/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

A programmer estimates his typical day of coding with Claude Code is equivalent to running the dishwasher an extra time, much more energy than a “median query”

Most of the discourse about the environmental impact of LLM use focuses on a ‘median query.’ What about a Claude Code session?

Simon P. Couch

Context & Ripple Effects

This estimate lands as Claude Code has moved from a niche coding assistant toward sustained, high-intensity use: an earlier hands-on found it could tackle difficult legacy-code work, while noting form-factor limitations in the early Claude Code experience.

The distinction between a single prompt and a coding session matters more as Claude’s audience expands and Claude Code capabilities are being carried into Cowork. Claude’s rapidly growing web audience makes session-level resource accounting a more relevant question than headline figures for a “median query.”

First-order effects

  • The estimate reframes the environmental footprint of Claude Code around a day-long, tool-using workflow rather than an isolated LLM request; developers and teams evaluating AI use now have a more relevant, if still anecdotal, comparison point.
  • It exposes how easily per-query energy claims can understate the cost of coding agents when they are used iteratively or continuously.

Second-order effects

  • AI vendors and enterprise buyers face pressure to distinguish lightweight chat usage from long-running agent workloads in their energy reporting and procurement assessments.
  • Usage controls become part of the resource discussion: Anthropic’s planned limits targeted the small share of users running Code continuously in the background, as described in its rate-limit plan for intensive Claude Code use.

Third-order effects

  • If coding agents become a standard development interface, AI energy accounting will likely shift from model- or query-level averages to workload-level measures that capture duration, iteration, and background execution.
  • That shift could make efficiency competition less about a model’s single-response cost and more about how much compute an agent consumes to complete a software task; this estimate alone cannot establish typical usage across developers.

The trend: The broader trend is the migration from one-off chatbot interactions to persistent agent workflows, making per-query AI-impact metrics increasingly inadequate.

Discussion

  • @simonwillison.net Simon Willison on bluesky
    This is great - it's about time someone updated the discourse on LLM energy usage to reflect that coding agents use massively more prompts than occasional questions to ChatGPT  —  Simon estimates that a day of coding agent usage comes out close to the energy needed to run a dishw…
  • @niccrane Nic Crane on bluesky
    Excellent post about Claude Code's actual energy usage 💡  —  Most AI energy posts only look at single queries.  Simon breaks down full coding sessions - much higher, but still about the same as running a dishwasher once a day.  —  simonpcouch.com/blog/2026-01-20-cc- impact/ …
  • @simonpcouch.com Simon P. Couch on bluesky
    Whenever I read discourse on AI energy/water use that focuses on the “median query,” I can't help but feel misled.  Coding agents like Claude Code send hundreds of longer-than-median queries every session, and I run dozens of sessions a day.  —  On my blog: www.simonpcouch.com/bl…