⚡ For GPU owners

Rent out your GPU.
Earn while it sits idle.

Rent out the GPU in your gaming PC, multi-GPU rig, workstation or Mac, and earn money while you are not using it. GridShare pays you for every hour your GPU is actually rented, and only for those hours. Setup takes 5 minutes; payouts go straight to your bank.

₹8.40–₹35.70
Per GPU, for every rented hour
≈₹12,800
A month · RTX 5090, rented 12 hours a day. An estimate, not a promise
5 min
To install
Daily
Bank payouts (or weekly / monthly)

Which GPUs can you rent out?

NVIDIA, AMD Radeon and Apple Silicon are supported, and the agent detects every GPU on install. The rates below are what your GPU earns for each hour it is actually rented — not a monthly guarantee. The rate follows the card's capability (VRAM and speed class), not its model year.

● LIVE
🖥️
NVIDIA GPUs
Gaming PCs, workstations, laptops (plugged in) and multi-GPU rigs, on Windows 10/11 (WSL2) or Ubuntu. In a rig, each card is listed separately and can serve a different buyer.
₹8.40 – ₹35.70/hr per GPU · RTX 2060 to RTX 5090
RTX 30/40/50 series · RTX A4000–A6000 · Quadro RTX
● LIVE Beta
🔴
AMD Radeon (ROCm)
RX 6000 and 7000 series on Ubuntu 22.04 with ROCm 6.1, set up by our AMD installer. RX 9000 series: email us for early access.
₹12.60 – ₹21.70/hr · RX 6800 XT to RX 7900 XTX
Ubuntu 22.04 only · install the AMD agent
● LIVE Beta
🍎
Apple Silicon
Any M1–M4 Mac on macOS 12+. Native agent, no Docker. Macs serve inference and LoRA fine-tuning (MLX, PyTorch MPS, llama.cpp); big unified memory suits large models.
₹8.40 – ₹37.10/hr · M1 base to M3 Ultra
Must stay plugged in while rented
Talk to us
🖧
Datacenter GPUs
A100, H100 or several machines? We onboard larger providers directly.
₹58.80 – ₹168/hr · A100, H100, H200
support@gridshare.in

How much can your GPU earn?

Select your hardware type and GPU model. Pick how many hours a day you expect it to be rented. Figures are an illustration, not a forecast. You earn only while your GPU is actually rented, and demand decides how often.

₹0
Illustration — you earn only for the hours rented
₹0
Daily
₹0
Monthly
₹0
Annual
Actual earnings depend on demand.
6× GPU rig ₹109.20/hr you earn
₹0
Illustration — you earn only for the hours rented
₹0
Daily
₹0
Monthly
₹0
Annual
Each GPU billed independently. Rig nodes can serve different buyers simultaneously.
₹0
Illustration — you earn only for the hours rented
₹0
Daily
₹0
Monthly
₹0
Annual
Pro GPU high-VRAM nodes attract premium ML workloads.
🔴 AMD ROCm Beta — Available Now. ROCm 6.1 + PyTorch 2.x. Ubuntu 22.04 required. AMD RX 6000/7000 series supported. Standard PyTorch and TF workloads run at 88–95% of equivalent NVIDIA performance. Install the AMD agent →
₹0
Illustration — you earn only for the hours rented
₹0
Daily
₹0
Monthly
₹0
Annual
AMD nodes are priced to reflect ROCm's smaller buyer pool. Rates will increase as ROCm adoption grows.
🍎 Apple Silicon Beta — Live Now. Native macOS agent (no Docker). Uses MLX + PyTorch MPS. macOS 12+ supported (macOS 13+ for full MLX). Any Apple Silicon chip, M1 base to M4 generation. Killer feature: M3 Ultra has up to 512 GB unified memory — run 70B+ LLMs that NVIDIA can't match. Earn ₹8.40–₹37.10/hr depending on chip.
₹0
Illustration — you earn only for the hours rented
₹0
Daily
₹0
Monthly
₹0
Annual
Apple Silicon nodes serve inference-only workloads — MLX, llama.cpp, Ollama, PyTorch MPS. Best for large LLM inference where unified memory is the bottleneck.

How to rent out your GPU in 4 steps

Works on Windows 10+ (with WSL2) or Ubuntu 20.04+. No technical knowledge required — our installer handles everything.

