Hard Problems I want to back in 2025
Every January I kick off the new year with my annual Hard Problems post summarizing the areas I’m most excited about exploring and investing in over the coming twelve months. As in years past, this lists cover a wide range of pressing issues and emerging opportunity areas. I may lose interest or wise up to the realities of a given area but this list does form the current state view of where I’m investing my mental cycles. With any luck that curiosity will create a pathway for new venture investments as was the case in 2024.
I’ll will say that when comparing to year’s past, 2025 feels different. The global tech ecosystem has a vibrancy and unabashed ambition that makes the preceding years feel sleepy. Between the juggernaut of AI development, the promise of relaxed regulation, and the unifying force of a shared global adversary the feeling of limitless opportunity has never been stronger, in my experience anyway. Or maybe it’s just Fly’s new €80 million fund! Either way I look ahead to 2025 more exuberant than ever to be a First Check / Deep Tech investor.
Hard Problems I want to back
Agent (Micro)Payment Rails - One of the questions I’ve been musing most actively with my team has been, how do we upgrade the internet when agents become ubiquitous? Agents capabilities are on a tear and seemingly every corner of the knowledge economy could be remade. That has a wide range of rabbit holes, but payment rails have been where I’m most intrigued. The sheer volume of transactions - not to mention payment size, authenticity needs, latency requirements, etc. - would completely dwarf what we see as “max load” today.
Industrial Robotic Leapfrogging - I spent a lot of last year thinking about Robotics and if you’re a fan of my newsletter this will come as no surprise. The macro opportunity from my investment thesis still holds and I remain as bullish as ever. With that said, the bar for robotics investment is unquestionably higher than it was twelve months ago. A surge of capital and an army of tourist VCs have driven us to peak Robotics hype. One sub-theme that I’m thinking about is the opportunity to leapfrog. In a world of rapid reshoring the West needs to establish a new means of production, which should equate to building with best in class as a first principle. Incumbent automation brings with it rigid ROI requirements and workforce inertia. I’d rather head for open waters.
Lab Automation - Biology could be the emergent language of AGI, which means we’re going to need a lot more wet lab capacity. Despite what I said above, Lab Automation strikes me as an ideal Robotics application for several reasons. However it’s the services business model that has stunted adoption in my view. For the past year I’ve grown increasingly bullish that a full stack / services model is the play to back. IMO the opportunity size and margin profile more than make up for the somewhat non-recurring revenue.
Physics-proof Compute - Over the past 24 months compute capacity has become a top issue across every board room of merit. For the past several decades computation capabilities have been driven by harnessing three core resources: transistor density, algorithmic development and data. In the last 18 months a new limiting factor has presented itself in a big way: energy. I find this fascinating, as does my partnership. Most of our thinking has gone to heat / energy unlocks, but decentralization, edge inference and classic efficiency systems all look interesting.
Human Authentication / Deepfake Detection - I’ve written about this in the past and my conviction has only strengthened as generative capabilities have evolved. I simply refuse to believe that this a) can’t be done or b) must be handled by the tech giants. When has self-regulation ever worked?
Scientific Acceleration - This is not so much a problem but it represents one of the most prominent interest areas for Fly VC. AI has enormous potential but pushing the edge of what’s possible scientifically is the most exciting. We’ve already made several investments that harness AI to push the frontier of scientific understanding (e.g. material science, drug development) and 2025 will be no different.
Extreme Weather Prediction - Climate change may have taken a backseat in the tech zeitgeist but the risks are still very present. Global weather systems are impossibly complex and interrelated but in silico intelligence is making serious inroads. Massive financial markets like insurance, property and commodity futures are materially affected by environmental models that are woefully outdated. If the past few years of extreme weather events have shown us anything it’s that the past does not reflect the future.
Related to this, though less clear in my mind, is Climate Adaptation. If you believe that >2°C increase is inevitable (as I do) then we need to have systems to combat the structural consequences. The challenge, of course, is the negative externality realities of the area.
EU Silicon Competency - As a general rule I don’t care for geo-oriented strategies and broadly more bullish on a US-oriented GTM motion, but this is might be an exception. Compute is a national security issue and the EU is really behind the ball. There is an argument for national champions, especially in Eastern Europe that has not received their representative investment from the GCPs. I don’t know if this area viably works for a pre-seed fund but I’m thinking about it.
Non-reactor Bio-production - One of my most bruising lessons from last year was just how in the s**t VC appetite for the bio economy. There is a graveyard of startups that were predicated on the view that the ESG movement would continue and that’s clearly unwound. I’m a believer in production through a renewable biological means but cost parity must be achieved in a very short timeframe. To me that only seems possible if leveraging non-bioreactor methods, which are inherently batch-based and comparatively inefficient. My view on the approach is quite nascent and accordingly fluid but I’m super rigid on the view that sustainable impact is a feature, not the product.
As you can see with all of these, generally I have no idea what the answer is. But that’s not my job. I (and Fly VC) am here to find, back and support the extreme technical talent that can will the future forward so if that sounds like you please reach out.