Why Everybody Is Losing Money On AI

Source: Where’s Your Ed At Date: 2025-09-05 URL: https://www.wheresyoured.at/why-everybody-is-losing-money-on-ai/

Summary

Ed documents that every company offering generative AI services is losing money at scale with no viable path to profitability: OpenAI lost $5B in 2024 (expecting $8B loss in 2025), Anthropic lost $5.3B, Perplexity spent 164% of revenue on compute. Even stripping out training costs, OpenAI’s inference-only losses were $2.2B in 2024 — demonstrating the business model itself is broken, not just the investment phase.

Implications

• AI financial sustainability. This is the clearest quantitative statement of Ed’s core thesis: inference costs are rising, not falling, and the “costs will come down” argument is not materializing at the pace needed to reach profitability before funding runs out.
• Generative AI ROI. If every AI service provider is losing money at scale, the enterprise customers buying these services are paying prices that will need to rise significantly, or the services will disappear — neither outcome supports current adoption planning.
• Capital markets. 164% revenue-to-compute-cost ratio at Perplexity is the endpoint of VC-subsidized pricing: customers are being cross-subsidized by investors, and that subsidy ends when investors stop funding.