Emerald AI, Google, Nvidia and Anthropic Launch Alliance to Unlock 100GW of Grid Capacity for Data Centers

Skynet Chance (+0.01 pp): Removing energy bottlenecks enables a larger aggregate compute base for frontier AI training and deployment, marginally increasing the scale of systems that could pose control challenges. The news concerns infrastructure logistics, not alignment or autonomy, so the effect on loss-of-control probability is small.

Skynet Date (no change): Unlocking up to 100GW of grid capacity accelerates the physical buildout that gates AI capability growth, modestly compressing the timeline for very large-scale AI systems. The effect is indirect and depends on actual deployment of demand-response programs.

AGI Progress (+0.01 pp): Grid interconnection has become a binding constraint on data center expansion, and making demand response routine addresses that bottleneck without waiting for new generation. This supports continued compute scaling, a key input to AGI progress, though it adds no algorithmic advance.

AGI Date (no change): Freeing substantial existing grid headroom lets compute capacity come online faster than new power plants would allow, pulling the compute-scaling curve forward. Emerald AI's chief scientist notes it blunts but does not eliminate the need for new generation, limiting the magnitude.