From SaaS to Robots

The end of the beginning for climate tech, Part 3

In Part 1 of this three-part series, I discussed how the end of the beginning has come for “climate tech”.

In Part 2, I shared my thinking on what this means for the investment landscape, with a focus on hardware — especially “deep tech” which pushes the boundaries of science & engineering.

The short story is that the last five years produced a cambrian explosion of exciting new ideas spanning nearly every category of ‘hard’ technology that matters to the climate. The task of the next five years, and beyond, is to scale a few of these up in all of the categories that matter. This will require getting very creative in order to finance the first few big manufacturing facilities & major projects for a select handful of solutions.

It will also require becoming even more discerning when it comes to financing the development of even more new ideas — i.e. theoretically “better mousetraps”.

But there’s also another, softer side to climate tech — namely, software. Although software alone is obviously, obviously not going to solve our myriad greenhouse gas problems, I have little doubt that the intelligence and automation that software enables can be an accelerant. I see it as a solid nominee for “best supporting actor”.

Yet in the case of software, there are much bigger, tectonic shifts at play than the end of the beginning for climate tech. The “tech” sector at large is also struggling with the end of an era: the construction of the internet. Climate tech is subject to many of the same forces channeling the broader tech market in new directions, and in fact, I’ve come to believe that climate tech could be at the leading edge of some of these opportunities.

Yes, this involves “AI”… but this is definitely not about chatbots.

Phew.

The end of the beginning for software, too

Unsurprisingly, software has always been the tie that binds climate tech to the rest of the tech sector most intimately. Hence, climate tech is now facing a similar reckoning with software investment as the broader tech industry: a kind of slow-brewing existential crisis. The well-trod SaaS and mobile playbooks that have defined the market for the past fifteen years have become increasingly difficult to rely on as the “end of the internet” has approached.

This is also true, albeit to a lesser extent, for software dedicated to the big, capex-intensive industries which are responsible for the majority of carbon emissions, such as electric power, heavy industry, building materials, and transport. Entrepreneurs and venture capitalists have sought to apply their familiar playbooks to specific problems in these sectors, too.

All told, the results from this excursion have been underwhelming: Building breakout software businesses focused on niche applications in “legacy” industries — which are often constrained by risk-averse cultures, and decades of technical debt — has turned out to be just as hard as it sounds. There have been some success stories with good, albeit modest outcomes for early investors, nearly all of which have been in the emerging electric transport segment. (For example, at EIP, we were early investors in Greenlots and Viriciti, both of which developed foundational electric mobility software platforms which quickly proved attractive to strategic acquirers.)

I’m sure there will be some more pure-play software success stories in climate tech. However, much of the terrain has now been explored, and found to be rocky. Hence, the bar for better software mousetraps in the sectors that matter most for the climate may now be even higher than it is for better hardware mousetraps.

Plus, there’s only so much that pure software solutions were every going to do to address our global climate problem. Electrons & molecules are still stubbornly physical things. That’s why I find many of the most interesting solutions in climate tech tend to sit right at the intersection of bits and atoms.

Of course, nearly every hardware company these days already has lots of software embedded in its products. The IoT wave of the last decade saw to that. But I’m convinced that the stars are now aligned for the next level of integration across the software-hardware divide. In fact, I believe this is one of the most fertile areas for the next wave of climate tech entrepreneurs to pursue. Frankly, I suspect it might be a candidate for The Next Big Thing in the rest of tech, too. Move over web3. Make room for the robots.

That’s right: Robots

Did you really think the point of all this investment in AI was just ripping off artists? (ooooohhhhhhhh)

Robotics is the perfect example of a field in which progress on elements of both software and hardware are coming to together to create a new primordial soup of innovation. AI is one of these elements, of course, but there are others. Basically every input that it takes to build a robot has become a lot cheaper in the past decade.

