Boosteroid is examining what a future GPU surplus could mean for cloud gaming. In a recent LinkedIn post, the company explored how lower GPU rental prices could make expanding into new regions more affordable, potentially bringing cloud gaming to people who currently live too far from a nearby server.
Its argument is that new GPU clusters might keep coming online even after supply catches up with demand. With suitable hardware cheaper to rent, more regions could become viable for local cloud gaming servers.
That could be welcome news if your nearest cloud gaming data centre is far away. But Boosteroid isn’t announcing that a surplus has arrived. Whether this scenario benefits cloud gaming depends on which GPUs become cheaper and what it costs to put them to work.
GPU Prices Aren’t Moving in the Same Direction
On September 15, Boosteroid increased its monthly Ultra subscription from €12.89 to €16.89. CEO Ivan Shvaichenko attributed the change to higher operating costs. He reported that GPU server prices had risen severalfold over the previous year, while data centre hosting expenses had also increased.
The annual Ultra plan remained €89.89, and Boosteroid said existing subscribers would keep their current rates for as long as their subscriptions remained active.
A surplus would change some of those economics. If companies bring more GPU capacity online than customers need, providers may have to lower rental rates to attract business. But we aren’t seeing that across every type of hardware today.
AIMultiple’s September 2026 Cloud GPU Rental Price Index tracks listed rates across 75 providers. Its newest-generation GPU group rose from a median $2.12 per GPU-hour in October 2024 to $4.72 in September 2026. Over the same period, its legacy GPU group fell from $1.77 to $0.95.
The index measures advertised rental prices, and the models and listings contributing to each group can change. The figures don’t establish a market-wide decline in GPU costs. They also don’t tell us whether the custom hardware needed for cloud gaming is becoming cheaper.
Cheaper GPUs Could Help Cloud Gaming Reach More Regions
Opening a cloud gaming location takes an investment before the first subscriber connects. A provider needs gaming-capable machines, a data centre with sufficient power and cooling, and a fast, reliable network.
Lower hardware costs could reduce the number of paying customers a new location needs to support itself. That could change the calculation in regions where demand currently isn’t high enough to justify a local cluster. Providers would still have to account for electricity, networking and the risk of servers sitting unused.
If a nearby server becomes financially viable, the benefit is fairly straightforward. Game traffic may have a shorter route, which can reduce latency and make controls more responsive. That matters most when you’re playing something where a split-second delay is noticeable.
The catch is that lower GPU prices don’t create customers overnight. Expansion still depends on whether enough people in a region will use the service regularly.
AI Infrastructure Isn’t Automatically Cloud Gaming Infrastructure
Much of the new GPU infrastructure drawing attention is being built for AI. Those systems aren’t automatically suitable for streaming games.
NVIDIA introduced DSX POD at GTC Berlin on October 21. It’s a reference design for Vera Rubin AI infrastructure that partners can assemble, connect and test inside modular containers before delivery. NVIDIA estimates that deployment at a prepared site could fall from six months to as little as 14 days. Its partners expect to begin delivering the systems in early 2027.
That’s a substantial change to how AI computing capacity could be deployed, but it isn’t a cloud gaming announcement. An AI training cluster may not have the graphics rendering and video encoding capabilities that real-time game streaming requires. CPU performance and low-latency networking also matter.
Boosteroid’s current hardware provides a useful contrast. AMD detailed its work with the service in June, including custom Radeon GPUs paired with EPYC processors. Its server configurations are optimized for real-time game rendering and streaming, with cooling and networking suited to that workload.
So even if AI-focused accelerators become cheaper to rent, cloud gaming services can’t necessarily use them. The hardware has to run games effectively, and converting an existing AI cluster may not make economic sense.
Nuage’s Launch Highlights Today’s Capacity Problems
We saw how difficult securing usable GPU capacity can be when we spoke with Nuage founder Jordy Kieto earlier this week. His cloud gaming service launched on October 16, but demand quickly exceeded the machines available to it.
Kieto told us roughly 670 sessions started on launch day, yet only about 6% reached a ready cloud PC. Most of the others ended because no GPU machine was available.
To improve access, Kieto spread machines across more US regions, used alternative GPU types where needed and requested higher limits from his provider. He also introduced a queue so requests wouldn’t end immediately when capacity ran out.
The setback exposed a different constraint. Nuage needed access to suitable GPU machines in the right regions, but regional allocations and account limits restricted what it could use. A future surplus might improve access to some hardware, although it wouldn’t automatically remove those limits.
Lower Infrastructure Costs Wouldn’t Guarantee Cheaper Subscriptions
Even if GPU rental rates eventually fall, cloud gaming subscriptions won’t necessarily follow. Hardware is only part of the cost of running a service. Electricity, cooling, network capacity, maintenance and data centre contracts still have to be paid for.
Those pressures extend beyond cloud gaming. Uptime Institute’s July 2026 survey found that high costs remained the leading concern for data centre management teams. Power availability and capacity forecasting were also growing concerns.
And any savings don’t have to turn into lower monthly prices. A provider might open a new regional location, replace aging machines, improve streaming quality or add capacity to reduce queues. Those changes could be more useful to you than a small subscription discount.
Boosteroid’s post examines a possible change in the wider GPU market rather than announcing changes to its own subscription pricing or regional expansion plans.
If suitable gaming hardware becomes cheaper, some regions that are too expensive to serve today might finally justify a local cluster. Until providers can turn those lower costs into actual new capacity, a GPU surplus remains a possibility rather than an improvement you can use.
As always, remember to follow us on our social media platforms (e.g., Threads, X (Twitter), Bluesky, YouTube, and Facebook) to stay up-to-date with the latest news. This website contains affiliate links. We may receive a commission when you click on these links and make a purchase, at no extra cost to you. We are an independent site, and the opinions expressed here are our own.
