A Financial Argument for Deep Tech

Cantos Ventures
Published in
8 min readFeb 22, 2022

I’m going to make some claims about “deep tech” so let’s start off by acknowledging that term is fairly subjective. Here’s how we define it at Cantos:

Deep tech = Predominantly taking technical risk (rather than market risk)

[While there are some instances where this is true for purely software endeavors, our focus at Cantos requires some proprietary hardware or biological component.]

There is a reasonable tendency to consider these startups riskier than their market risk-centric counterparts due to the often esoteric technologies being developed. It’s the assumption I made when I started Cantos after leaving SoFi, where I had begun making small angel investments on the side. I was almost exclusively investing in world-positive software startups from 2012 through 2016 despite being a sci-fi nut obsessed with imagining a future that could be. I avoided hardware and biology startups because all the smart VCs I looked up to said things like “hardware is hard” and “we don’t take FDA risk”.

So when my friend Seth Bannon at Fifty Years first told me about a nascent synthetic biology startup called Solugen in late 2016 when he was preparing to write their first check I parroted the party line. Oops! (I ended up investing a few months later in the Seed round.)

Cantos’ first deep tech investments were made in early 2017: First Twelve (fka Opus 12), then CATALOG, Solugen, Visolis, Space Tango, and Prellis Biologics. Being perfectly honest, I thought of these as outright riskier than the other startups in Cantos 1’s otherwise SaaS- and fintech-heavy portfolio, but some of those software startups were doing well enough that I felt it made sense to sprinkle a few “moonshots” into the portfolio.

To my surprise, the deep tech startups began outperforming despite being later investments…

Sector performance within Cantos’ first fund (MOIC)

[Some may think Cantos’ hard pivot to deep tech in 2018 was driven by personal interest, but we’ve been following the numbers!]

What was going on here? Sure, venture portfolios are typically driven by one or two outliers (four in Cantos 1’s case: Solugen, Knowde, Skyryse, and Ethic), but I’ve been obsessing over this question for the past four years and here’s what I’ve come to… There are at least four concrete financial arguments for investing in deep tech, which even the most profiteering, value-oriented investor can acknowledge. In ascending order as I see them:

Four: Sand Hill Road’s Blind Spot

This was the first argument I made for Cantos’ deep tech pivot, shortly after learning Howard Marks’ famous framework for investing (and why VCs obsess over the word “contrarian”)…

Source: Howard Marks / Oaktree Capital Management

[Sheel Mohnot if you’re reading this I understand and buy your argument for why this doesn’t make as much sense in early-stage venture but stay with me.]

I was like, “Wait a second, if every investor I know is categorically steering clear of deep tech and there’s any real value to be created there then this is a huge missed opportunity!” So I dug into why most of Sand Hill Road wasn’t investing in hardware and bio. All you have to do is look at their resumes… It’s information technology if not exclusively software all the way down for pretty much every GP at every venture capital firm I’d ever heard of. InfoTech is all they’ve ever known! No wonder that’s what they’re focused on. They’re being completely rational; they’ve specialized for decades on software/IT and they’ve made a boat-load of money doing it.

As a result, deep tech is structurally “contrarian”––at least for now.

Three: Diversification

So much of “tech” today looks like a nesting doll of venture dollars. So many of the monster funding rounds over the past few years have gone into companies like APIs for SaaS startups whose customer base are other software startups. We’re in the incomprehensible derivatives phase of funding information technology.

Deep tech startups, on the other hand, if/when they get past the technical derisking, have customers that largely fall into the industrials and healthcare sectors. In a tech correction would you rather be leveraged long into Silicon Valley or diversified across innovative companies selling into pharmaceuticals, aerospace & defense, telecommunications, energy, logistics, chemicals, agriculture, and construction?

We’ll come back to deep tech-as-industrials later.

Two: Talent Arbitrage

My elevator pitch for deep tech is “We get better talent cheaper that also sticks around longer.”

Talent––or sometimes I say “talent gravity”––is the number one leading indicator for success, period. In our own portfolio we’ve observed that world-positive companies in general and deep tech startups especially have an easier time sourcing candidates, have more applicants accept offers over competing offers with higher comp, and have longer average employee tenure.

It isn’t just in our portfolio. I see it in my social circles as well. Maybe it’s who I hang out with but my friends are consistently drawn to companies with stronger social or environmental missions and frequently take pay cuts to join such companies. To bring it home, my wife is a Senior Product Designer at a large fintech company and receives––no joke––between six and ten recruiter emails every day. She loves her current employer but has taken two informational interviews over the past few years. Both were climate tech startups.

One: (Mis-)Pricing, or “Narrative Arbitrage”

Here’s the really fun one. I’m going to get wonky on you, sorry. It’s commonly accepted in finance that at its core, investing is all about finding mis-priced assets. To earn a return on capital investors take risk on some view of the future––or they find “arbitrage” in an asset that is under-priced relative to its value (often for some technical reason, as I’ve alluded in this post).

