A Framework for AI Monetization

UBS defines generative AI as a shift toward a token-driven economy currently facing a compute deficit where token demand exceeds supply. The report identifies four key demand buckets—existing business models, agentic AI, consumer apps, and training—and recommends broad positioning across the AI supply chain.

Key Takeaways

• 1.The fundamental unit of generative AI consumption is the 'token', and a 'compute deficit' persists where demand for tokens outstrips supply.
• 2.AI token demand is driven by four key buckets: acceleration of existing models, enterprise AI agents (agentic AI), personal AI, and training compute.
• 3.Capex for major AI spenders is projected to reach USD 821bn in 2026, growing to USD 986bn in 2027.

Table of Contents

• The acceleration of existing business models
• The emergence of agentic AI and its ramp
• Consumer markets need further maturity until we reach personal AI assistants
• Training demand and capex remain well supported by incremental model releases
• Appendix
• Required Disclosures
• Equity selection system
• Equity selection: An assessment relative to a benchmark
• Global asset class preferences definitions
• Statement of Risk
• Risk information

Document Preview

Page 1 of 5

Subscribe for full accessStart Free Trial