Deep Tech 2020: What Really Has Our Attention (and Interest) This Year?
Looking at tech trends is always fun. Particularly over longer periods of time, as we can then begin to see the pendulum swings. Let’s take computing as an example.
Consolidating the on-prem legacy chaos into a large cloud platform where basically anything can be managed for you economically has evolved into those same IT assets being distributed over hyper-scalers, virtual/private, hybrid, and edge clouds. Thousands — more like millions — of edge clouds. Now, we’re seeing companies move key workloads back to on-prem, as it’s simply more economical. Even our own portfolio companies spend their first VC check on buying a ton of hardware, something I did 20+ years ago (and a clear startup narrative violation).
Pendulum swings are unstoppable and unavoidable. That’s why at Speedinvest, our Deep Tech team looks at the ever-changing tech world roughly through 4 lenses:
1) Enterprise AI (Enterprise software)
- AI/ML applied to software for common, horizontal enterprise processes
- Vertical AI
2) Infrastructure: Computing, Data and Networking
- Things that give DevNetSecOps superpowers
- Whatever assists data scientists and first-world citizens and helps them be more effective operationally
- Tools that help developers and data scientists work better together
- Including AI applied to cybersecurity, IoT/Edge security, and securing AI
4) Frontier Tech
- For example, quantum technologies, DLT, lensless imaging, securing AI
Investment strategy: The who, what, when and why
We invest at seed stage and write initial tickets of €1M — 1.5M. The companies we invest in often have rich IP that is either hard to reproduce or well protected, which provides a valuable competitive advantage and/or moat. These ideas can take longer to take to market and may require more capital to implement, which is why we focus on technologies that will have a large, long-term impact. The teams typically have strong academic credentials or a rich operational background specific to the area they are working in.
What really has our attention in 2020?
We will continue to use the above criteria when investing in 2020, but there are several topics and trends that have caught our eye this year. Specifically, areas where we need to better understand how recent technology breakthroughs enable new use cases or allow for improvements to existing use cases. Where are the actual scalable solutions starting to emerge, and what are the market opportunities they present? Here goes.
5G: McKinsey estimates that network related CAPEX will grow 60% in the next 5 years to roll out 5G. It’s easy to dismiss 5G as just a “fatter pipe,” but more interesting to look for new use cases and “over the top” services that will emerge or be enabled when these networks come online at scale.
New AI breaking through: New AI methods, technology — and challenges, frankly — continue to occur at high velocity. We’re not only interested in looking at Vertical AI applications but also further upstream so we can more clearly see what’s coming down the pike. Area’s like GANs, transfer learning, federated learning, explainability, and security. Where does new research emerge that creates “magic”?
Democratization: Regardless of what you call them, we believe there’s a continued demand from non-expert enterprise users to get access to state-of-the-art tools that are currently in the exclusive domain of developers and data scientists, with an amazing UX.
Automation: In enterprise software automation, we believe automation will continue to improve quality, efficiency and UX — either in the form of co-pilot or auto-pilot solutions. For DevNetSecOps, these solutions will be the gatekeepers and help swing the pendulum back. Maybe not towards something simpler, but at least something more manageable.
Edge Computing: Compared to the Apollo 11 moon lander, an iPhone has millions of times more memory and 100K times more computing power. Technically speaking, your SIM card could run quantum-resistant crypto today. While IoT growth has turned out slower than forecasted, it is hard to imagine a world where everything is NOT connected, and huge workloads NOT move to powerful edge devices and (some say millions of) microdata centers running very capable software.
Bioinformatics: Just as movie creation will be 100% CGI virtualized (The Al Pacino on Netflix today may not look much like the real thing, but we can assume he owns the rights to his avatar), software is now munching on biology and life science. For example, there is a very exciting opportunity to use software/AI and data to engineer new drugs creating an estimated $8B market in the total $80B drug discovery market in 2026.