We’ve primarily focused on the desktop Blackwell GPUs, but Nvidia will naturally also have laptop variants. It has even announced the names and a few of the specifications, even though the actual laptops are still a month or two out. The various models are basically taking the same ‘desktop GPU minus 10’ approach as the RTX 40-series.
The top mobile GPU will be the RTX 5090, with 82 SMs and 10496 CUDA cores. That’s basically the same as the desktop 5080, minus two SMs, and the 1824 ‘AI TOPS’ indicates boost clock speeds should also be similar. Real-world clocks, with a TGP of 95–150 watts, will likely be lower. But what’s most interesting is that the 5090 laptop GPU will have 24GB of memory on a 256-bit interface, meaning it has 24Gb (3GB) chips, unlike any of the other announced Blackwell GPUs.