Nicolas Sauvage’s Vision for AI and Venture Capital
Nicolas Sauvage, the founder of TDK Ventures, posits that it takes four years for the most promising technological investments to become apparent. This perspective was shared during a recent talk at StrictlyVC’s San Francisco event, co-hosted by TDK Ventures.
Building a Corporate Venture Fund
Since its inception in 2019, TDK Ventures has expanded to manage $500 million across four separate funds. One of its standout investments is in Groq, an AI chip startup valued at $6.9 billion following its latest funding round. Remarkably, Sauvage recognized the potential of Groq early in 2020, even before the generative AI surge highlighted the importance of infrastructure investments. Groq was founded by Jonathan Ross, an engineer instrumental in developing Google’s Tensor Processing Units, and specializes in the realm of inference—essentially, the computational tasks that occur every time a model responds to data.
Ross meticulously designed his chips, starting with the compiler to ensure that any part removed would compromise functionality. While some investors may have perceived Groq’s focus as niche, Sauvage identified a significant opportunity. Unlike consumer hardware that often faces inherent limits, the demand for inference grows consistently with the introduction of new applications and models. Unexpectedly, the demand for inference has surged in 2023, driven by AI agents that tackle complex tasks across multiple queries rather than single prompts.
An Unlikely Partnership
Interestingly, a Japanese electronics company renowned primarily for its magnetic tape might not appear to be a natural ally for a cutting-edge AI startup. Sauvage describes the formation of TDK Ventures as a very unlikely scenario. After attending a couple of influential lectures at Stanford that analyzed both the strengths and weaknesses of corporate venture capital, he envisioned a fund aimed at addressing a critical question: what innovations are essential for TDK’s future, and what might threaten its viability?
After persistent advocacy, he won the green light to establish this fund—a persistent pursuit, especially given his lack of Japanese language skills or direct ties to Tokyo.
Innovative Technologies in TDK Ventures’ Portfolio
Sauvage’s portfolio features a plethora of technologies that have gained increased attention among venture capitalists over the past year. These include solid-state grid transformers, sodium-ion batteries intended for data centers, and alternative battery chemistries designed to circumvent the geopolitical issues surrounding lithium and cobalt supply chains. The guiding principle throughout these investments has been to pinpoint bottlenecks projecting four years ahead and align them with founders already tackling these challenges.
What’s on the Horizon?
As for future trends, Sauvage remains vigilant about the emergence of physical AI—not all robotics but specifically robots designed for dedicated tasks. His portfolio includes Agility Robotics, which specializes in moving items within warehouses facing labor shortages, and ANYbotics, which builds rugged robots capable of operating in hazardous environments that humans cannot safely enter. The common thread is clarity of mission; these robots are engineered to perform one specific function with high reliability.
Additionally, Sauvage foresees a transformation in the computing architecture landscape. While GPUs have historically dominated model training, he believes that inference chips, such as Groq’s, will reshape performance metrics when models operate. With the evolution of AI capabilities, he argues that CPUs, typically seen as less powerful than GPUs, could be poised for a renaissance due to their flexibility and suitability for complex decision-making processes that require orchestration across various tasks.
The Competitive Landscape and Future Manufacturing
A recent report from Eclipse, a venture firm Sauvage closely monitors, highlighted a phenomenon he refers to as “vibe manufacturing.” This term captures the rapid, AI-assisted iteration in the prototyping of physical hardware that mirrors the agile development seen in software. Notably, Chinese manufacturers are achieving quicker design-build-test cycles than their Western counterparts, raising the stakes for innovation in hardware manufacturing.
Sauvage suggests that the ability to iterate swiftly in the physical domain could provide a competitive manufacturing edge to the companies and countries that master this art. Current limitations in dexterity present a challenge; while AI models improve rapidly, physical AI innovation still faces hurdles in achieving the necessary fluency. The firms that can solve these problems are likely to emerge as front-runners in the manufacturing landscape.
For a deeper insight into Sauvage’s journey and TDK Ventures’ strategic direction, check out the full article here.
Image Credits:Slava Blazer for TechCrunch/StrictlyVC /
Image Credit: techcrunch.com
