Compute is bottlenecked by energy, not chips
Why the constraint on the next wave of computing is power, and what that means for a country with abundant stranded gas and a grid that has never exceeded roughly 6 GW.
By Exusia Digital
Ask anyone building AI infrastructure what their constraint is and they will not say chips. They will say power, and specifically a megawatt they can energise this decade.
The queue is for interconnection, not silicon
Grid interconnection queues in mature markets now run to years. That is the real gating factor on new compute capacity, and it is why the most valuable thing a developer can hold is not a GPU allocation but a signed route to power.
Nigeria's version of the problem is sharper
The national grid has never exceeded roughly 6 GW for a population above 200 million. Anyone proposing to pull hyperscale load from it is proposing something the system cannot deliver. But the same country flares gas that earns its producer nothing, which is energy sitting in exactly the wrong place for a grid and exactly the right place for a modular compute site.
What follows from that
If power is the constraint and the power is stranded, you do not build where the fibre is and hope for a connection. You build where the energy is and bring the network to it. That is the whole of our deployment model, and it is why we describe ourselves as an energy company that happens to sell compute.