Nicolas Sauvage was never chasing the headline deals. While other investors crowded around flashy AI applications and consumer chatbots, he was writing checks into the plumbing. The infrastructure. The stuff that makes everything else actually work.
Since 2019, Sauvage has been building a portfolio anchored in the less glamorous corners of AI. Chips, tooling, data pipelines, the foundational layers that most VCs historically found too technical, too slow, or simply too dull to get excited about.
That calculation is shifting. Over the past year, the venture community has started paying serious attention to exactly the kinds of companies Sauvage backed years ago. What was once considered niche has become a priority as enterprises move past the experimentation phase and start demanding reliable, scalable AI systems that actually hold up in production.
The timing is not accidental. As AI deployments have matured, the bottlenecks have become clearer. It turns out that building a capable model is only part of the challenge. Getting it to run efficiently, cost-effectively, and at scale requires a whole stack of technologies that did not exist five years ago and are only now being built properly.
Sauvage's thesis was essentially that the boring parts would become critical parts. That has proven out. His portfolio now reads like a map of the infrastructure gaps the industry is scrambling to fill.
It is a reminder that in technology investing, being early to something unglamorous can be more valuable than being early to something exciting. The exciting stuff gets crowded fast. The foundational work tends to compound quietly, until suddenly everyone needs it and there are only a handful of serious players in the room.
Sauvage appears to have made sure his companies are among them.
