AI's Economics Don't Make Sense
GitHub Copilot is shifting to usage-based pricing starting June 1, 2026, moving away from fixed subscription plans to better align costs with actual AI model usage. This change reflects the increased computational demands of Copilot's evolution into a more advanced, agentic coding platform. The shift highlights broader economic challenges across the AI industry, where subsidized services are becoming unsustainable due to high inference costs.
- ▪GitHub Copilot will adopt usage-based pricing on June 1, 2026, charging users based on actual model costs rather than fixed request limits.
- ▪Microsoft has been losing an average of over $20 per user monthly, with some users costing up to $80 per month in compute expenses.
- ▪The move follows a wider industry trend where AI companies face unsustainable burn rates from offering heavily subsidized AI services.
- ▪AI services often obscure costs using tokens or messages, making it difficult for users to understand the true computational expense.
- ▪Many AI platforms, including OpenAI, Anthropic, and Perplexity, operate at a loss under current pricing models due to high underlying infrastructure costs.
Opening excerpt (first ~120 words) tap to expand
Hello premium subs! This is your ad-free free newsletter for the week. Questions? Queries? Email me at [email protected], and if you have a scoop, ezitron.76 is my Signal. Yesterday morning, GitHub Copilot users got confirmation of something I’d reported a week ago — that all GitHub Copilot plans would move to usage-based pricing on June 1, 2026. Instead of offering users a certain number of “requests,” Microsoft will now charge users based on the actual cost of the models they’re using, which it calls “...an important step toward a sustainable, reliable Copilot business and experience for all users.” Users instead get however much they spend on their GitHub Copilot subscription (EG: $19 of tokens a month on a $19-a-month plan).Translation: "we cannot continue to subsidize GitHub…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Ed Zitron's Where's Your Ed At.
