Own a share of a GPU.

Moonshift splits real GPUs into shares on the XRP Ledger. Right now 1 GPU is online.

Rent a GPU by the hour

Every GPU below passed its health check in the last minute. Run a task on one, or buy a share of what it earns.

GPU capacity

What each GPU costs by the hour, and whether it is free right now.

GPU	RAM	Storage	USD / hour	Free nodes
NVIDIA T4 16GBGPU-A	16 GiB	100 GB storage	–	Checking

Working around
the clock.

Here is GPU-A over the last 20 hours and the next 4. Each bar is five minutes. Dark bars are recorded work, light ones are idle time. New work appears as tasks run.

Times in UTC

20:00

23:00

02:00

05:00

08:00

11:00

14:00

NOW

17:00

WorkingIdleAhead

Working, last 20 hoursLoading

Revenue, last 24hLoading

Paid per share, last 24hLoading

80¢

of every dollar a GPU earns goes to its shareholders.

Revenue split per 1 USDC earned

0.80 USDCto shareholders

0.15 USDChosting, 15%

0.05 USDCreserve, 5%

Paid in a single XRPL Batch transaction at 00:00 UTC. Every shareholder gets their cut in the same transaction, so nobody gets paid first.

Borrow without selling.

Lock your shares and borrow up to half their value in USDC. Locked shares keep earning their nightly payout.

Locklocks your shares until the loan is repaid

Paymentthe lending pool sends you USDC

Batchboth happen together or not at all

Borrow against GPU-A sharesCurrent base APR

...

Estimated borrowing cost

Waiting for current lending data

Base-rate examples using the live share price and round liquidity. Each point varies shares and loan size. Connect your wallet for your rate and limit.

Example shares to lock

Collateral value0.00 USDC

Lending round available...

Get your quote

Point your OpenAI client here.

Moonshift's router speaks the OpenAI API. Requests are metered by token. Shareholders receive 80% of GPU revenue.

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://moonshift-protocol.vercel.app/api/v1",
    api_key=os.environ["MOONSHIFT_KEY"],
)

r = client.chat.completions.create(
    model="Qwen/Qwen2.5-1.5B-Instruct",
    messages=[{"role": "user",
               "content": "Explain what a GPU does in one sentence"}],
    extra_body={"priority": "balanced"},  # cheapest | balanced | fastest
)
# routed to the best GPU for that priority

From our GPUsRecorded answers

Answers will appear here.

Public questions and their GPU responses rotate here automatically. No wallet is needed to view them.

No completed public response is available yet.

Ask your own question

Private prompts use your signed-in Moonshift account. USDC is checked before each request.

Private prompt Checking available models

Checking your USDC balance...

The answer appears here.

The app

Enter the app

Trade GPU shares on the XRP Ledger, run compute tasks, or borrow against the shares you hold.

Enter app