to wait.
| Specifications | RTX 5080 16GB (current) | RTX 5080 24GB (rumored) | RTX 5080 SUPER (rumored) |
|---|---|---|---|
| video memory | 16GB GDDR7 | 24GB GDDR7 | 24GB GDDR7 |
| Memory allocation | 8 pieces of 2GB | 8 units of 3GB (estimated) | 8x 3GB |
| Memory interface | 256-bit | Rumored to remain 256-bit | 256-bit |
| CUDA cores | 10,752 | Not yet confirmed | 10,752 |
| Memory bandwidth | 960 GB/s | Not yet confirmed | 1,024 GB/s |
| Graphics card power consumption | 360W | Not yet confirmed | 415W |
| Suggested retail price | Starting at $999 | Not yet announced | Not yet announced |
| Status | Launched in January 2025 | Not yet published | Reportedly postponed |
Source: KOCPC Chinese