Why AI Companies Are Moving to Rust
9.8× channel medianLong-form5:20EducationPublished Oct 17, 2026, 8:08 AM · 2d ago
Views
50,751
+285 views/hour
Views / hour now
285
measured 44m ago
Peak views / hour
2.1K
16h ago
First 24 hours
—
not tracked from upload
Engagement
1.3%
likes + comments / views
Likes
578
1.1% of views
Comments
56
0.11% of views
Est. ad revenue
$355–$812
lifetime, $7–16 RPM
Checked every 3 h while the video is this age · last check 44m ago · next in 135 min. YouTube itself updates public counts with a short delay.
Avg views / day
19.3K
since upload
vs channel median
9.8×
this video's views ÷ the channel's usual
Rank in channel
#2
of 25 tracked
Views over time
24 readings
Total views Views gained per hour
Description
In September, GitHub said one engineer and a team of AI agents moved the Copilot agent runtime to Rust. Bun, the runtime under Claude Code, was rewritten in Rust with Claude agents. OpenAI's Codex CLI went Rust in 2025. But almost everyone in AI writes Python, so what's going on?
In this video we follow one prompt down the stack and look at where the Rust actually is: the agent CLI, the tokenizer, the model weights, the server front door and your Python tools. Then we look at why Python servers and C++ both run into trouble, why AI agents suddenly made big rewrites cheap, and where Rust doesn't help (Cloudflare's outage, compile times, Microsoft's 2030 walk-back).
Short version: Python stays where people think and try things, Rust takes the parts that run all day, and CUDA C++ still does the GPU maths.
All numbers are the companies' own, as stated in their posts:
- GitHub, Migrating the GitHub Copilot runtime to Rust: https://github.blog/ai-and-ml/generative-ai/migrating-the-github-copilot-runtime-to-rust-using-copilot/
- Bun, Rewriting Bun in Rust: https://bun.com/blog/bun-in-rust
- OpenAI Codex CLI is going native: https://github.com/openai/codex/discussions/1174
- tiktoken: https://github.com/openai/tiktoken
- Safetensors joins the PyTorch Foundation: https://huggingface.co/blog/safetensors-joins-pytorch-foundation
- vLLM Rust front-end RFC: https://github.com/vllm-project/vllm/issues/40846
- uv: https://github.com/astral-sh/uv
- What's new in Python 3.14: https://docs.python.org/3/whatsnew/3.14.html
- Chromium memory safety: https://www.chromium.org/Home/chromium-security/memory-safety/
- Google, Rust in Android: https://blog.google/security/rust-in-android-move-fast-fix-things/
- LWN, The state of the kernel Rust experiment: https://lwn.net/Articles/1050174/
- Cloudflare outage, Nov 18 2025: https://blog.cloudflare.com/18-november-2025-outage/
- The Register on Microsoft's 2030 post: https://www.theregister.com/2025/12/24/microsoft_rust_codebase_migration/
- Astral to join OpenAI: https://astral.sh/blog/openai
## Chapters
0:00 Three coding agents, one language
0:34 Follow one prompt down the stack
1:21 Python is fine, Python servers aren't
2:15 Why not just C++
3:06 Agents made rewrites cheap
3:58 Where Rust doesn't help
4:45 How it fits together
Tags
rustrust programmingpythonai infrastructuregithub copilotclaude codecodex clibunrust rewritetiktokensafetensorsvllmuvruffpydanticmemory safetypython gilfree threaded pythoncloudflare outageai agentscode bear
Details
- Published
- Oct 17, 2026, 8:08 AM
- Length
- 5:20 (320 s)
- Format
- Long-form
- Category
- Education
- Audio language
- EN-GB
- Quality
- HD
- Captions
- No
- Made for kids
- No
- Licensed content
- Yes
- Video ID
- kq7_2bDLtls
More from Code Bear
Reverse Engineering Reverse Anything (REA)463 views · 3h ago · +171/h
NCM: The LLM Killer?11K views · 7h ago · +1.5K/h
AI Is Flooding Open Source With Bug Reports!1.3K views · 17h ago · +74/h
My AI Influencer Reel Hit 100K Views: Here's How I Make the Videos1.5K views · 1d ago
OpenCode Got Hacked BADLY ( Explained! )157K views · 1d ago · +475/h
Higgsfield Ads Studio is CRAZY!2.5K views · 2d ago · +2/h
Tesla's Flying Car!4.1K views · 3d ago · +3/h
Tesla's Flying Roadster Explained: Suction, Air Jets and a Lot of Hype5.2K views · 3d ago · +6/h
Statistics from the YouTube Data API, tracked by FameX. Views per hour, revenue and comparisons are FameX estimates.