Bottom-up technology for everyone — why AI should cost $2, not $20

Bottom-up technology for everyone — why AI should cost $2, not $20 There's a pattern in tech history that keeps repeating. First, a new technology emerges. It's expensive, accessible only to well-funded companies and wealthy individuals. Then, over time, costs drop. The technology becomes democratized. Everyone gets access. But with AI in 2024–2026, something different is happening. The costs are dropping — rapidly. But the prices aren't. The economics don't add up Claude Haiku — one of the most capable AI models available — costs roughly $0.25 per million input tokens to run. A typical power user sends maybe 50,000 tokens per month in chat messages. That's $0.0125 in raw API cost per month per user. ChatGPT charges $20/month. That's a 1,600x markup . This isn't about covering costs. This is about extracting maximum value from users who have no alternatives — or who don't know alternatives exist. Who pays the real price For a developer in San Francisco earning $180,000/year, $20/month