Unruly Q3'26 LP Letter: Deeptech's Normalization

Don’t forget Unruly Futures is happening Nov 4th with still some spots left! Learn more at https://www.unrulyfutures.com/

—

TL;DR (although I hope the full post is worth a read):

• Everybody recognizes there are no more moats in software
• Deeptech hardware is instead all moats
• Software and hardware scaling costs are now similar, with AI even clearly more expensive
• Deeptech hardware costs are still substantial at scale, but much lower at seed/validation stage
• Large tech cos used to be the only ones that could afford these moonshots, now independent firms can try moonshots and large companies are the only ones left to do bleeding-edge AI
• Vanilla VCs moving en-masse to deeptech to chase outperformance
• We will have many more hardware companies: founders get meaningful validation in the market that deep hardware solutions can raise capital
• A lot of people will get very hurt
• Everything will get more expensive

—

Deeptech's Normalization

This might be the fastest the venture space has ever shifted (aside from insane fads like NFTs or similar), as if you closely look at today's venture funds target pipelines you can really only find two things: AI and deeptech. There’s nothing else that matters.

It's not a coincidence that VCs have started calling robotics and hardware intensive industries "physical AI". Makes it sound more software-y, when in reality it's pure deeptech.

I’ve manifested a feeling to my team members quite a few times over the past year: we’re not early to deeptech anymore. It has normalized. There’s a distinct feeling when you’re not early anymore. You start seeing the same ideas getting funded over and over again. You see large rounds happening with people you’ve never heard of. You feel overwhelmed with new companies to look at. Suddenly companies in your portfolio no one wanted to touch a couple years back can now raise very large rounds with decent success.

It’s really what we live for as early stage “contrarian” investors, but it becomes a bit of a problem if every other investor in the world now likes the same stuff you do (while also calling themselves “contrarian”, obviously).

What cemented it all for me, and I think for everyone else now, is Bessemer announcing their new fund family, “$5.75B to back what’s next”. I’m so old I remember a time when Bessemer was the most SaaSified of investors. When even them announce a new fund family by showing videos of self-driving cars, spaceships, network switches, maritime drones, space vehicles, Si/C batteries, hypersonic airplanes, inspection robots; all of it while using the tagline “forged in steel” you know that their software bets are now the odd ones out compared to the hardware ones.

How deeptech got normalized

If you look at the big generalist VCs portfolios, you can find anything from nuclear fusion to humanoids and space companies.

If you had looked just a few years ago, it would have been all calendar scheduling tools and AI-enhanced CRMs.

So what happened? There's a few things that happened, all in rapid succession.

Software moats got cratered by AI: investing in normal SaaS feels now as ancient as doing networking switch (although this is so old that it's now hot again!).

AI is a clear alpha-field. But AI is expensive as heck. All the megafunds are raising tens of billions, USV and Benchmark had to expand their fund sizes for the first time in decades.

This means that very few firms can compete to do the $50M pre-seeds that new AI-intensive companies require, and therefore need to look elsewhere for investment areas. But it's not going to be normal software.

Deeptech investing actually worked. Everybody wants to become the new Founders Fund and be early to the next SpaceX (obviously everybody forgets just how completely out there that investment was even for a future-forward fund like Founder Fund, but I digress). Everybody wants FF II's DPI and the validation of it coming from deeptech means they now have permission to do it as well and can defend large deeptech bets with their LPs.

Deeptech is getting cheaper. Just as AWS and similar infrastructure made it much cheaper to launch a startup, all of the recent advances in physical engineering made by the large corporates means that it's quite cheap to build a prototype of a drone or any other hardware enabled device.

China. None of this could have happened without Chinese factories. While a lot of companies are building sovereign solutions, Made-in-EU/US, the reality is that most of the components to enable cheap deeptech come from China and its epic contribution to batteries, displays, chips and so on.

Deeptech is what the world needs. There's also generally been a narrative shift in tech, surely helped by A16Z's American Dynamism, but also by the Russian War, the China threat, and all of the other various macro plays, that made all of us collectively realize just how much we neglected real hardware innovation in the long euphoria and inebriation generated by the software boom. Feels like a bit of a hangover now, and a wake up. As we wrote in our Q1 Letter.

So here we are, the result of all of this is that a lot of formerly "normal" VCs are getting very comfortable with deeptech investing.

What this means for small pre-seed deeptech venture funds

For a small fund like us, who’s been investing in hardware/deeptech since day one (and as an angel since 2018), this is a key moment to really sit down and reflect on the evolution of the space.

We could argue we might have moved into deeptech too early, but good things come to those who wait (in the right place, and not forever).

On the positive side: this is absolutely epic news. We have been making some pretty out there and very very deep tech investments in our Fund I, and the biggest question we had from LPs (and quite frankly often in our heads as well) was always: where is all the follow on capital going to come from?

Now suddenly the answer is incredibly clear: every generalist traditional software VC is a follow on target lead in our spreadsheets. Many of them have already led follow-ons in our companies, and we couldn’t be happier.

Our deeptech companies benefit from getting people on board that have seen what the company building pace for traditional software businesses should be, as well as the capital needed.

Where the problem starts to show is when, soon enough, this trickles down to the seed stage. As soon as the follow-on risk is removed, suddenly everyone will want to do the types of deals we’ve been doing for years - and a lot of people already are.

This means that:

1. we’re not so special anymore
2. there’s a lot more capital chasing the same deals (although the number of deals also grows for the same reasons, it does so slower than the amount of capital)
3. everything gets more expensive

While we have removed some risk (eg. follow on capital), the risk/reward profile can skew to a not-interesting level when everyone shows up to the in-vogue market at the same time.

But regardless, even assuming risk/reward someone stays balanced, everything just gets harder: more deals to sort through, harder to even see everything, more VCs to compete with, harder for all of these companies to assemble the dream teams they need, and so on.

What's next for deeptech investing?

Now, I want to make it very clear: I don’t think investing in deeptech is now EV negative. I strongly believe that a whole lot of the most valuable companies of the next decades will be deep tech companies.

One of the most interesting dynamics on the "mega" firms will be to see if the generalists will be able to make good decisions and adapt to this new frontier, or if the smaller "deeptech native" firms will scale to become the "mega" firms of tomorrow.

One easy prediction for us to make is that the mega firms in 10-20years will be mostly about deeptech investing.

But at the same time it's healthy to recognize that the actual outsized alpha at the very early stages is diminishing. Excess return is going to be harder to generate as returns will disperse through the market.

We believe we’ve built expertise, network and structures that can still provide us with alpha and EV-positive investing in deeptech at the seed stage for quite a while. But learning from the past, recognizing when suddenly it’s not as easy as it was has always served me well.

Fortunately, we're just built in a very specific way, and we've already started exploring where the new frontier moves towards.

We have already started to develop some ideas about it and have already started deploying heavily into it from our second fund, but don’t yet want to give a full blueprint to everyone else about it.

That’s the topic for another quarterly letter.

I'll leave you with a couple of recent tweets that validate the feeling:

Bonus if you got down here, Unruly Futures' video:

and a discount code to use at UF: UNRULYFRIENDS

Unruly Futures 2026 - State of the Future · Luma

A future history of civilisation Last year, we gathered 250 audacious minds to explore what it means to be human in the age of exponential technology. This…

Francesco Moiraghi