Don’t call it “deep tech”
They’re the specialists. We are the generalists.
Every few years I rear my head up from working with founders and talk to a bunch of Limited Partners. It’s always interesting to find out where the views of those allocating capital at the highest levels are relative to the startup ground game.
Last time around (2021–2022) historically low interest rates had everything going to the moon, climate tech was all the rage, SpaceX and Anduril were gaining attention, AI hype was starting, and hardware and biotech startups were SPACing left and right. This had many LPs seriously considering “deep tech” for the first time.
I’ve honestly never loved that term, or at least categorizing Cantos as such, but we’ve mostly embraced it out of convenience––to date. LPs tend to put us in their deep tech bucket, anyway. I don’t really blame them considering our portfolio mostly looks like this…
The phenomenon I’m noting most when talking to LPs in 2025––aside from the massive liquidity crunch––is that most of them now have a deep tech bucket! That’s pretty cool for those of us who believe the best way to solve the world’s biggest problems is to build in the real world, right?? I thought so at first, but now I’m not so convinced.
Having a “bucket” means the area so-labeled is necessarily constrained. Many LPs will talk about their deep tech “exposure” having invested in just one or two such firms. I find this relegation of the most interesting and obvious technological advancements cognitively dissonant for a few reasons.
First, because the original venture-backed companies were hardware-centric. The history of venture capital is intrinsically tied to the likes of Fairchild Semiconductor, Digital Equipment Corporation, Teledyne, Tandem Computer, Apple Computer, Intel, 3Com, Cisco, and Genentech. Software came later, enabled by hardware––from semiconductors to servers.
It wasn’t until the advent of cloud computing in the mid-2000s (AWS launched to third parties in 2006) that software as we conceive it became synonymous with “tech” and came to predominate venture capital. Fast forward 10–20 years and entrepreneurs building in the real world started to look like oddballs, but it is software-only companies that are actually the historical anomalies.
Second, it occurs to us that two-dimensional products are narrower in scope than those with a z-axis. I’ve started asking LPs whether they think a firm with investments across mining, oncology, lumber, defense, food, nuclear power, fertility, robotics, shipping, bio-sensors, computing, aviation, and chemicals is more or less constrained than those that limit their investments to screen-based products. They’re the specialists. We are the generalists.
Third, it’s fairly obvious to those of us on (or at least near) the ground in 2025 that the biggest economic opportunities are IRL. This is by no means limited to “deep tech” investors. Where is the cutting edge of AI going after chatbots? It’s robotics or “physical AI” and biology… at least according to Elon Musk, Yann LeCun, Sam Altman, and Jensen Huang. Many of the smart “generalists” appear to believe this, with massive investments in startups like Physical Intelligence (Thrive, Khosla, Lachy Groom, Redpoint, Sequoia), Skild AI (Sequoia, Coatue, Lightspeed, Menlo, CRV), and Fei Fei Li’s World Labs (a16z, NEA). (We have an unannounced investment in this space.)
Fourth, acknowledging the importance of real-world technology (“real tech”?) isn’t limited to private markets. One of the most crowded trades on Wall Street––or at least on Reddit––is nuclear power for AI data centers, even after DeepSeek took some air out of them. (Look at Oklo, NuScale, Constellation Energy, and Nano Nuclear’s stock charts over the past few quarters.)
Those are small cap equities, however (aside from Constellation at $22B), so let’s look at this “bucket” top-down. Hey Perplexity, what are the 20 largest publicly traded companies?
Only a third of those are predominantly software companies, and that’s including Visa and Mastercard. If Apple, NVIDIA, Tesla, TSMC, or Broadcom were seed-stage startups today, would we give them a cute label like “deep tech” given their 3-dimensionality and cordon them off to a subset of a subset of our overall portfolio?
So here’s my call to my fellow “deep tech” investors…
Stop calling it “deep tech”! Don’t squeeze yourselves into someone else’s bucket. Flip the script. You are the true generalists, practicing old school venture capital, backing the most ambitious founders building tomorrow’s most important companies. The future of technology is in the physical world, not confined to our screens. Innovating in the real world isn’t a bucket. It’s the main thing, the star of the show.
Deep tech? Never heard of her. We’ve used “near frontier” for years but that isn’t quite right either. Real tech? Warmer, but how about this:
Venture capital. Has a nice ring to it, don’t you think?
Don’t call it “deep tech” was originally published in Cantos Ventures on Medium, where people are continuing the conversation by highlighting and responding to this story.
very true! the point about fairchild semiconductor and intel being "deep tech" by today's definitions is the one that reframes everything