The Great AI Replacement Hit a Spreadsheet: Microsoft and Uber Can’t Afford Their Own Agents

2026-07-06 · Source: Towards AI - Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Novice, quick

Summary

Microsoft, a major investor in Anthropic with a \$5 billion stake and a seller of its models, has begun removing Claude Code licenses for its own engineers, particularly within the Windows and Microsoft 365 division, by June 30. This move highlights a significant cost problem with AI agents: even the companies developing and selling these tools find them too expensive for internal use. Similarly, Uber's use of Claude Code rapidly depleted its full-year 2026 AI tools budget in approximately four months, prompting the company to cap spending at \$1,500 per employee per tool per month. Industry analysis from Gartner indicates that agentic tasks consume 5 to 30 times more tokens than standard chatbot interactions, underscoring the substantial operational costs associated with advanced AI agent deployment.

Key takeaway

For Directors of AI/ML evaluating internal agent deployment, recognize that even major tech companies struggle with AI agent operational costs. Your budget planning must account for agentic tasks consuming 5 to 30 times more tokens than chatbots. This high consumption can deplete annual budgets rapidly. Implement strict spending caps per employee and tool. Prioritize thorough cost-benefit analyses before widespread adoption to avoid unexpected financial strain and ensure sustainable AI integration.

Key insights

AI agents, despite their promise, incur prohibitive operational costs, even for their developers and major adopters.

Principles

• AI agent operational costs are high.
• Token consumption scales significantly with agentic tasks.

In practice

• Cap employee AI tool spend.
• Evaluate token usage for agentic tasks.

Topics

• AI Agents
• Operational Costs
• Microsoft Azure
• Anthropic Claude
• Uber AI
• Token Consumption

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

Related on AIssential

• What you'll pay for AI agents will be wildly variable and unpredictable — AI agents incur vastly higher and unpredictable token costs, primarily driven by input tokens and cache reads, without guaranteeing improved performance.
• Why Uber has Already Burned Through its AI Budget — Aggressive AI tool adoption can rapidly deplete budgets, necessitating cost management and strategic expansion.
• The Pulse: AI token spending out of control – what’s next? — Exploding AI agent token spend and vendor demand outstrips budgets, while Meta faces internal morale issues.
• When AI Costs More Than the Engineer — AI compute costs are rapidly outpacing engineer salaries, creating a new dominant cost structure for advanced AI development.
• Token Illiteracy Is the New Data Illiteracy — and It’s About to Cost You the Same — Token consumption is a rapidly growing, ungoverned enterprise cost center requiring immediate, cross-functional operational discipline.
• Microsoft launches AI agent with pay-as-you-go pricing — AI agent services are shifting to usage-based billing due to high, variable computational costs.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Towards AI - Medium.