Nvidia announced this week that the latest version of its subscription service, GeForce Now Ultimate, has officially gone live for several cities in the US, rolling out to San Jose, Los Angeles, and Dallas, as well as Frankfurt, Germany. Areas surrounding these cities will also be able to connect to the new Ultimate tier servers.
This version upgrades GeForce Now’s premier RTX 3080 tier and rebrands it to Ultimate membership, offering the same benefits of the RTX 3080 tier but upgrading the cloud rig to an RTX 4080 GPU.
The service is powered by the Lovelace GPU architecture and, according to Nvidia, streams at up to 240 FPS with NVIDIA Reflex, up to 4K 120 FPS with support for DLSS 3 and RTX ON, and ultrawide support at up to 3,840 x 1,600p resolution at 120 FPS.
We punched in the numbers and found that if you paid for the Ultimate subscription tier in six-month increments for six years ($99.99, about £85/AU$145), it would cost the same as buying the RTX 4080 graphics card at its current MSRP. This makes it an excellent option for those with a solid internet connection who wants the performance of the current-gen graphics card without having to pay over $1,000 for it.
“After the start of the rollout of the RTX 4080 SuperPODs today, it’ll start rolling out to other regions, with wider release expected throughout Q1,” an Nvidia spokesperson told TechRadar. “On our weekly GFN Thursday blog, we’ll be giving updates each week on which regions are getting RTX 4080 performance.”
Could this be the future of PC gaming?
We previously tried out the RTX 3080 tier for our Acer Chromebook 516 GE review and found the performance on one of the best Chromebooks we’ve tested to be near indistinguishable from actually running a laptop with the best GPU on the market.
And when we went hands-on with the new Ultimate tier for CES 2023, we found that the performance is even better, as it addresses latency issues that have held the subscription service back. Not only would the upgraded servers bring system latency beneath that 60ms threshold, but Nvidia also claims that by incorporating Nvidia Reflex into its server-side processing, it can bring it down as low as 35ms, which is on par with an actual gaming PC running local hardware.
If this turns out to be true, that would be absolutely huge and make an already great service perfect for even hardcore and eventually competitive gaming, maybe even beating out even the best gaming PC you can get for a comparable price.