I Backed Jonathan Ross Before—And I’d Bet on His Advice Again

Seven years ago, I invested in Groq before they even had a product. At the time, it was a highly contrarian bet. Almost no VCs were investing in AI, let alone focusing on "AI inference." But my thesis was simple: the acceleration of large machine learning models was inevitable, and therefore, the demand for training these models would skyrocket. However, the second order implication - that many overlooked - was that inference, the need to run these models in real-time for real applications - would be just as critical. I saw the need for new, optimized hardware infrastructure to support not just model training but also low-latency inference, which would be essential for deploying AI applications at scale.

But the real bet?

Jonathan Ross. A visionary founder who could see the future before most others and who actually had the ability to execute against this vision. He wasn’t just building for the AI landscape of today, he was shaping the AI landscape of tomorrow.

Fast forward to today, and Groq is now one of the most talked-about AI companies. Recently, Jonathan shared some invaluable insights for founders on a 20VC podcast. Here are his five biggest takeaways:

1. Position Yourself for the Wave

Don’t chase what’s already obvious. Instead, find a large, unmet customer need that no one else is focusing on yet.The best opportunities emerge when technology unlocks something new and underserved.

Groq saw inference as a bottleneck before most customers even knew it was a problem!

2. Have a Mission That’s Much Bigger Than You

A truly ambitious mission fuels success. It attracts top talent, capital, and the resilience needed to win. When your mission is big enough, there is no choice but to be successful.

Groq believed that scaling inference was the key to ensuring AI wouldn’t be monopolized by a few dominant players. High stakes, high conviction.

3. Ask Yourself: What happens when something scarce or expensive suddenly becomes abundant and cheap?

This is one of the first questions I ask my founders in this new AI-era.

You need to skate to where the technology puck is going, not where it is now. As AI models get better, does your product get much better or commoditized?

Groq’s core insight was that AI was improving faster than the 2x that Moore’s Law suggested. They realized that the output of chips was doubling every two years, compounding into a 4x improvement. Instead of relying on expensive High Bandwidth Memory which Nvidia had cornered from a supply perspective, Groq completely re-architected its chips to keep model parameters live on-chip, turning computation into a more efficient pipeline. The benefit? Massive speed and scale at much lower costs.

4. If You’re Competing, You’re Already Behind

If you’re in direct competition, someone has already solved the problem. The biggest wins come from seeing what others don’t.

When Groq launched, most AI chip startups were fixated on Nvidia’s stronghold in training, a highly lucrative market. But Groq focused on the less obvious inference market which was considered “lower-margin” and niche, at the time. Seven years later, Groq still has minimal competition, and even Nvidia doesn’t have the right product for that space.

5. Becoming One of The “Mag 7” vs Staying a “Mag 7” are 2 Different Things

There are 3 stages on the road to becoming one of the “Mag 7.”

A. Solve an urgent, unmet customer problem better than anyone else.

B. Build world-class GTM execution to acquire customers and own the market.

C. Establish a defensible moat (ideally, one of the “7 Powers”).

But eventually, staying at the top means solving the next big customer problem. Even if it means disrupting yourself.

It remains to be seen if Groq will become one of the “Mag 7,” but my money is on Jonathan.

Build for the Future, Not Just for Today

I believe that the next quantum leap in AI will come from building insanely useful and usable products that impact the 7 billion people who don't currently truly benefit from technology today. I call this "AI for the real world."

If you’re a founder building in AI and thinking at this scale, let’s connect.