Gemma 4 shipped Apache 2.0 in April — a 31-billion parameter open model that matches proprietary systems twenty times its size. METR’s updated time-horizon methodology now shows AI capability doubling every 89 days since 2024. Truly open, self-evolving agent harnesses are here.
The $650 billion concentration thesis rests on assumptions that no longer hold the way they did eighteen months ago — and the infrastructure bets placed on those assumptions are starting to show it. MIT Technology Review is now writing about what most capital allocators still haven’t internalized.
The Grove Foundation is a 501(c)(6) standards body publishing the open governance stack for distributed AI. The methodology is CC BY 4.0. The measurement framework — Λ — quantifies which AI architectural deployment patterns show structural viability.
The Autonomaton Pattern is the first published standard, with a detailed open source roadmap available to members. The thesis remains the same: the infrastructure future is more open, more distributed, and more contested than the headline-grabbing $650 billion bet assumes.