1
Run the Installer
One command installs Docker, NVIDIA Container Toolkit, and the GridShare Node Agent. Works on Windows 10+ with WSL2 or Ubuntu.
⏱ ~5 minutes · Automatic NVIDIA driver detection
2
Auto-Detect All GPUs
The agent finds every NVIDIA GPU in your system — each is registered as a separate node. A 6-GPU rig = 6 nodes that can serve different buyers simultaneously.
🔍 All NVIDIA GPUs detected · Per-GPU monitoring
3
Pause When You Need It
Pause any GPU from the dashboard and it takes no new rentals. A rental already running carries on; message us if you need it ended. Your hardware, your terms.
⏸ One-click pause · Support ends a live rental on request
4
Get Paid
Earnings accumulate in real time. Choose daily, weekly, or monthly payouts. Money lands directly in your bank account via IMPS/NEFT — any Indian bank.
💰 Daily from ₹150 · weekly from ₹300 · monthly free from ₹500

Gaming PC — Full earnings table

What you take home when your GPU is rented 10 hours a day, and 14 hours a day. Figures are your earnings — nothing is deducted from them.

Your GPU You earn/hr Monthly, rented 10 h/day Monthly, rented 14 h/day Yearly, rented 10 h/day
RTX 2060, RTX 3060 (8 GB), GTX 1660 Ti, GTX 1080 Ti · small models only ₹8.40 ₹2,520 ₹3,528 ₹30,240
RTX 3060 · 12 GB ₹10.50 ₹3,150 ₹4,410 ₹37,800
RTX 2070, RTX 2070 SUPER ₹11.90 ₹3,570 ₹4,998 ₹42,840
RTX 3060 Ti, RTX 2080 ₹12.60 ₹3,780 ₹5,292 ₹45,360
RTX 3070 ₹15.40 ₹4,620 ₹6,468 ₹55,440
RTX 2080 Ti, RTX 3070 Ti ₹16.80 ₹5,040 ₹7,056 ₹60,480
RTX 3080 ₹18.20 ₹5,460 ₹7,644 ₹65,520
RTX 4070 ₹19.60 ₹5,880 ₹8,232 ₹70,560
RTX 3080 Ti ₹20.30 ₹6,090 ₹8,526 ₹73,080
RTX 4070 Ti, RTX 4070 SUPER ₹23.10 ₹6,930 ₹9,702 ₹83,160
RTX 3090 ₹21.00 ₹6,300 ₹8,820 ₹75,600
RTX 4080 ₹29.40 ₹8,820 ₹12,348 ₹1,05,840
RTX 4090 ₹33.60 ₹10,080 ₹14,112 ₹1,20,960
RTX 5090 ₹35.70 ₹10,710 ₹14,994 ₹1,28,520

You earn only while a buyer is running on your machine. Amounts shown are what reaches your bank, before your own income tax.


Multi-GPU rig — the highest-yield provider setup

Your 6-GPU rig creates 6 independent nodes — all earning simultaneously.

How the rate is set: by the card's capability tier — VRAM and throughput class — not by model year. A newer card in the same class earns the same as an older one (a 5070 Ti and a 4070 Ti SUPER are both 16 GB-class "Pro Max" — each earns you ₹25.90/hr, plus a little more if the machine has 32 GB+ of system RAM). Buyers pay for what the GPU does, so providers with last-generation cards aren't undercut by a label.

Rig Config GPUs You earn/hr (all GPUs) Monthly, rented 10 h/day Yearly, rented 10 h/day Nodes created
Starter Rig 2× RTX 3080 ₹36.40/hr ₹10,920 ₹1,31,040 2 nodes
Classic Rig 4× RTX 3080 Popular ₹72.80/hr ₹21,840 ₹2,62,080 4 nodes
Full Rig 6× RTX 3080 Most Common ₹109.20/hr ₹32,760 ₹3,93,120 6 nodes
Full Rig 6× RTX 3080 Ti ₹121.80/hr ₹36,540 ₹4,38,480 6 nodes
Max Rig 8× RTX 3080 Highest Yield ₹145.60/hr ₹43,680 ₹5,24,160 8 nodes
Power Rig 2× RTX 4090 ₹67.20/hr ₹20,160 ₹2,41,920 2 nodes
4090 Rig 4× RTX 4090 ₹134.40/hr ₹40,320 ₹4,83,840 4 nodes

Yearly = 12 × monthly. Each GPU earns independently at its tier rate.


Workstation & professional GPU earnings

Professional GPUs with large VRAM attract premium ML fine-tuning and inference workloads. If your A6000 sits idle after 6 PM, it should be earning.

