AI's Economics Don't Make Sense [Ad Free]

2026-04-28 · Source: Ed Zitron's Where's Your Ed At · Field: Finance & Economics — Economic Analysis & Policy, Capital Markets & Investment Management, Artificial Intelligence & Machine Learning · Depth: Advanced, extended

Summary

The article argues that the economics of generative AI are fundamentally broken and unsustainable, leading to a "subsidy scam" where companies hide true costs. It highlights GitHub Copilot's shift to usage-based billing on June 1, 2026, as a key indicator. The author details how monthly subscriptions for LLM services like Anthropic's Claude and OpenAI's ChatGPT are unprofitable, with providers losing significant money per user (e.g., Microsoft losing over \$20/month per Copilot user, some costing \$80/month). The article also scrutinizes the economics of AI data centers, citing the \$52.8 billion Stargate Abilene project for OpenAI by Oracle, which is years behind schedule and relies on OpenAI's "ridiculous" projection of \$852 billion in revenue/funding by 2030. It concludes that the entire generative AI industry is unprofitable, unsustainable, and built on deception, with major players like OpenAI and Anthropic facing severe financial viability issues.

Key takeaway

For investors evaluating AI companies, recognize that current generative AI business models are largely unsustainable. The shift to token-based billing, exemplified by GitHub Copilot's June 1, 2026 change, exposes the true, high compute costs previously subsidized. Scrutinize financial projections, especially for companies like OpenAI and Anthropic, which rely on massive, unproven revenue growth to cover escalating infrastructure debt and operational losses. Your due diligence must account for the inherent unpredictability and high cost of LLM inference, which often lacks clear ROI.

Key insights

Generative AI's subscription model is economically unsustainable, masking massive compute costs and leading to inevitable price increases.

Principles

• LLM services are inherently unpredictable and costly.
• Monthly subscriptions fail with wildly variable compute costs.
• AI data centers require 100% utilization for profitability.

In practice

• Token-based billing reveals true, high LLM usage costs.
• Unprofitable AI startups rely on continuous VC funding.
• Data center projects face high debt and obsolescence risks.

Topics

• Generative AI Economics
• Token-Based Billing
• AI Data Center Costs
• OpenAI Financial Viability
• Anthropic Compute Deals
• Oracle Stargate Project

Best for: CTO, VP of Engineering/Data, Executive, Director of AI/ML, Investor, Tech Journalist

Related on AIssential

• AI's Economics Don't Make Sense — Generative AI's subscription models and compute infrastructure are economically unsustainable due to massive subsidies and hidden true costs.
• Anthropic will pay xAI $1.25B per month for compute — AI companies are adopting a "neocloud" model to monetize excess compute capacity and offset infrastructure costs.
• Nvidia VP Says AI Costs ‘Far’ More Than Human Employees — AI compute costs can exceed human labor, challenging economic viability despite massive tech investments.
• Anthropic is about to become the first profitable AI lab — Anthropic is nearing profitability driven by coding tools and agentic AI, despite increased user costs and compute demands.
• 📈 Why AI bills rise as costs fall — AI costs escalate despite falling token prices due to demand elasticity and agents' hidden, amplified token consumption.
• I think Anthropic and OpenAI have found product-market fit — Frontier AI labs have found product-market fit for coding agents, driving significant enterprise revenue through aggressive pricing.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Ed Zitron's Where's Your Ed At.