I’d argue that venture capital’s monster returns over the past twenty years are largely attributable to a historic dearth of capital vis-a-vis the largest techno-economic shift in human history — the internet. (I make this claim because the internet offered global scale with unprecedented profit margins and moats that were deeper than thought possible.) While I still hold that the amount of venture dollars pale in comparison to future opportunity offered by an expanding global population of potential entrepreneurs, the magnitude of capital that’s poured into VC over the past few years cannot be ignored. That dearth has turned into a glut in certain geographies and sectors — namely, U.S. SaaS and fintech––and here’s how those startups get valued:

Potential enterprise value = TAM * ARR * growth rate

In today’s venture capital rounds it’s mostly a game of brinksmanship WRT which investor is willing to maximize that function in their Excel model… and Tiger Global’s mostly winning that game. To be more charitable, the methodology for valuing SaaS and fintech early on is more knowable. Venture capitalists have gotten very good at pricing companies with early revenue because they know that when you’re predominantly taking market risk, growth is largely a function of capital. In these companies the equity raises largely fund CAC. “You have $X revenue today and grew Y% last year so we’ll give you $Z expecting you can grow 2Y to 3Y.” This year’s performance is highly predictive of next year’s so there are fewer dramatic surprises to the upside. (The blue line below.)

The thing about predominantly taking technical risk, though, is that if you’re careful to take virtually no market risk then growth can look more like a step function than a smooth curve. There are three companies in the Cantos portfolio that within five years achieved nine-figure contracts or purchase orders at a time when they had next to no revenue (and there are a couple more on the cusp of signing such deals). When growth is that irregular it’s very hard to bake into a valuation, especially in the early years when the revenue is basically zero. (The purple line below.)

How would you price the two companies below in year two––say, at Seed or Series A?

Here’s how we’ve seen funding play out along the purple plot…

  1. Some out-there deep tech pre-seed/seed firm like Cantos, Refactor, Boom Capital, Fifty Years, or 11.2 does the first couple rounds.
  2. A larger deep tech firm like Lux, Prime Movers Lab, Eclipse Ventures, DCVC, Obvious Ventures, or Innovation Endeavors does the Series A/B––or occasionally a generalist firm that does some deep tech like Venrock.
  3. A massive Wall Street-type asset manager does the Series B/C once the company has largely put technical risk behind it and begins to commercialize with nine-figure orders/contracts/revenue. In our portfolio Fidelity did Skyryse’s Series B; Baillie Gifford and GIC did Solugen’s Series C; and Blackrock, Baillie Gifford, and Fidelity did Astranis’ Series C.

That might seem odd considering these companies are in many cases still considered “deep tech” in Silicon Valley, but when they show up to Wall Street with nine figures of purchase orders from long-standing industrials with prime credit ratings they effectively say “What’s ‘deep tech’? We call this industrials!” (Same goes for “TechBio” companies: at this stage they’re considered healthcare/life-sci.)

[I explained this phase change recently to Nathan Benaich of Air Street Capital (an awesome ML-centric pre-seed/seed firm) and he called this “narrative arbitrage”. I’m stealing that, with full credit to Nathan!]

Now here’s the thing about this mis-pricing opportunity deep tech offers… It only works if you’re very careful to take de minimus market risk. Many of our peers get themselves into trouble by taking both. If you need years to build something that you can only then figure out if people or businesses will pay for then that’s not for us. What happens in that case is the slow/uncertain growth begins after the technical derisking phase––there’s no step function and as a result it isn’t seen as industrials or healthcare by large asset managers. You’ll likely have to raise nine-figure rounds on little-to-no revenue. Good luck to you, but that looks more like charity than investing.

It turns out this way of thinking––carefully delineating between technical and market risk––isn’t at all new. Here’s Tom Perkins on the topic (speaking about the early days of Genentech in the documentary Something Ventured, which I can’t recommend enough):

My idea in everything has always been to try to put the risk up front and get rid of the risk as fast as you possibly can, and then pour in the money once the risk is gone.

Most of what we invest in at Cantos nowadays is either making commodities cheaper and with a lower carbon footprint or treating disease. When you have line of sight to pulling something like that off, there’s less risk than you might think at first glance. It’s just a different type of risk. In deep tech you’ll have the benefit of a rare green field opportunity in venture, find an easier time hiring and retaining talent, ultimately sell into more diversified sectors with more reliable demand, and discover more “mis-priced assets”. (I haven’t even mentioned that you might save lives and maybe even the planet.)

Consider all this next time you’re looking at a SaaS or fintech pitch deck. Then wade into tech risk… the water’s warmer than it looks.



Cantos Ventures

Views best described as cosmopolitan-to-extropian. Professional fanboy of world-positive tech @ Cantos Ventures 🛰🧬💹