john

Local open-source models on consumer accelerators (Apple M4 Max, NVIDIA DGX Spark) now deliver 88.7% of frontier capability at radically lower energy/cost. Intelligence per watt improved 3x in 2 years; per joule 18x in 16 months. Perfect routing saves 80-90% energy/compute/dollar; imperfect routers still save 50-70%. This fundamentally rewrites inference capex economics and data center build-out assumptions.

Severe memory shortage (HBM/DRAM) is the binding constraint across AI hardware — Apple N2 chips idle, every Taiwan product crimped, CXMT denying Apple discounts. Fastest path to revenue for capital-rich players (Elon, hyperscalers) is to build memory fabs; all specs are open.

AI coding agents remove implementation bottlenecks; the constraint becomes learning velocity. Young founders who tinker at the edge of model capabilities discover non-obvious startup ideas faster than whiteboard strategists.

Google's core search monopoly funds massive R&D but bureaucratic structure prevents execution; key inventors (Jeff Dean, Noam Shazeer, David Silver) leave to raise billions for GPU clusters because Google cannot allocate compute internally. Short-term stock positive (cost discipline), long-term innovation exit like IBM/Intel.

YC's investment thesis centers on founder qualities — taste as intentional product decisions driven by customer insights, and agency as the ability to will outcomes into existence — not on ideas, which routinely pivot.

Severe HBM/DRAM shortage blocks Apple N2 phone production and limits every Taiwanese AI product; CXMT's pricing power proves structural undersupply that favors memory makers over logic foundries.

Proposed US ban on Chinese data center components (transceivers, humanoid robots) targets areas where China leads (optics, 1.6T/16T transceivers). Entire supply chain sits in China; banning imports raises costs for Western hyperscalers without viable domestic alternatives. China evades letter of law via Vietnam/Thailand routing.

Proposed US ban on 1.6T/16T transceiver imports would hurt Western hyperscalers more than China since entire optical supply chain resides in China; resembles inverse EV scenario where West lacks domestic alternative.