Wright's Law
Wright's Law holds that for every cumulative doubling of units produced, costs decline by a constant percentage — typically 20–30%. First observed by Theodore Wright in 1936 studying aircraft production, the pattern has since been confirmed across semiconductors, solar panels, lithium-ion batteries, genome sequencing, and now AI inference.
Unlike Moore's Law (which predicts a fixed time-based pace of improvement), Wright's Law ties cost reduction to cumulative experience. The more you build, the cheaper each unit becomes — not because of economies of scale alone, but because production experience drives process innovation, yield improvements, tooling optimization, and design simplification. This makes Wright's Law a more reliable predictor of technology cost trajectories: research by the Santa Fe Institute found it outperformed Moore's Law in forecasting across dozens of technology domains.
Why Wright's Law Matters for Digital Economies
Wright's Law is the engine behind deflationary technology. When production costs fall predictably with cumulative volume, the strategic question shifts from "can we build it?" to "how fast can we scale adoption to drive down costs?" This creates a powerful feedback loop: lower costs expand the addressable market, more users drive more cumulative production, and more production drives costs lower still. It is the same flywheel that made smartphones ubiquitous and is now making AI inference a commodity.
In AI, Wright's Law applies at multiple layers. Hardware costs (GPUs, custom ASICs) decline as fabrication experience accumulates. Inference optimization — quantization, distillation, speculative decoding — improves with each generation of deployment experience. And the software layer benefits from Wright's dynamics too: as more teams deploy AI agents, the ecosystem of tools, frameworks, and best practices deepens, reducing the effective cost of building the next agent.
Implications for Platform Strategy
Wright's Law creates winner-take-most dynamics in production-heavy domains. The company (or ecosystem) that accumulates production experience fastest enjoys a structural cost advantage that compounds over time. Tesla's battery strategy, TSMC's semiconductor dominance, and the aggressive pricing of cloud AI inference all reflect Wright's Law thinking: price ahead of the cost curve to maximize cumulative volume.
For virtual economies and gaming, the implication is that infrastructure costs — compute, bandwidth, AI inference — will continue their exponential decline. Features that are prohibitively expensive today (real-time AI NPCs, persistent simulated worlds, on-demand content generation) become economically viable as cumulative deployment drives costs down the learning curve. Platforms that invest early in these capabilities build an experience advantage that late entrants cannot easily replicate.
The agentic web compounds this further: as AI agents drive more automated transactions and workflows, they accelerate the cumulative volume of inference calls, pushing costs down the Wright's Law curve faster than human-only usage would. Each agent interaction is another unit of cumulative production, feeding the deflationary engine.