Can AI answer the $3 trillion question?

Key Points:

• Sequoia partner David Cahn estimates that by 2026, AI infrastructure spending will reach $1.5 trillion, requiring the AI industry to generate $3 trillion in revenue to justify the investments in chips and data centers.
• Despite significant revenues reported by AI companies like Anthropic ($60 billion ARR) and OpenAI (up to $20 billion ARR in 2025), there remains a substantial revenue gap to cover the massive infrastructure costs.
• Major hyperscalers such as Google, Meta, Microsoft, and Amazon anticipate large free-cash flow increases by 2028, expecting returns on their extensive AI chip investments.
• Apollo’s chief economist Torsten Slok warns that if these hyperscalers fail to meet their cash-flow targets, it could trigger severe market consequences, potentially leading to an economic recession and a correction in the S&P 500.
• The growing use of cheaper open-weight AI models, often from Chinese sources, and improvements in token efficiency may reduce revenue for AI infrastructure providers, posing additional risks to the sector’s financial outlook.