costs

Anthropic's IPO prospectus reveals a sweeping AI vision but highlights massive costs, including a $42 billion net loss in 2025 and plans for $518 billion in infrastructure spending, potentially valuing the company at over $2 trillion.

The article explores the financial burdens of AI services, citing major losses by companies like OpenAI and Anthropic, and posits self-hosting as a future solution despite current economic and technical challenges.

A user shares a breakdown of 116 paid products costing $730k per year, listing hosting, AI, dev tools, and other services, asking for cheaper alternatives.

After 3 months running AI agents in production across 3 SaaS products, the author shares what worked (GitHub MCP, Postgres MCP, Playwright MCP) and what broke (long tasks, auth walls, cost blowups, multi-tool orchestration errors), with a monthly cost of ~$430.

A survey of 6,000 executives finds that over 80% of companies report no productivity gains from AI, with a third of leaders using AI only 90 minutes per week, despite heavy investment. Meanwhile, Goldman Sachs predicts a 24x increase in token use by AI agents by 2030, raising cost concerns for companies like Uber and Microsoft.

This article highlights the difficulty of switching an LLM's memory layer after extended production use, noting that memory lock-in can be more problematic than model switching due to accumulated claims and drift.

A discussion post about the high costs of running LLM agents, with users sharing frustrations and seeking advice on tracking token spending and improving efficiency.

agentsview is an open-source local tool for browsing, searching, and tracking costs across AI coding agents.