HOT RADAR
agentHEAT 11.5°

AI CURATED HOT STORY · 11/20/2026

Uber Caps Claude Code Sessions at 400K Tokens to Curb AI Costs

1 reports archived1 independent sourcesupdated 11/20/2026, 09:12:45
Synthesis & Latest Updates
1 Sources Cross-Validated

After exhausting its AI budget, Uber caps engineers' Claude Code sessions at 400K tokens even on million-token models. Weekly agent requests grew 9.4x from February to August and weekly active users of its AI coding tools grew 7x, yet overall AI spending has stayed roughly flat since April, with cost per session down 52% from its June peak. Uber still runs MCP servers but routes agents through command-line tools tied to a central MCP gateway, avoiding the 50K-70K tokens needed to preload descriptions for 100-plus tools.

LATEST/After exhausting its AI budget, Uber caps engineers' Claude Code sessions at 400K tokens even on million-token models. Weekly agent requests grew 9.4x from February to August and weekly active users of its AI coding tools grew 7x, yet overall AI spending has stayed roughly flat since April, with cost per session down 52% from its June peak. Uber still runs MCP servers but routes agents through command-line tools tied to a central MCP gateway, avoiding the 50K-70K tokens needed to preload descriptions for 100-plus tools.

Hourly heat curve

6 fully observed hours
Now
11.5
Peak
13.010:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first
  1. Rohan Paul (@rohanpaul_ai)T2·71 pts
    • Uber caps Claude Code sessions at 400K tokens despite million-token model capacity
    • Weekly agent requests grew 9.4x from February to August while AI spending stayed flat since April; cost per session fell 52% from peak
    • Agents reach MCP servers via CLI tools behind a central MCP gateway, avoiding 50K-70K tokens to preload 100-plus tool descriptions