Sources: Source: Bloomberg, “Lidar Sensor Price Plunge Leads to Orders From GM,Volkswagen”, Oct 2022.  Less Wrong, “Trends in GPU price-performance”, Marius Hobbhahn, Tamay, July 2022.  BNEF, “Lithium-ion Battery Pack Prices Rise for First Time to an Average of $151/kWh”, Dec 2022. Stanford University, “AI Index report 2022” (“Distribution of Robotic Arm Prices”). Microsoft, “2019 Manufacturing Trends Report”, 2018 (IoT Sensor cost)

At EIP, we’ve ended up with a fairly sizable cohort of robots in our portfolio, despite having no particular intention to do so. We just keep on falling for robotics companies — which is what you’d expect as the market is gearing up for a mega-trend. Already, the robots in our portfolio are mowing lawns, tending greenhouses, fulfilling orders, stringing electric conductors, and building solar farms.

For example, check out this little guy from Scythe Robotics.

Source: Scythe Robotics

The thing about robots these days is that they don’t usually look like “robots” — at least, not the humanoid kind which the majority of science fiction movies have trained us to expect. Instead, robots are mostly highly specialized machines, packed with embedded intelligence which enables them to sense & navigate complex environments. This makes robotics companies ideal examples of what the tech blogger Packy McCormack has labeled “vertical integrators” in an interesting series of posts. By his definition:

“Vertical Integrators are companies that:

1. Integrate multiple cutting-edge-but-proven technologies.

2. Develop significant in-house capabilities across their stack.

3. Modularize commoditized components while controlling overall system integration.

4. Compete directly with incumbents.

5. Offer products that are better, faster, or cheaper (often all three).”

An excellent example of a vertical integator from the EIP portfolio is Quilt, whose product doesn’t look at all like a robot… but might just qualify as one.

Quilt recently launched the most efficient two-zone mini-split heat pump on the market… and that’s just the hardware. The product is also chock full of embedded sensing and intelligence, which enables it to learn from & respond to the pattern of life in a home, the weather, and possibly even signals from an electric utility — thereby delivering thermal comfort dramatically more efficiently.

I could go on, because the list of emerging climate tech robots (and quasi-robots) is long. I’m very excited about this space.

But robotics isn’t the only domain in which the marriage of software and hardware is becoming more exciting. Another area that merits attention is the application of Generative AI techniques to foundational scientific research. Back in March of this year, a technical fellow at Microsoft named Chris Bishop asserted the following in a keynote address:

“In my view, the most important use case of AI will be scientific discovery. And the reason I believe this is that it’s our understanding of the natural world obtained through scientific discovery, together with its application in the form of technology, that has really transformed the human species.”

We’re already beginning to see examples of GenAI accelerating the kind of scientific research that underpins materials discovery. Nearly a year ago, Google DeepMind researchers published a paper on the use of AI to discover hundreds of new crystalline structures with encouraging theoretical properties, including a subset that are very relevant to climate tech — e.g. battery electrolytes and superconductors.

My colleague Eve Hanson has been keeping close tabs on this space, and just shared a more recent paper from Professor Susana Garcia (with whom she’d collaborated at Rocky Mountain Institute). The paper describes an AI-driven process for discovering new candidates for carbon capture sorbent materials, one of which has already been singled out for commercialization.

Source: “A holistic platform for accelerating sorbent based carbon capture” Nature, Garcia et al, July 17^th 2024

In sum, moving forward, I believe that the most exciting opportunities in climate tech software will be very closely tied to hardware. In fact, I suspect that this will be the case for the rest of the tech sector, as well. New modes of employing software to transform the physical world could create entirely new paradigms for company creation — which might just be the Next Big Things that tech has been searching for. And climate tech could be at the vanguard.

The end of an era

From deep tech to software, climate tech has reached the end of an era. Everyone working in this sector needs to understand the ramifications.

It’s time to stretch our legs, take stock of the building blocks we’ve already built, and refocus our attention on the next big set of opportunities:

1. FFOAK financing for the very best of the last generation of startups.

2. Better mousetraps — but only if they’re much, much better. (Plus, the discipline to pass up on marginally better options.)

3. New combinations of software & hardware which put advances in AI to work in the physical world — e.g. ROBOTS.

Oh yeah, and a carbon tax of some sort would be nice, too…

Onward!

P.S. If you’re into climate tech, and you haven’t read my series of posts on the Ten Biggest Questions in Climate and Energy Tech… you should!

And subscribe, please.

(Part 1)

(Part 2)