Training-efficiency gains will reduce aggregate compute demand (threatening "more compute = more value").

◷ Share view, as of June 2026: confidence, rationales and priors are hidden. The questions, stances, claims, tests and evidence stay visible.

would falsifyFrontier capability stays tightly coupled to raw compute; efficiency gains don't dent aggregate demand.

Tail into Market.

Fed by claims (in either domain)

Evidence

19 support7 contradict

Supports· weight medium· Behind the Term Sheet: Moonshot AI (cathaycapital.com)· via Moonshot AI's K2.5 achieves frontier coding and agentic benchmarks at one-fifth the inference cost of leading closed models· Technical Capability
Neutral· weight medium· AI copyright battles enter pivotal year as US courts weigh ... (reuters.com)· via US courts set to issue landmark rulings on AI fair-use in copyright disputes spanning music, visual art, and text in 2026· Regulatory & Legal
Neutral· weight medium· The next chapter of the Microsoft–OpenAI partnership (blogs.microsoft.com)· via Microsoft secures 27% equity stake in OpenAI PBC at $135B valuation; extends model IP rights through 2032, decouples from AGI declaration timeline· Market & Valuation
Neutral· weight medium· Two California district judges rule that using books to train AI is fair use | White & Case LLP (whitecase.com)

Signals on the board

Research bearing on this · advisory

Neutral· weight medium· AI Infrastructure Spending Forecast to Be Over a Trillion Dollars Over the Next Five Years, According to Dell’Oro Group - Dell'Oro Group (delloro.com)· via Dell'Oro Group revises global data center capex forecast to 24% CAGR through 2028, projects over $1 trillion AI infrastructure spending over five years· Market & Valuation

Contradicts· weight medium· AI labs and investors organize around the possibility of recursive self-improvement as OpenAI's Lilian Weng leads new self-improvement research and Richard Socher's Recursive Superintelligence raises hundreds of millions (axios.com)· via AI lab leaders and investors organize around recursive self-improvement as institutional milestone, August 2026· Technical Capability