// infrastructure costs

Source: The Pomp Letter

As AI chip capacity expands, spending is consolidating around inference and deployment rather than model training—a shift that disrupts the traditional model where chip makers counted on sustained demand from data centers running expensive GPUs for R&D. Inference-optimized chip designs, edge deployment providers, and software layers handling model serving now compete for dollars that previously flowed to a handful of training-heavy hyperscalers. Price pressure on chip makers intensifies when customer spending fragments across dozens of inference use cases with different hardware requirements, not simply when capacity expands.

Source: The Verge

Matthew Prince is floating a radical restructuring of internet economics—reviving HTTP 402 (Payment Required) to make bot operators pay for the infrastructure they consume rather than socializing those costs across human users. If bot traffic truly reaches 1000x human levels as Cloudflare's data suggests, the current model where search engines and AI scrapers effectively steal bandwidth becomes unsustainable. The alternative is paywalls for bots or selective blocking, including Google. Edge computing platforms could shift from service providers to toll collectors, forcing AI trainers and search indexers to bear infrastructure costs now absorbed by human users.