// resource constraints

All signals tagged with this topic

Apple's AI Token Cap Exposes the Waste Problem

Apple's decision to limit token consumption across its AI operations shows that compute efficiency has become a board-level concern at scale. The economics of training and serving large language models are forcing even well-capitalized tech giants to weigh model sophistication against operational cost. Smaller companies without Apple's infrastructure will face sharper decisions about whether to build or buy AI services. This constraint may accelerate vendor consolidation, pushing more organizations toward APIs and edge models rather than custom implementations.

London's AI data centers strain housing, power, and water

London's position as Europe's premier data center hub is colliding with the city's basic infrastructure limits—power grids, water systems, and residential space are already under pressure before the next wave of AI compute demands arrives. Local governments and utilities are actively negotiating trade-offs between hyperscaler expansion and livability. This tension reveals that proximity to capital, talent, and existing networks no longer automatically wins site selection when power and water become the limiting resources.