Why GeForce NOW put a 100 hour cap on your gaming
A time limit on cloud gaming is not a random choice. It comes straight from how the network is built.
On January 1, GeForce NOW added a 100 hour monthly cap. Play past it and you are done for the month, or you buy more time in 15 hour blocks: 2.99 dollars on the Performance tier, 5.99 on Ultimate. Up to 15 unused hours roll over. Day one Founders members keep unlimited play for life as long as their subscription never lapses, which tells you NVIDIA knows exactly how much that perk is worth.
It is easy to read the cap as stinginess. It is more useful to read it as physics.
Central data centers have a ceiling
Most cloud gaming runs on a handful of large data centers packed with GPUs. Every hour you play, a GPU somewhere is running flat out just for you. That hardware costs money to buy, power and cool, and there is a fixed amount of it.
NVIDIA says fewer than 6 percent of players will ever hit 100 hours. Believable. But averages were never the problem, peak hours are. Friday evening, everyone logs on at once, and the exact number of GPUs in the building decides who plays and who waits. A cap trims the heaviest users so the peak fits the hardware.
When more people want to play than the data centers can serve, a company has two levers. Raise the price, or cap the usage. 2025 was the year of the first lever, across the whole industry. 2026 opened with the second.
The faster servers do not fix it
GeForce NOW also rolled out faster hardware this year, up to 5K at 120 frames per second. Better pixels, same ceiling. Faster GPUs still sit in the same buildings, and there are still only so many of them. Speed and scale are two different problems, and a data center upgrade only solves the first one.
Another way to build it
The ceiling exists because the compute is stacked in one place. Spread that compute out, closer to players and across far more machines, and the ceiling moves. Adding capacity stops meaning breaking ground on a building and starts meaning adding rigs to a network.
That is the direction we find interesting, and it is why Walter is not shipping a play time cap in early alpha. We would rather grow the network than grow the meter.