Why we start with mining, when Crusoe stopped
The category leader sold its entire bitcoin mining division in 2025. Here is why the same decision would be wrong for Nigeria today, and what would change our mind.
By Exusia Digital
Crusoe Energy invented this category and then, in March 2025, sold its entire bitcoin mining division to focus on AI. We start with mining anyway. Here is the reasoning, in full.
Crusoe exited into demand we do not have
It moved because US hyperscaler demand was contracted, immediate and enormous. Nigeria has no equivalent queue of AI offtakers today. Making the same move here would leave the power earning nothing while we waited.
Mining tolerates an energy profile that AI cannot
Associated gas volumes fluctuate. A training run cannot be interrupted; a mining load can stop and restart in seconds. Until a gas supply is proven at a constant rate, mining is the only workload that matches it.
It funds the harder build
Revenue from a mining site underwrites AI capacity without needing an anchor tenant to sign before construction. That is how a development-stage business gets to high-value capacity without a chicken-and-egg problem it cannot solve.
The site is the asset, not the racks
Land, power infrastructure and modules are workload-agnostic. Converting a site from mining to AI is a rack-level change. We are not building a mining company; we are building sites, and starting them on the workload that pays from day one.
What would change our mind
A creditworthy AI offtaker signing for capacity, or hashprice falling through the floor published in our model. Both are in the risk register, and the second has a defined response.