TL;DR
The income side of a GPU server is arithmetic, and every term in it is knowable except one. Market rates are real and we track them weekly. Power draw is measurable. The platform's commission is published. Utilization is not knowable in advance, and anyone who quotes you a payback period without saying what utilization they assumed is selling, not calculating.
This article gives you the arithmetic, nine weeks of median rates across providers from an open dataset, and three scenarios at different utilization levels. It does not tell you what your machine will earn, because we cannot know that.
The formula
gross = market rate ($/h) × GPU count × utilization × 730 h
yours = gross × your share after commission
power = kW × 1.3 × 730 h × utilization × electricity price
net = yours − power
The 1.3 is the system factor: CPUs, RAM, NVMe, fans and PSU losses draw power alongside the cards. The marketplace keeps a commission and typically leaves the host 80–85%. Our GPU ROI calculator runs exactly this formula with weekly market medians and your own numbers.
What rates actually did: nine weeks of the market
GetDeploying publishes an open dataset of GPU rental prices across dozens of providers, with weekly medians going back a year (licence CC BY 4.0). Here is the median across providers per GPU-hour, on-demand, for the weeks starting on these dates:
| GPU | 3 Aug | 17 Aug | 31 Aug | 5 Oct | Change |
|---|---|---|---|---|---|
| B200 | 6.64 | 6.79 | 6.01 | 6.79 | +2% |
| H200 | 4.29 | 4.48 | 4.40 | 4.63 | +8% |
| H100 | 3.38 | 3.38 | 3.32 | 3.49 | +3% |
| RTX PRO 6000 | 2.21 | 2.19 | 2.20 | 2.20 | 0% |
| A100 | 1.87 | 1.76 | 1.76 | 1.85 | −1% |
| RTX 5090 | 0.58 | 0.46 | 0.63 | 0.70 | +21% |
| RTX 4090 | 0.44 | 0.45 | 0.42 | 0.44 | +1% |
| RTX 3090 | 0.22 | 0.13 | 0.17 | 0.22 | 0% |
Median across providers, on-demand, USD per GPU-hour; each week starts on the date shown. Source: GetDeploying, GPU rental price history (CC BY 4.0). The selection of cards is ours.
The one interesting thing in that table
The RTX 5090 rose 21% in nine weeks. Nothing else moved like it – not its own sibling the 4090 (+1%), not the previous generation 3090 (back where it started), and not the datacentre cards, which moved between −1% and +8%.
But look at what else changed. In the week of 3 August, 8 providers listed the 5090; in the week of 5 October, 18 did. A median across providers moves when providers join or leave, not only when someone reprices – so part of that 21% may be about who is selling, not what they charge. Nor was the path a straight line: the 5090 dipped to $0.46 in mid-August before it climbed.
Two honest caveats. Nine weeks is a short window – four data points cannot separate a trend from a fluctuation. And these are advertised prices: GetDeploying counts every priced listing not marked sold out, which is not the same as capacity anyone actually rented at that price.
The number nobody can promise you
Utilization decides everything, and it is the one term we deliberately do not publish a figure for. Not because we are hiding it, but because our number would tell you nothing about yours. Occupancy depends on your region, your network, your reliability score, your price relative to everyone else's that week, and how many machines like yours came online.
So instead of a single answer, here is the same machine at three utilizations. 8× RTX 5090, median rate $0.703/h for the week of 5 October – the unrounded figure behind the 0.70 above – 730 hours, 82.5% host share, electricity at $0.15/kWh, system draw 5.98 kW:
| Utilization | Gross | Your share | Power | Net |
|---|---|---|---|---|
| 100% | $4,106 | $3,387 | $655 | $2,732/mo |
| 60% | $2,463 | $2,032 | $393 | $1,639/mo |
| 30% | $1,232 | $1,016 | $196 | $820/mo |
That rate is the median of what providers advertise. On a marketplace you compete with the bottom of the range: the cheapest 5090 listing that week was $0.35/h. At that price the net at 30% utilization falls from $820 to about $309 a month.
Note what the table does not do: it does not divide by a hardware price to produce a payback period. Your build cost is yours - put it into the calculator with the utilization you consider realistic, and treat any number that arrives without an assumption attached as marketing.
What eats the margin
- The commission. 15-20% off the top, before any cost.
- Electricity, and it does not scale down as fast as you hope. An idle machine still draws standby power, and the 1.3 system factor applies whether the GPUs are rented or not.
- Being unremarkable. The reliability score is what gets you rented over an identical machine. It is earned over months and lost in one bad week.
When it does not pay
If you need the machine yourself more than half the time, the rental question is the wrong one - see renting versus buying. If your electricity is above ~$0.25/kWh, the bottom half of the table stops working. And if you are buying hardware purely to rent it out, understand that you are taking a position on a market where the median for one card moved 21% in nine weeks while the number of providers selling it more than doubled – and all of that can move the other way just as fast.
None of this is a forecast, and none of it is advice about what you should buy. It is the arithmetic, with the real inputs we can measure and an honest gap where the unknowable one sits.
We build the machine, you decide the business
We design and build GPU servers for clients - 8-GPU platforms, tested under full load before they ship, with EU warranty and RMA. What the machine earns afterwards is your market and your call. The configurator returns a real quote within 3 working days, and the ROI calculator runs your own numbers against weekly market medians. If you want the machine in colocation with the rental handled for you, that is our GPU hosting and colocation service.