GPU VRAM You earn/hr Monthly, rented 8 h/day Yearly, rented 8 h/day
RTX A4000 16 GB ECC ₹16.80/hr ₹4,032 ₹48,384
RTX A5000 Recommended 24 GB ECC ₹22.40/hr ₹5,376 ₹64,512
RTX A6000 48 GB ECC ₹35.00/hr ₹8,400 ₹1,00,800
Quadro RTX 5000 16 GB ECC ₹11.90/hr ₹2,856 ₹34,272
Quadro RTX 6000 24 GB ECC ₹18.20/hr ₹4,368 ₹52,416

ECC memory makes workstation GPUs preferred for long-running ML jobs. High-VRAM nodes attract the most demand.


Payouts to your bank account

Choose daily, weekly, or monthly payouts. Sent straight to your bank account via IMPS/NEFT — fast, direct, any Indian bank.

📅 Daily
₹10 flat per payout · pays out from ₹150
Sent every day at 2 AM IST once your balance crosses ₹150. The convenience option — a flat ₹10 covers the transfer cost, whatever the amount.
📆 Weekly
₹5 flat per payout · pays out from ₹300
Sent every Monday. One small transfer fee instead of seven — about ₹20–25 a month.
🏦 Direct Bank Payouts
IMPS / NEFT to any Indian bank
No wallet in between · pays from ₹150 (daily) / ₹500 (monthly, free)
📊 TDS (Tax Deducted at Source)
Deducted only once your yearly earnings cross the limit in Indian tax law
Keep a valid PAN on file, or tax is withheld at a higher rate · Form 16A issued at year-end
📈 Real-time Tracking
Live earnings in Provider Dashboard
Per-GPU breakdown · Export to CSV

System requirements

NVIDIA live, AMD Radeon and Apple Silicon in beta. Pick your platform.

🎮
GPU Hardware
  • NVIDIA CUDA GPU (RTX, GTX, Quadro, or A-series)
  • Minimum 4 GB VRAM (8 GB+ strongly recommended)
  • NVIDIA driver version 525 or newer
  • Single GPU or multi-GPU rig both supported
💿
Operating System
  • Ubuntu 20.04, 22.04, or 24.04 (recommended)
  • Windows 10 or 11 with WSL2 enabled
  • Docker installed (auto-installed by our script)
  • NVIDIA Container Toolkit (auto-installed)
🌐
Network
  • 50 Mbps+ upload minimum
  • 100+ Mbps for multi-GPU rigs
  • Static or dynamic IP (NAT OK)
  • No port forwarding needed (reverse tunnel)
  • Port 443 outbound open
⚡
Power & Uptime
  • No minimum uptime (higher = more income)
  • PSU adequate for GPU sustained load
  • UPS recommended for high-uptime setups
  • Thermal monitoring enabled automatically

Is it safe to rent out your GPU?

Temperature, power and fan limits on every GPU, and every buyer job walled off from your files. How we keep your machine safe →

🌡️
Thermal Monitoring
Real-time temperature tracking per GPU. Every overheat is recorded and listed in your portal’s Safety tab, and we email you if a machine hits a critical temperature.
⚡
Power Limits
From 10 PM to 7 AM an NVIDIA GPU’s power is capped (60% by default). Change the hours and the cap in Settings.
🔒
Full Isolation
Every buyer job runs in an isolated Docker container. Buyers cannot access your files, network drives, or any data outside their container.
🪪
Verified, Traceable Buyers
Every buyer verifies their email and pays through an RBI-regulated payment gateway, so usage traces back to a real, funded account. Illegal workloads are prohibited by the Terms of Service and monitored at the platform level.
📊
Fan & Health Monitoring
Fan speed, memory errors, and GPU health metrics monitored continuously. Abnormal readings trigger alerts before hardware damage can occur.

Frequently asked questions

Everything you need to know before you start listing.

