Contrary to the view that AI token intensity will drop after the initial coding boom, the move from simple queries to autonomous 'agentic' workflows will cause an order-of-magnitude (10x) increase in token usage per task. This applies across all knowledge-based jobs, ensuring sustained and explosive demand for compute.
The market often misinterprets AI progress as linear. However, a clear 'scaling law' dictates that a tenfold increase in the computing power used to train LLMs results in a twofold capability improvement. This exponential relationship means future advancements will be far more disruptive and surprising than incremental projections suggest.
The economics for enterprises adopting AI are incredibly favorable. A task costing $55 in human labor can be completed by an LLM for a fraction of the $5 cost of a million tokens. This massive arbitrage creates a powerful incentive for adoption and justifies large-scale infrastructure spending.
Despite rapid advances in AI models, the average corporate user has not yet caught up, creating a gap between capability and widespread implementation. This lag means the significant revenue inflection for hyperscalers' massive AI investments is not imminent but is more likely a 2026 event, once enterprise adoption matures.