R1 — Capital → compute flywheel (reinforcing). The spine. Profits fund the cash to burn, cash scales compute, compute inflates the frontier training budget, and that budget becomes a barrier nobody else can pay. Concentration compounds.
R2 — Data advantage flywheel (reinforcing). Users generate proprietary data, data deepens the moat, the moat concentrates the market, and market power buys more users.
R3 — Investment momentum (reinforcing). Market cap raises shareholder expectations, expectations become pressure to deploy, deployment wins users, users become profits, profits become market cap.
R4 — Productivity cost-savings (reinforcing). AI adoption raises productivity per worker, cuts operating costs, lifts corporate profits, and profits fund more AI.
R5 — Layoff savings (reinforcing). Productivity makes layoffs thinkable, layoffs cut operating costs, the savings lift profits, and profits fund more automation.
R6 — Price competitiveness → demand (reinforcing). Lower costs make prices competitive, demand grows, revenues grow, and profits fund the next round of adoption.
R7 — GPU scarcity & moat (reinforcing). Compute hunger drives GPU demand, scarcity raises prices, high prices raise the cost of entry, and the moat deepens for whoever already owns compute.
R8 — Capital-fueled AI hiring (reinforcing). Capital bids up demand for AI talent, salaries inflate the frontier training budget, and the budget locks out everyone who cannot pay it.
R9 — Academic brain-drain (reinforcing). Big-tech salaries open a compensation gap, the gap drains researchers out of universities, weaker academia attracts less funding, and the gap widens. Vicious.
R9b — Strategic tech leadership (reinforcing). States fund compute for strategic leadership, the industry grows, tax revenues and defence stakes grow with it, and the public funding gets renewed.
R10 — Political capture (reinforcing). Site-added name for a cycle the board draws but never labels: concentration buys political influence, influence sets political priorities, priorities direct public investment into compute, and the returns concentrate further. This is the loop that keeps the brakes weak.
B1 — Electricity price brake (balancing). Compute demand raises electricity prices, and prices push back on demand. The grid is the one brake nobody can lobby away.
B2 — Cost-of-growth brake (balancing). Growth forces hiring, hiring raises operating costs, costs erode price competitiveness, and demand cools. A real but slow brake.
B3 — Reskilling vs layoffs (balancing). Reskilling blunts the incentive to lay off, and layoffs create pressure to reskill. Today this loop barely runs.
B4 — Reskilling vs unemployment (balancing). Reskilling drains unemployment, and unemployment creates demand for reskilling. A brake that needs public investment to spin.
B6 — Antitrust brake (balancing). The weak brake (numbering site-assigned from the parked registry). Concentration raises barriers, barriers invite antitrust pressure, remedies lower barriers a little, and the loop resets. Pressure without teeth, so far.
C1 — Executive compensation (outcome chain, not a loop). Not a feedback loop, an outflow. Profits lift the share price, the share price sets executive pay, and pay drains into concentrated wealth at the top.
C2 — Profits → concentration (outcome chain, not a loop). Not a feedback loop, an outflow. Corporate profits become shareholder wealth, and shareholder wealth drains into concentration of power in the AI era.