Stages
6
Demand pulls hyperscaler spending — the check writers — onto chips, memory, networking and power, then circles back to the labs.
One machine, mapped stage by stage: who needs the compute, who pays for it, who receives the checks, who builds and powers the silicon — and how the profits loop home. Click any block to trace its connections and the KOPS oracle that tracks it.
Six stages, left to right. Capital flows downstream onto the receivers; chip & cloud profits recycle back to the labs. Click any block to light up everything it connects to.
Where to look as the money moves — each lens is becoming a KOPS oracle.
Hyperscaler capex — and the backlog (RPO) behind it — funds the entire build-out. That spend, and the ratio of receivers to writers, is the single largest forward signal. Tracked by Capex & Check-Flow.
As money flows down the chain, margin pools wherever supply is scarce — accelerators and HBM keep the largest slice, commodity layers the least. Tracked by Value-Capture.
Electricity, transformers and cooling now gate how much compute can run at all — the bottleneck has moved from silicon to power. Tracked by AI-Power.
The blocks in the map are tracked by these settleable KOPS oracles.
From the labs that need compute to the power plants that run it — and the money that loops back home.