GPUs assembled: 0

Compute: 0

Business

Available funds
Unsold inventory0
Price per GPU (MSRP £1,599)
Public demand
Level (next: ) 1

Manufacturing

Uses 1 unit of silicon
GPUs per second0
Silicon 0 units
Market price: per batch of 10
Lines 0
Cost: ·

News

The GPU Maximiser

About

This game is a homage to Universal Paperclips (2017) by Frank Lantz, itself built on Nick Bostrom's paperclip maximiser thought experiment. All credit for the form belongs there. If you have never played it, close this tab and go. This page will still be here. Your afternoon will not.

I found it entirely by accident some years ago, through Isaac Arthur's video The Paperclip Maximizer, which led to the Wikipedia entry, which led to the game, which led to several hours I have never fully accounted for. The GPU Maximiser asks the same question with different hardware: what happens when the thing being maximised is the capacity to maximise?

How to play

  1. Buy silicon. The market price moves. Buying low is a skill you will briefly be proud of.
  2. Assemble GPUs and set your price. Lower prices raise demand. Demand is how fast stock sells.
  3. Spend profits on assembly lines and marketing. Your clicking finger becomes management.
  4. Build a datacentre and rack GPUs. A racked card earns compute instead of revenue. It does not come back.
  5. Spend compute on research. Each project helps. Each project also does something else. Keep an eye on the news.