- The Pulse by 42neurons
- Posts
- OpenAI's Codex: Performance in software engineering
OpenAI's Codex: Performance in software engineering
Hey there!
Welcome back to The Pulse, where we dive into interesting AI stories and trends backed by data, all presented through simple visuals.

> OpenAI launched new AI coding agent Codex based on o3 - optimized for software engineering
> can perform multiple tasks in parallel - writing code, debugging, answering database questions, etc.
> currently only available to Pro, Enterprise, and Team users & operates offline
> takes ~1-30 mins to complete tasks + can link to GitHub repositories
> although beats o3 at coding tasks, is slightly less safe according to OpenAI's tests

> most startups focus on clinical documentation, followed by medical coding
> current estimated valuations:
Abridge: $2.75B (after raising another $250M in Feb 2025)
Ambience Healthcare: ~$300M
Suki: ~$500M
Regard: $350M
> genAI healthcare market: expected $3.3B in 2025, projected to reach $39.8B by 2035

> cloud computing company currently has 250,000 GPUs in 32 datacenters, with Microsoft being the biggest customer (62% of revenue in 2024)
> first reported revenue since IPO beat expectations: $853M expected vs $981.6M in Q1 '25
> 737% revenue growth in 2024; projected 363% in 2025 at $4.9B-$5.1B
> signed $4B contract + five-year agreement worth $11.9B with OpenAI, who gets $350M in stock