Nicolas Sauvage Is Betting On The Boring Parts Of Ai

Nicolas Sauvage believes it takes 4 years for nan champion bets to look evident — reasoning that he shared connected shape past week astatine StrictlyVC’s San Francisco event, which TDK Ventures co-hosted.

It’s a mentation he’s been moving to beryllium since 2019, erstwhile he founded nan firm task limb of nan Japanese electronics giant, which is now managing $500 cardinal crossed 4 funds. The AI spot startup Groq, weighted astatine $6.9 billion during its astir caller backing information past fall, is nan highest-profile illustration of this thinking.

In 2020, good earlier nan generative AI roar made infrastructure bets look obvious, Sauvage wrote a cheque into nan company, which was founded by Jonathan Ross — 1 of nan engineers who built Google’s Tensor Processing Units. Groq was focused from nan commencement connected inference: nan computational dense lifting that happens each clip a exemplary responds to a query. Ross had designed his spot by building nan compiler first, stripping nan architecture down until, arsenic Sauvage describes it, “you can’t region 1 portion and person it still work.”

It mightiness person looked niche to some, but knowing what he did astir his genitor company’s constraints, Sauvage saw asymmetry. Unlike user hardware, which has a earthy ceiling, request for conclusion keeps compounding pinch each caller exertion and each caller model. Sauvage couldn’t cognize past that request for conclusion would detonate this year, acknowledgment to each AI supplier that plans and acts crossed dozens of calls (where a azygous query utilized to suffice).

But successful immoderate ways, Ross was besides making a bet, too. After all, a Japanese electronics conglomerate champion known for magnetic portion is not, connected its face, nan astir evident investing partner. In fact, Sauvage describes TDK Ventures’ ain beingness arsenic very unlikely. But aft 2 back-to-back Stanford lectures — 1 making nan lawsuit for firm VC, 1 cataloguing each logic it fails — Sauvage, who is French and joined TDK successful Silicon Valley done an acquisition, sounded nan thought to Tokyo office contempt having nary evident opinionated to do so. (“I’m not Japanese. I don’t speak Japanese; I don’t unrecorded successful Tokyo.”)

After refusing to return nary for an answer, he yet received nan greenish ray successful 2019 to build a money whose instruction was to reply 1 question: What’s nan adjacent large point for TDK, and what mightiness termination it?

Image Credits:Slava Blazer for TechCrunch/StrictlyVC /

The portfolio he has since assembled is dotted pinch technologies that person go much wide absorbing to VCs complete nan past year: solid-state grid transformers, sodium-ion batteries for information centers, replacement artillery chemistries that sidestep nan geopolitical fragility of lithium and cobalt.

The subject down each of it is nan same: place nan bottleneck 4 years out, past find nan founders already moving connected it.

The question, of course, is what’s next. For his part, Sauvage is watching beingness AI intimately — not each of robotics but robots pinch a highly circumstantial occupation to beryllium done. Agility Robotics, for example, successful his portfolio, focuses connected nan single, mundane task of moving things from 1 spot to different successful warehouses facing workforce shortages. Another portfolio company, Swiss portfolio ANYbotics, builds ruggedized robots for environments excessively hazardous for quality workers — places wherever nan occupation meaning is fundamentally to spell wherever group can’t. The through-line is clarity of purpose. The robots Sauvage is betting connected don’t effort to do everything; instead, they do 1 difficult point reliably.

Sauvage says he’s besides watching nan compute stack displacement again. GPUs dominated training — nan massive, parallel computation of school a model. Inference chips for illustration Groq’s are reshaping what happens erstwhile that exemplary speaks: faster, cheaper, astatine scale. Now, Sauvage argues, CPUs are owed for a renaissance. They’re not nan astir powerful chips aliases nan fastest. But they’re nan astir elastic and champion suited to nan branching, decision-making logic of orchestration. When an AI supplier delegates a task, checks connected its progress, and loops backmost crossed dozens of steps, thing has to negociate nan full choreography. That something, increasingly, looks for illustration a CPU.

And past there’s China. A caller study from Eclipse — a task patient he follows intimately — documented what Sauvage describes arsenic “vibe manufacturing” — nan rapid, AI-assisted loop of beingness hardware prototyping, mirroring what vibe coding did for software. Chinese manufacturers, nan study found, are compressing nan design-build-test rhythm for beingness products successful ways Western proviso chains aren’t yet equipped to match.

For Sauvage, it’s a bottleneck awesome — and 1 he’s already moving connected pinch TDK Ventures’ various investments. One remaining unsolved problem, he says, is dexterity. Models are improving accelerated capable that beingness AI feels inevitable; what’s still missing is nan beingness fluency to match. The countries and companies that fig retired really to iterate connected atoms arsenic accelerated arsenic others iterate connected codification will person a manufacturing advantage.