Will It Run AI · Cost Calculator
Calculate electricity cost, compare it to cloud API pricing, and find out exactly when your GPU investment pays for itself.
Section 1
How much does it cost to run your GPU for AI inference? Based on TDP, hours of use per day, and your local electricity rate.
TDP: 450W · MSRP: $1,599
Using: $0.160/kWh
Daily kWh
3.6 kWh
Monthly cost
$17.28
Yearly cost
$210.24
CO₂ / month
43.2 kg
avg. grid intensity
Summary
Running RTX 4090 24GB for 8h/day at $0.160/kWh costs $17.28/month and emits approximately 43.2 kg CO₂/month.
Section 2
Compare what you'd pay for API access against the electricity cost of running the same workload locally.
Input: $2.50/1M tokens · Output: $10.00/1M tokens
Assumes 25% input / 75% output token split.
Cloud API
$24.38
per month
Daily: $0.81
Yearly: $292.56
100,000 tokens/day
Local GPU
$17.28
per month (electricity)
Daily: $0.58
Yearly: $207.36
450W · 8h/day
Verdict
Local inference with RTX 4090 24GB is $7.10/month cheaper than OpenAI GPT-4o at your usage level.
Section 3
Factor in the GPU purchase price to find exactly when local inference becomes cheaper than paying for API access over time.
Pre-filled with MSRP: $1,599
Payback
Your RTX 4090 24GB pays for itself in 226 months
GPU cost
$1,599
API / month
$24.38
Electricity / month
$17.28
Savings after 12mo
−$1,513.80
Cumulative cost over 24 months
Methodology
Electricity cost is based on GPU TDP (max power draw) × hours of use per day × your electricity rate. Actual draw may be 60–90% of TDP under inference workloads.
API cost assumes 25% input tokens and 75% output tokens. Real costs vary by prompt length. Prices sourced from provider pricing pages as of 2025.
Payback period is when cumulative API cost savings exceed the GPU purchase price plus electricity. It does not include setup time, hardware maintenance, or idle power draw.
