Productify — Rent a GPU by the hour · Earn with your hardware
Pproductify.
Bag

Rent a GPUby the hour

Rent now
Inside a gaming PC lit in rainbow colours, with a white GeForce RTX graphics card across the middle
Poster for Charge, an animated short made in Blender: an old man and a robot in neon light
Rendered on GPUs: Charge, a Blender open movie

Run your AI code on real NVIDIA graphics cards in people's PCs. Pay by the second.

$5 free credit to start

Or join the waitlist

This is a
graphics card.

Gamers buy them to play. The same chips run modern AI: training models, fine-tuning them, testing them.

An MSI GeForce RTX 3070 graphics card with three fans, standing on a desk

What GPUs like these power

Games, 3D and film are what a provider's GPU is built for. On Productify, renters run AI work: PyTorch and Python today, Blender rendering later.

  • Veloren: a voxel character overlooking mountains and forestGames Veloren
  • 0 A.D.: ships off a tropical beach, seen from aboveGames 0 A.D.
  • SuperTuxKart: a racer on a desert track at the start lineGames SuperTuxKart
  • Blender 3.0 splash art: an animated girl holding a small yellow bird3D Blender
  • Red Eclipse 2: a first-person arena with a fireballGames Red Eclipse 2
  • Luanti: a blocky desert landscape under the game logoGames Luanti (Minetest)
  • Unvanquished: a first-person view inside a sci-fi baseGames Unvanquished
  • Veloren: a glider flying over a forest at nightGames Veloren
  • Poster for Sprite Fright, a Blender open movie: teenagers in a forest of glowing mushroomsFilm Sprite Fright
  • OpenTTD: a busy isometric city with railways and bridgesGames OpenTTD
  • Xonotic: a first-person arena shooter in a dim corridorGames Xonotic
  • Battle for Wesnoth: a hex map with units and a character portraitGames Battle for Wesnoth
  • SuperTuxKart: a go-kart on a sunny beach trackGames SuperTuxKart

Most days,
it sits idle.

While its owner is at work or asleep, the GPU inside a gaming PC does nothing at all.

The inside of a black gaming PC: cooler, memory and graphics card slots, lit in white

Your laptop
can't train it.

AI work needs a GPU with enough memory. Buying one for a single project rarely makes sense.

nvidia-smi output in a terminal: NVIDIA GeForce RTX 4060, 54 degrees, 754 of 8188 MiB memory used

We connect
the two.

The owner pairs their PC with our app and decides when it's available. You rent its GPU in a locked-down container, billed by the second.

The Productify Node app on a provider's laptop: Docker running, GeForce RTX 4060, 52 °C, limits for renters, activity log

One afternoon,
one RTX 4060.

On 19 October 2026 we rented out a provider's gaming laptop end to end. This is the log — including the two things we found and fixed. Tap a row.

  5. Real work on the real GPU: the renter's code ran on the provider's card, and the GPU graph moved in Task Manager.

Start free.

1 credit is US$1. Instances use it at the machine's hourly price, by the second, and stop when it runs out. Paid packs open when payments go live — we're in test mode during the beta.

Free

$0

$5 credit, once

  • About 10 hours on a $0.50/hr GPU
  • PyTorch and Python environments
Claim it on the pricing page

Starter

$20

$21 credit

  • 5% extra credit
  • Credits don't expire
  • Email support
Coming soon

Builder

$50

$53 credit

  • 6% extra credit
  • Credits don't expire
  • Priority email support
  • For longer training runs
Coming soon
A Gigabyte GeForce RTX 3090 graphics card on a wooden desk

Provider spec sheet

What renters get

GPU
Your NVIDIA card, via CUDA
Environments
PyTorch or Python
Internet
Public sites, ports 80 & 443

What renters can't do

See your files
No access
Reach your PC or router
Blocked
Run as administrator
User 1000
Share your GPU
One renter at a time

What you control

Availability
Pause any time
Emergency stop
One click
Hourly price
You set it
Platform fee
10% (Free plan)
Earnings
Not guaranteed
How providing works

Questions

What can I run?

Python and PyTorch on an NVIDIA GPU, from a browser terminal. JupyterLab is on the way. Gaming and full remote desktops aren't supported.

How am I charged?

The machine's hourly price, metered by the second while the instance runs, plus $0.0002 per GB of disk per hour. Instances stop automatically when your credits run out.

Can the provider see my work?

Your job runs in an isolated container without access to the provider's files, but the provider owns the hardware. Don't upload secrets you can't afford to expose.

Is it live?

It's a closed beta: some machines are listed, they go online and offline, and payments are in test mode.

What does a provider need?

An NVIDIA GPU, Windows 10 or 11, Docker Desktop, a current NVIDIA driver and a steady internet connection.

Still curious?

Read the full FAQ, how isolation works on the security page, or contact us.

Have an NVIDIA GPU at home? Make it rentable while you're not using it — on your hours, at your price.

Become a provider

Tell us about your hardware. We review every application and help you pair your PC.

Apply as a provider
Photo creditsPhotographs via Wikimedia Commons: "2025 Wnętrze komputera PC" by Jacek Halicki (CC BY-SA 4.0); "MSI GeForce RTX 3070 VENTUS 3X OC" by TheStriker (CC BY 4.0); "Gaming-PC 20240426 HOF2514" and "Gigabyte GeForce RTX 3090 Eagle OC" by PantheraLeo1359531 (CC BY 4.0); "Charge" and "Sprite Fright" posters and Blender 3.0 splash by Blender Studio (CC BY 4.0); Game screenshots via Wikimedia Commons: Veloren by Antimundo (CC0); 0 A.D. by Wildfire Games (CC BY-SA 3.0); SuperTuxKart by the SuperTuxKart contributors (CC BY 1.0, CC BY 3.0); Red Eclipse 2 by TucanHolmes (CC BY-SA 4.0); Luanti (Minetest) by cisoun (CC BY-SA 4.0); Unvanquished by the Unvanquished developers (CC BY-SA 2.5); OpenTTD by the OpenTTD contributors (CC BY-SA 4.0); Xonotic by OliwierJaszczyszyn (CC BY-SA 4.0); Battle for Wesnoth by the Wesnoth developers (CC BY-SA 3.0). Screenshots and app photo: Productify test, 19 Oct 2026.