Please find Sapphire Ventures’ latest Market Memo: The Token Economy – AI Costs in Focus.

This month, we explore how AI economics are evolving as the industry shifts from building models to deploying them at scale. While infrastructure investment and model training costs continue to climb, enterprises are increasingly focused on the economics of inference, token consumption, and AI ROI.

Here’s a preview:

• AI infrastructure spending remains unprecedented: Hyperscalers are on pace to invest more than $700B annually, with ecosystem-wide commitments approaching $1T per year.
• Training frontier models is increasingly expensive: The latest generation of models may require $5B+ per training run, highlighting the growing costs of pushing the AI frontier.
• The token economy is exploding: Token consumption is growing exponentially as agentic workflows, reasoning models, and enterprise AI adoption accelerate.
• AI costs are becoming a boardroom conversation: Variable usage, pricing, and model selection are creating new challenges around budgeting, forecasting, and measuring ROI.
• Optimization is the next phase of adoption: As organizations move beyond experimentation, model diversification and cost efficiency are becoming critical priorities.