We think a lot about venture investing. Our core thesis on the topic is that the US venture community has grossly underweighted hardware over the past ten years. Over the next ten years, 60% of the revenue from “technology” will accrue to the hardware sector, but over the past ten years, hardware has only attracted 10% of venture dollars. So in recent years we have spent a lot of effort searching for ways to bridge that gap, and take advantage of what should be a massive opportunity. Does AI change any of that?

The shot answer is no. All that AI needs to run on hardware, and most indicators point to AI needing a lot of hardware. So the shift to neural network based machine learnings reinforces our thesis. But there is a more complex, nuanced answer because the rise of AI represents a fundamental change in the foundations of all our technology, which hints at a profound shift in the economics of technology.

To be more precise, we see the opportunity not in hardware or semis, but in Deep Tech – which we admit is not a great term – but combines hardware and software into complete systems. We have written about this many times in the past, so it is gratifying to see the idea percolate through so many large companies today.

We see DeepTech opportunities arising along a number of axis.

At the lowest level, there is a need for many more sensors. The latest crop of large language models (LLM) are now morphing into large vision models (LVM), and however you want to term them, they are going to require immense amounts of input data. We see that already in the sensor fusion in cars. For LVMs to work, these sensors need to be very affordable and power efficient, fortunately there are many new companies springing up to make this possible.

In that vein, anything that can bring about material, orders of magnitude improvements to power consumption are critical. This includes batteries, new energy systems and sources, new materials and new connectivity methods.

Moving up the stack, we will also need a whole new category of processors. There are valid reasons why venture investors have shunned this sector, but when you do the math, the investment case looks surprisingly robust for certain types of chips. This will still be a challenging sector in which to invest, with very large competitors and even larger customers, but none of them are nimble, and there are many white spaces out there on which to build giants of the future.

At the highest level, the rise of AI is fundamentally altering the economics of data centers, industrial systems, robotics and many other markets. There are fortunes to made here but they are going to be hard to find. In recent weeks, we have spent a lot of time listening to presentations, keynotes and conference floor pitches. One thing is clear from this, no one really knows what is coming next. There is incredible work being down across the AI research landscape and traditional venture investing models have not had to dig this deep into fundamental technology in a very long time.

We know companies across all these sectors and know where to find more. The rise of machine learning, and the radical decrease in costs of these systems we call AI offers what has the potential to be a generational opportunity, and we are excited to be here for it. Please drop us a line if you would like to discuss or need some ideas.