Interactive
Seven parameters change the branches in this project. Move them and the three panels update. The result is written in states, not numbers.
These assumptions also drive every counter, bar, and cluster highlight on the scenario & map the combined scenario page.
Relative speed of frontier progress, especially agentic systems and automated R&D.
Stringency of controls on advanced chips, tooling, and cloud access affecting ASEAN and East Asia.
Growth of AI-capable capacity and the share of it firmly matched with renewable generation.
Speed and strength of convergence on shared AI governance, evaluation, and compute principles.
How quickly agentic workflows absorb Tier-1 and Tier-2 outsourcing and digital service tasks.
Annual capacity to move workers into AI deployment, evaluation, and governance roles.
Whether chips, clouds, and safety tooling sit with a few frontier entities or a diversified mix including regional actors.
Displacement arrives through sustained attrition instead of mass layoffs. It is easier to absorb and harder to see, so policymakers respond later.
Transition programmes exist but reach a small share of those affected, and skew toward workers who were already the most employable.
Entry-level pathways narrow but survive, leaving a route from front-line work into supervision, quality, and process design.
Compute capacity grows steadily and unevenly, concentrated in two or three hubs, with the rest of the region importing inference.
Grid stress stays manageable. So does the region's bargaining position, because little local capacity leaves little to trade with.
Licensing friction, not prohibition, does most of the work. Compliance ambiguity slows shipments as effectively as any ban.
A few states adopt common technical annexes while others hold out. The result is a workable core and a persistent periphery.
Control is uneven. A few providers dominate inference while regional actors hold real positions in packaging, hosting, and evaluation.
The region can audit what it deploys. Accredited regional evaluators make pre-deployment tests and disclosure rules enforceable.
This sandbox does not run a simulation. Each parameter falls into a low, moderate, or high band. The panels select written statements from that combination. This lets you inspect the scenario's conditional logic.
It does not produce probabilities, employment numbers, or capacity projections. Decimal outputs would imply a precision the project does not have. Test automation intensity against reskilling throughput. That pair creates the largest differences in outcome.