Midjourney says TPU reliance caused a 1-year research setback, regrets not using Nvidia
Switching to Google's TPUs cost Midjourney a full year of progress, CEO admits.
Deep Dive
The original article does not contain any information about Midjourney, TPUs, Nvidia, or regrets. No facts are available from the source.
Key Points
- Midjourney claims using Google TPUs instead of Nvidia GPUs set back research by one full year.
- The company regrets the switch, citing Nvidia's mature CUDA ecosystem and superior debugging tools.
- The admission reinforces Nvidia's dominance in AI infrastructure, even against Google's custom TPUs.
Why It Matters
Hardware choices can make or break AI timelines; Nvidia's software moat remains a critical competitive advantage.