Nvidia’s GeForce RTX 5050 is official: 8GB, $250, coming in July | usagoldmines.com

“Affordable graphics card” seems like something of an oxymoron these days. But after six months after the launch of the RTX 5000 series, Nvidia finally has one that broadly fits into that category. The GeForce RTX 5050 is now official, in both desktop and laptop flavors. The former is an 8GB card, the same as the RTX 5060, and will land at $250 when it launches next month.

The Blackwell-series card is a notable step downward from the 5060, which only makes sense. That 8GB of memory is running at GDDR6 speed instead of GDDR7, and you get only 2560 CUDA cores versus 3840. Nvidia rates the ray tracing cores at 40 teraflops of capacity, about two-thirds of the more expensive card, with a maximum wattage of 130 and an identical 128-bit memory bus. Like all RTX 50-series cards, it can handle DLSS 4 and the associated tech.

There’s no mention of a Founder’s Edition launch, so like most mid-range and budget GPUs, expect this to be handled entirely by Nvidia’s graphics card vendor partners. You can find Nvidia-provided RTX 5050 performance claims versus the older RTX 3050 in the image above, though note that the comparisons lean heavily on DLSS 4’s Multi-Frame Generation feature. The RTX 30-series didn’t support AI Frame Generation whatsoever.

There was no RTX 4050 on the desktop. At this price it’ll compete with the 8GB version of AMD’s RX 9060 XT ($300) and Intel’s Arc B580 ($250).

Laptops equipped with a discrete RTX 5050 will also get 8GB of memory, strangely faster at GDDR7, but with the same 2560 CUDA cores. Acer has already confirmed that it will upgrade its popular Nitro V series of gaming laptops with the new card starting at $900, a bit cheaper than the $999 starting MSRP Nvidia is touting. Those will also land in July, along with offerings from other Nvidia partners.

Editor’s note: This article was updated to add Nvidia’s performance data.