Theo Questions Claude.ai’s Idle GPU Use

Theo says an idle Claude.ai tab uses a surprising amount of GPU capacity on his laptop.

TBTheo BrowneSpeaker · Ping Labs
Channel
Theo - t3․gg
Published
Aug 6, 2026
Video
0:13
Read
1 min
Original
English
Open on YouTube
Scroll to read and the video shrinks to the bottom-right corner. Captions follow your reading language.
Reading languageAvailable in 2 languages
You're reading the original English.
SUMMARY

In 30 seconds

Theo Browne criticizes Claude.ai’s apparent idle resource usage. He reports that each empty tab uses about 10% of the GPU on his $8,000 laptop; the short clip does not establish a cause or whether the result generalizes.

  1. 1An idle-page complaintTheo criticizes the resource use he observed on an empty Claude.ai page.
  2. 2His reported measurementHe says each idle tab uses about 10% of the GPU on his $8,000 laptop.
  3. 3Limited evidenceThe short clip does not establish the cause or whether other setups behave the same way.

Takeaways

  1. Theo reports high GPU use from idle Claude.ai tabs on his laptop.
  2. His estimate is around 10% of the GPU for each tab.
  3. The clip does not identify a cause or establish results across other devices and browsers.
FULL TEXT

Read the edited text by chapter

Transcript only
01

A complaint about idle resource use

Theo describes a GPU-use figure he observed with Claude.ai tabs. He criticizes the apparent cost of leaving an empty page open.

I think Anthropic needs better engineering here: an empty, idle Claude.ai page uses about 10% of the GPU on my $8,000 laptop for each open tab. That is my observation on my setup; this short clip does not establish the cause or whether the same result occurs on other devices and browsers.

More from this channel

5 slides38:47EN
Theo - t3․gg· Oct 28, 2026Theo’s Long-Running Agent Experiments: Verification, Compute and CostTheo Browne · Ping LabsA skateboard game and two compiler projects show how repeated checks and flexible compute can sustain agent work—and how expensive those loops can become.Coding agentsAgent engineeringAI cost & performance+2
5 slides36:07EN
Theo - t3․gg· Oct 27, 2026Theo on Token Costs, Coding Agents, and the Price of AI CapabilityTheo Browne · Ping LabsA viral estimate of Lauren's AI spending prompts a closer look at cache pricing, benchmarks, and the changing work of coding agents.AI cost & performanceCoding agentsInference+1
5 slides30:37EN
Theo - t3․gg· Oct 26, 2026Theo’s Three Skills for Reviewing and Shipping with Coding AgentsTheo Browne · Ping LabsHow sol-loop, full-send and full-stop encode review, verification and different levels of merge authority.Agent skillsCoding agentsCode review & verification+2
2 slides0:43EN
Theo - t3․gg· Oct 23, 2026Theo Distinguishes Security Vulnerabilities from Active AttacksTheo Browne · Ping LabsA short explanation of why disclosed software flaws and reported attacks are different kinds of security news.CybersecuritySoftware engineering
6 slides29:22EN
Theo - t3․gg· Oct 22, 2026Theo on Security Vulnerabilities, Stolen API Keys and AITheo Browne · Ping LabsA look at the gap between vulnerability disclosure and patching, and a speculative account of why stolen inference keys may be valuable.CybersecurityAI securitySoftware engineering+1
4 slides31:13EN
Theo - t3․gg· Oct 20, 2026Theo on Why Smarter Models Help More Than DevelopersTheo Browne · Ping LabsFrom game modifications to compiler experiments, Theo revisits his assumptions about who benefits from more capable AI.Coding agentsDeveloper productivityHuman-AI Collaboration+1