Can I list my 6-GPU rig?+
Yes. Install the agent once; it finds every NVIDIA GPU and lists each card separately, so one rig can serve several buyers at once. Example: 6× RTX 3080 earns ₹109.20/hr while all six are rented. Rented 10 hours a day, that is about ₹32,760 a month. Rigs suit this well because they are built for sustained load.
Can I still game while my GPU is rented out?+
No. You choose the hours your GPU is available; outside them no new rentals start. When you sit down at the PC we detect it within 30 seconds and stop charging the buyer, but their session keeps running in the background and still holds GPU memory. If you need the whole GPU back, message us and we will end the rental — a running rental is stopped only by the renter or by us, because someone is paying for that work. Many owners list overnight only.
Can renters see my files or access my PC?+
No. Every buyer job runs in an isolated container with GPU access only. Buyers cannot see your Windows or Linux files, cannot reach network drives, and cannot install anything outside their container. On a Mac the job runs in macOS's built-in app sandbox instead.
How much money can I earn by renting out my GPU?+
GridShare buys your GPU's time. Your machine shows a “You earn” rate per hour, set automatically by GPU model, and GridShare pays you that rate for every hour it is rented. Example: an RTX 4090 earns ₹33.60/hr, so 10 rented hours earn you ₹336. Monthly payouts to your bank are free; faster payouts cost a flat ₹5 (weekly) or ₹10 (daily), charged only when a payout is sent. Balances below the minimum (₹150 daily / ₹300 weekly / ₹500 monthly) roll over, fee-free.
Will running AI workloads damage my GPU?+
Sustained load is normal for a GPU — games run them at full load for hours, and rigs are built for it. GridShare adds per-GPU temperature monitoring (auto-pause at 85°C), a night-time power cap and fan monitoring that most people don't have when gaming. The main risk is electrical surges, so a UPS is recommended for machines that stay online a lot.
How does TDS work for provider income?+
GridShare deducts TDS on provider payments as the Income Tax Act requires, only once your earnings for the financial year cross the limit that applies to you. Keep a valid PAN on your account — without one, tax is withheld at a higher rate (Section 206AA). We issue Form 16A for TDS deducted, and you claim the credit when you file your ITR.
What happens if a buyer runs something illegal on my GPU?+
Every buyer verifies their email and pays through an RBI-regulated payment gateway, so every job traces back to a real, funded account. Our Terms prohibit illegal use, including crypto mining, DDoS, spam and CSAM, and network traffic is monitored at the platform level. Your rights and duties as a provider are set out in the Provider Agreement.
Can I list both a gaming PC and a multi-GPU rig on the same account?+
Yes. Install the agent on each machine. They all appear in your Provider Dashboard as separate machines, each with its own controls, earnings and schedule.
I have an AMD Radeon GPU — can I list it? What is ROCm?+
Yes — AMD support is live in beta. ROCm is AMD's compute platform, the equivalent of NVIDIA's CUDA. RX 6000 and RX 7000 series cards run PyTorch and TensorFlow workloads with ROCm 6.1. It needs Ubuntu 22.04; Windows is not supported for AMD. To get started, sign in to the portal and copy the install command shown there.
I have a Mac — can I list it on GridShare?+
Yes — Apple Silicon support is live in beta, for any M1, M2, M3 or M4 chip (base, Pro, Max or Ultra). Our native macOS agent (no Docker) serves AI workloads with MLX and PyTorch MPS, and large unified memory lets a Mac hold models that don't fit on most consumer GPUs. You need macOS 12 or newer (13+ recommended), Python 3.10+, and the Mac must stay plugged in while rented. To get started, sign in to the portal and copy the install command shown there.
How does the Apple Silicon agent work without Docker?+
Each buyer job runs as a separate process in macOS's built-in app sandbox, with access to one scratch folder only. Metal and MLX calls go straight to the GPU. So Macs serve inference (MLX, PyTorch MPS, llama.cpp) and LoRA fine-tuning (MLX-LM), not arbitrary Docker images. More on safety →
How do I remove GridShare from my PC?+
Windows: download GridShare-Uninstall.bat and double-click it (it asks for administrator permission once), or in PowerShell as administrator: irm https://gridshare.in/dl/uninstall.ps1 | iex — Mac: curl -fsSL https://gridshare.in/dl/uninstall-mac.sh | bash — Linux: curl -fsSL https://gridshare.in/dl/uninstall-linux.sh | bash. It removes the agent and its auto-start entries and restores the power settings the installer changed. On Windows it leaves Ubuntu (WSL2) and Docker in place because other software may use them; wsl --unregister Ubuntu-24.04 deletes that whole environment too. Your account and pending earnings are unaffected, and you can reinstall any time.

Ready to start earning?

Setup takes 5 minutes. No commitment. Stop anytime.
Whether it's one GPU or an 8-card rig — we handle everything.

Add your machine — sign up / sign in → Read the Docs
NVIDIA CUDA (live) · AMD ROCm RX 6000/7000 (Beta · Live Now) · Apple Silicon M1–M4 (Beta · Live Now) · Free to list