AI Doesn't Have ROI

2026-06-02 · Source: Ed Zitron's Where's Your Ed At · Field: Finance & Economics — Capital Markets & Investment Management, Corporate Finance & Treasury, Economic Analysis & Policy · Depth: Intermediate, extended

Summary

An analysis published on June 2, 2026, asserts that Artificial Intelligence (AI) lacks a measurable return on investment (ROI), citing escalating and unpredictable costs. Uber's COO found AI spending hard to justify, while one company reportedly spent \$500 million on Anthropic models in a month due to unchecked usage. GitHub Copilot's recent shift to token-based billing has also led to user frustration over rapidly depleted credits, even during promotional periods. The author contends that AI companies like OpenAI and Anthropic deliberately obscured the true operational costs of their services, leading to enterprises "freaking out" after a Q1 2026 transition to token-based billing. Unlike the Dot Com Bubble, the AI bubble is predicted not to leave behind useful infrastructure, as specialized hardware like NVIDIA Blackwell and Vera Rubin GPUs are ruinously expensive to run and difficult to repurpose. Claims of AI-driven job loss and "dark output" are dismissed as unsubstantiated, with a Bain & Co. study revealing 44% of companies fund future AI investments based on unmaterialized past savings. OpenAI CEO Sam Altman's response to cost concerns is criticized as evasive.

Key takeaway

For executives and investors evaluating AI initiatives, recognize that current AI spending often lacks clear ROI and carries significant financial risk due to obscured costs and unproven value. You should demand transparent, measurable economic impacts from AI deployments, rather than relying on theoretical benefits or unmaterialized past savings. Prioritize solutions with demonstrable cost-efficiency and tangible output, and critically assess vendor claims to avoid compounding unsustainable investments.

Key insights

The AI industry's economic model is fundamentally flawed, built on obscured costs and unproven ROI, leading to unsustainable enterprise spending.

Principles

• AI costs are inherently difficult to measure.
• Subsidized AI usage distorts perceived value.
• Specialized AI hardware lacks repurposing utility.

In practice

• Scrutinize AI vendor billing models.
• Avoid AI adoption based on unproven ROI.
• Question claims of AI-driven job displacement.

Topics

• AI Economics
• Return on Investment
• Large Language Models
• AI Cost Obfuscation
• GPU Infrastructure
• AI Bubble

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

Related on AIssential

• AI Is Too Expensive — AI's current economic model is unsustainable due to massive hyperscaler investments, high operational costs, and unproven profitability of AI labs.
• Exploring AI’s Value and Profitability — AI ROI is about augmenting human work and fostering innovation, not just incremental productivity.
• From AI Hype To AI Math: The Market Just Changed The Rules — AI investment focus has shifted from ambition to demonstrable economic returns and sustainable cash flow.
• IBM CEO: $8 Trillion AI Data Center Spend Yields Zero ROI — Unchecked AI data center capital expenditure risks zero ROI without fundamental shifts in efficiency and monetization.
• 😺 Over $2B in AI funding hit in a single news cycle — AI investment is shifting to enterprise verticals and agent infrastructure, with a global competitive landscape.
• I built a tracker of AI company spend vs revenue. Everyone is losing A LOT of money — Most AI companies are currently unprofitable, with significant spending outweighing revenue, while Nvidia profits.

Open in AIssential →

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