Betting on the 'boring' parts of AI — my take on inference and narrow robotics

Friends, I want to share an observation from AI and robotics.

Nicolas Sauvage on StrictlyVC described TDK Ventures' playbook: identify narrow, scalable bottlenecks with ~4-year horizons and back them.

- Example: an early bet on Groq — inference chips for faster, cheaper model execution.
- Portfolio: single‑task robots (Agility, ANYbotics) and alternative energy technologies.
- Trend: after GPUs and inference chips, CPUs are returning as orchestration layers; rapid Chinese prototyping is changing the game.

Why it matters: these solutions scale with application growth and need capital now.

How do you view bets on 'single‑task' physical AI?

#AI #robotics #investing #hardware