AI Explainer: what is cloud gaming?
TL;DR
Cloud gaming involves streaming a game from a remote GPU server straight to a screen, no console or gaming PC required. This piece covers how that works, why it’s become the industry’s first real proof point for edge GPU infrastructure, how fast the market is growing and what it takes to deliver it well.
Cloud gaming means playing a video game that isn’t running on the device in a player’s hands at all. The game runs on a remote server packed with GPU power, and what reaches a phone, TV or laptop is a live video stream of the gameplay, with inputs sent back to the server in the other direction. There’s no cartridge or download, and no console under the TV. Just a screen and a connection fast enough to keep up.
That’s different from online gaming in the traditional sense, where the game still runs on the player’s own device and the internet only carries multiplayer data back and forth. In cloud gaming, the device does almost nothing, all the processing that would normally happen locally happens on the server instead.
None of this is new in concept. OnLive tried the same in 2010 and gutted itself within two years, undone by latency and network infrastructure that wasn’t ready for it. Google’s Stadia tried again in 2019 and shut down by 2023 for largely the same reason.
What’s different now is the infrastructure itself. Faster, more widely available broadband and a much bigger footprint of GPU compute sitting close to users.
How it works
The moment a player starts a session, the service allocates a slice of a remote GPU to run it. That server renders every frame exactly as a console or gaming PC would, encodes it as video and sends it to the player’s device. The device decodes the video and displays it, then sends the player’s inputs, a joystick tilt, a button press, back up to the server, which renders the next frame accordingly.
That round trip has to happen dozens of times a second, and it has to feel instant. Services like Xbox Cloud Gaming and NVIDIA’s GeForce NOW are the clearest consumer face of this today. If the loop lags, player input feels disconnected from what’s on screen, and the game is unplayable no matter how good the graphics are.
Why it’s the industry’s proving ground for edge infrastructure
Cloud gaming runs on the same requirement that’s about to define AI inference: compute has to sit close enough to the user that latency stays invisible.
That overlap is what makes cloud gaming useful as a signal beyond gaming itself. It’s a workload with a paying, latency sensitive user base that already exists at scale, running on the same distributed, low latency GPU infrastructure that AI inference at the edge depends on.
That dynamic played out in public in July 2026, when Microsoft and Amazon both revised their cloud gaming strategies within the same week. Microsoft began testing ad-supported streaming that lets players stream games they already own for free, and Amazon folded its Luna service into Prime Video, dropping it as a standalone destination. Neither change was about which platform has the better game library. Both were attempts to solve the same problem, getting more people into a service that only works if the infrastructure underneath it holds up well enough to feel instant.
Telcos are looking at cloud gaming as a proven way to put existing 5G network investment to work, ahead of enterprise AI inference following the same path.
The scale of the shift
Estimates for how big the cloud gaming market is vary. Grand View Research puts global revenue at $4.9 billion in 2026, growing to $62.9 billion by 2033 at a 43.7% CAGR. Boston Consulting Group’s own December 2025 forecast is more conservative, projecting the market from $1.4 billion in 2025 to $18.3 billion by 2030, with the player base growing from 5 million to 65 million over the same period.
Fortune Business Insights, using a broader definition, puts 2026 revenue alone at $23.79 billion. The spread says more about how loosely the category gets defined industry wide than it does about direction, which every firm agrees on: cloud gaming is a small slice of the broader $374.8 billion gaming market today, and the fastest growing part of it.
What it takes to deliver
Delivering cloud gaming well is an infrastructure problem before it’s a software one. It means positioning GPU compute close to where players are, and running a routing layer that sends each session to the fastest available server as network conditions and load shift in real time.
A federated network of operator infrastructure, spread across many points of presence rather than concentrated in a handful of centralized facilities, gives that routing decision a reach no single facility could offer alone. It’s the same infrastructure logic AI inference is now following, with cloud gaming a step ahead in proving it can carry a paying, latency sensitive audience at commercial scale.
For a closer look at why proximity is becoming non-negotiable for AI too, read our short read on low latency AI by clicking below.
.avif)