Compute 2030: Four Scenarios for the Compute Transition

Introduction

The compute transition is not a technological event. It is a governance event. The question is not whether large-scale compute will reshape economic structures, labor markets, and the distribution of productive capacity. It will. The question is whether our regulatory institutions will have sufficient variety to absorb the disturbances it introduces.

W. Ross Ashby's Law of Requisite Variety, formalized in 1956, provides the most precise statement of this challenge available in the scientific literature. A regulator can reduce the variety of outcomes in a system only to the extent that it possesses at least as much variety as the disturbances it must absorb. Applied to compute governance: a regulatory body that cannot model the full range of compute applications cannot govern them.

This report presents four structural scenarios for the compute transition through 2030. Each scenario is defined by a distinct configuration of compute power concentration, regulatory institutional variety, and international coordination capacity. Each carries different implications for governance, equity, and democratic accountability.

"Only variety can destroy variety." — W. Ross Ashby, 1956

The Analytical Framework

Our scenario methodology draws on two theoretical traditions: Ashby's cybernetic framework for analyzing regulatory capacity, and the structural scenario planning methodology developed at Shell and subsequently adapted for policy analysis by the RAND Corporation and others.

The key analytical variable is the variety ratio: the ratio of regulatory variety to disturbance variety. A ratio below 1.0 indicates a variety deficit — the regulator cannot fully absorb the disturbances it faces. A ratio above 1.0 indicates regulatory excess — the regulator has more capacity than necessary, which is typically associated with over-regulation and efficiency losses.

We estimate current variety ratios for compute governance across five dimensions: technical modeling capacity, jurisdictional coverage, enforcement authority, institutional independence, and adaptive capacity. Across all five dimensions, existing regulatory bodies exhibit significant variety deficits.

Scenario I: Concentrated Dominance

In Scenario I, frontier compute remains concentrated in a small number of firms and states. Regulatory institutions fail to develop sufficient variety to govern the full range of compute applications. The variety deficit widens as compute capabilities advance faster than institutional capacity.

This scenario is characterized by: persistent concentration of frontier compute in 2–3 firms; regulatory capture of compute governance bodies; widening compute equity gaps across geographies and sectors; and increasing use of compute as a geopolitical instrument.

We assess Scenario I as the most likely near-term trajectory absent deliberate institutional intervention. The structural forces driving concentration — network effects, capital requirements, talent concentration, and regulatory arbitrage — are strong and self-reinforcing.

"A regulator that cannot model its system cannot govern it." — Conant & Ashby, 1970

Scenario II: Multilateral Fragmentation

In Scenario II, compute governance fragments along geopolitical lines. Multiple competing compute blocs emerge, each with distinct regulatory frameworks, technical standards, and access regimes. The result is a fragmented global compute landscape with high coordination costs and significant efficiency losses.

This scenario is characterized by: 3–5 competing national compute blocs; incompatible technical standards and regulatory frameworks; significant efficiency losses from fragmentation; and reduced capacity for international coordination on shared governance challenges.

Scenario II is more likely than Scenario III (Governed Transition) but less likely than Scenario I in the near term. The conditions for fragmentation — geopolitical competition, divergent regulatory philosophies, and strategic compute nationalism — are already present and strengthening.

Scenario III: Governed Transition

In Scenario III, international coordination succeeds in establishing shared governance frameworks for compute. Regulatory institutions develop sufficient variety to govern the full range of compute applications. The variety deficit narrows as institutional capacity advances in parallel with compute capabilities.

This scenario requires: successful international coordination on compute governance standards; significant investment in regulatory institutional capacity; effective mechanisms for compute equity redistribution; and sustained political commitment to governance over a multi-year horizon.

Scenario III is the most normatively desirable outcome but requires deliberate, sustained institutional effort. The Good Regulator Theorem implies that achieving it requires regulatory bodies to develop genuine modeling capacity for the full range of compute applications — a significant institutional challenge.

Scenario IV: Diffuse Proliferation

In Scenario IV, compute capabilities diffuse rapidly across a wide range of actors. The concentration of frontier compute decreases as hardware costs fall and open-source models proliferate. Regulatory challenges shift from governing concentrated power to governing distributed capabilities.

This scenario is characterized by: rapid diffusion of compute capabilities; proliferation of capable open-source models; shift in governance challenges from concentration to distribution; and increased difficulty of enforcement.

Scenario IV is the most uncertain of the four scenarios. Its likelihood depends heavily on hardware cost trajectories, open-source model development, and the effectiveness of export controls — all of which are subject to significant uncertainty.

Governance Implications

Across all four scenarios, the central governance challenge is the same: building regulatory institutions with sufficient variety to govern the full range of compute applications. The scenarios differ in the urgency, form, and feasibility of this challenge.

Three institutional priorities emerge from the analysis. First, regulatory bodies must develop genuine technical modeling capacity — not just the ability to regulate existing applications, but the ability to anticipate and model novel ones. Second, international coordination mechanisms must be designed to function under conditions of geopolitical competition, not just cooperation. Third, compute equity mechanisms must be built into governance frameworks from the outset, not added as afterthoughts.

The Good Regulator Theorem provides a precise criterion for evaluating regulatory proposals: does this proposal increase the variety of the regulatory body relative to the variety of the disturbances it faces? Proposals that fail this criterion, however well-intentioned, will not produce effective governance.