What do SpaceX, Nvidia, and ASML have in common? They all boast robust technology moats, giving them a formidable edge over their competitors. However, technology moats are not the impenetrable fortresses they sometimes appear to be. In fact, they could be the Achilles’ heel that leads to a startup’s downfall. How? Let’s take a look.

For a deep tech startup, a technology moat is more than just a competitive edge. In their early days, it’s pretty much the only thing the founders have. Technology moats are based on unique or privileged access to a technology breakthrough or advantage, patentable or not, that can create business value by itself (e.g., by solving an urgent, unaddressed market need). However, this advantage does not typically last forever: competitors catch up, patents expire, other ways of achieving the same result are discovered. Consider carbon capture: Carbon Engineering and Climeworks were pretty much alone in the world for several years; carbon capture startups are now commonplace. Sometimes, knowing that something can be done is what spurs more innovation.

Tesla and ASML can help us take a look at the extremes. Tesla’s approach to open-sourcing its patents, trading traditional patent protection for ecosystem growth, was revolutionary at the time. However, one could argue that by the time Tesla open-sourced its patents, the leadership team knew their future market leadership was well established and no longer exclusively based on technology. That is, Tesla knew that its technology moat would die anyway, so it used its downfall to its advantage. The conditions they imposed on the open-sourcing process reinforce this conclusion, as leveraging Tesla IP also prevented users from trying to enforce their patents.

On the other hand, ASML’s dominance in the EUV lithography market remains unchallenged and is likely to stay that way unless and until a new technology paradigm arrives (e.g., free-electron laser lithography). The paradigm change is particularly important for this case. The current technology is so complex and the barriers to entry so high (billions invested, companies acquired, processes developed) that ASML is essentially guaranteed a monopoly in the current paradigm¹. However, ASML’s efforts to stay at the cutting edge and invest aggressively in technology development reinforce the same conclusion: had they stopped investing and taking risks, someone else could have decided that the potential reward was worth the hassle².

Thus, I typically think of technology moats as temporary. They provide breathing room for startups to scale up both their technology innovations and their businesses so that by the time the technology moat gets eroded, they have other sources of defensibility (e.g., sticky customer contracts with high switching costs, control of some valuable market niche, significant economies of scale). As with anything in the cutting edge, longevity isn’t guaranteed. Companies must walk the tightrope between maintaining a technological edge and evolving beyond it.

Because of the above, I am generally skeptical of the ‘proven technology’ sales point during deep tech startup pitches. If the technology is ‘proven,’ what prevents deeper-pocketed rivals from erasing you from the map once the concept is proven?³ Some industries have sleepier giants that may prefer to become your customers or partners, but in others, incumbents are as good and as fast at R&D as many startups are (read: Nvidia, CATL).

In the end, there is no one size fits all. Some industries lend themselves more to long-lasting technological moats, and others don’t. Whatever may be the case in your industry, remember: technologies, by themselves, tend not to be very valuable. Businesses are.

Some Good Reads

David Keith on why carbon removal won’t save big oil but may help the climate – The Economist

The controversy over oil & gas players investing on carbon capture got more intense with the recent acquisition of Carbon Engineering by Occidental Petroleum. I’m biased – we were investors in Carbon Engineering (yay!) and I believe that more money into climate is better than less money into climate, even if the incentives are not perfectly aligned with our respective ideologies. Whatever your thoughts, this piece from one of Carbon Engineering’s co-founders is pretty clear on how far the ‘greenwashing’ here could go: permanent removal is likely to cost over $150/ton of CO2 for the foreseeable future, adding ~$70/barrel to the price of oil. Brent oil, by the way, is at ~$80/barrel today.

Profits would drop nearly 50% if companies paid for emissions – Fortune

Following the above thread, economists estimated in August that paying $190 per ton of CO2 emitted could wipe out almost half of total corporate profits. Even a generous long-term target of $100/ton would still erase over 20% of total corporate profits. Voluntary carbon markets are a tricky beast, indeed…

How Jensen Huang’s Nvidia Is Powering the A.I. Revolution – The New Yorker

One of the best interviews I’ve read this year. Don’t miss this one. Plus, tech moats!

First plasma fired up at world’s largest fusion reactor – Science

The JT-60SA tokamak fusion reactor in Japan just started operations. It will take a couple years for the machine to get where it needs to be, but it’s currently the largest fusion reactor in the world, and its data will be immensely valuable for the entire fusion community. Worth a quick read if you are interested in fusion energy.

Plant-by-plant decarbonization strategies for the global steel industry – Nature

Researchers compiled a database of iron and steel facilities worldwide, assessed plant-specific characteristics, and analyzed potential decarbonization strategies. Interestingly, in most cases, transitioning primary steel plants (i.e., those that make steel directly from ore) to secondary plants (i.e., those that make new steel from steel scrap) is the most cost-effective way to decarbonize these plants. This is influenced by the fact that primary production is limited in how much better it can get via e.g., efficiency upgrades and how much more emissions primary steelmaking generates (~2 tons of CO2 per ton of steel for primary vs. <0.5 tCO2/tSteel for secondary). On the other hand, the viability of this switch depends on the availability of scrap. The lesson: we still need to figure out a better way to make primary steel.

An oscillating reaction to produce clean fuels – Science

The Fischer-Tropsch process converts syngas (a mixture of carbon monoxide [CO] and hydrogen [H2]) into liquid (read: more complex) hydrocarbons. It is an alternative pathway to hydrocarbon fuels and chemicals from feedstocks other than oil, such as natural gas, coal, and biomass. While its commercial relevance today is limited, it has gained relevance more recently as one of the most promising approaches to future clean fuels. The recent paper discovered an interesting oscillating reaction that takes place in the process, highlighting the fact that despite having been in use since the late 1920s, we don’t fully understand the specifics. This easier-to-digest perspective provides more context into the process, and would recommend the read to anyone with an interest in clean fuels. The in-depth paper is this one.

Thanks for reading! See you next time.

If this resonates and you are building a breakthrough science company – get in touch here! Can’t wait to learn more.

For suggestions, feedbacks, and reading recommendations – please reach out here.

To ensure you never miss one of these, subscribe above and follow me on X or Linkedin.

Footnotes

¹ The best piece on ASML I’ve ever read is this one from The Generalist. Highly recommended. Go back.

² The next paradigm, however, may be different enough from the current one to make the barriers to entry orders of magnitude lower. If so, challengers are bound to emerge: there are already a few startups working on the free-electron laser alternative. Whether they’ll succeed is a different story. Go back.

³ Apple does this almost annually. The recent unveiling of OpenAI’s GPTs is another good example. Network effects and switching costs are among the few things that can provide protection here. Go back.