The pre-registered threat to the GPU-rental market has partially fired: the biggest demand is migrating to lab-led and hyperscaler-tier builds — OpenAI's Stargate shows ~7 GW and $400B+ announced over three years. But figures are announced, not contracted; financing unsecured; spend pledge cut from $1.4T to $600B with a Texas expansion dropped. Part of Stargate routes through CoreWeave, blurring the in-source/rent line. Scorecard: the rental layer is a genuine overflow valve — paid, useful — but valves earn fees, not moats.