My 2025 Investing Letter

How I'm thinking about the world

Happy 2026. Historically I used to write annual investor letters to the shareholders of the companies which I was CEO of. Annual reflections might seem slightly quaint in 2026 but I feel news cycles create so much over-sampling noise that longer observation periods remain a useful exercise. As I’ve moved from ‘builder who also invests’ to more of an ‘investor who also builds’, I’m continuing this format as an update on what I’m doing and some views on how I see the world. Feedback is deeply appreciated.

For context to all of this my investing activities are currently in three broad vectors. I’ll unpack those a bit more later but briefly they are:

• LFG Holdings (with Max): investing in opportunities across Discord and User-Generated Content (UGC) Gaming (e.g. Roblox, Fortnite).

• 10xHumans (under wraps but active with Max and Daria): investing in AI Enablement professional services (training/consulting/implementation).

• Personal/family office: PE, VC, private credit, special situations. Occasionally some direct company investments, usually very early, always founder driven.

I also sit on a couple of boards (MTG, a publicly listed mobile games publisher and Levellr, the Discord social listening platform) and advise a few others. So my observations come from a reasonably broad spread of data and behavior.

Broadly speaking, there are three themes I’m obsessing about.

The uninvestability of new digital spaces

One of the reasons that Max and I set up LFG Holdings in 2024 was in response to seeing new growth markets (especially new digital spaces) that were not conventionally shaped for either VC or PE. There are a variety of contributing factors to this phenomenon

Capital efficiency and business miniaturisation

Thanks to device saturation (distribution), attention saturation (content niches) and expanded monetization (particularly to younger audiences), the atomic unit of business has become smaller than it ever has been. Particularly true in digital media spaces, this has become more broadly true elsewhere and compounded by a general bootstrapping/seedstrapping mentality which rejects a lot of VC principles. AI moves this business miniaturization trend forward by an order of magnitude.

Business models

A bias towards bootstrapped building tends to go hand in hand with business models which many investors (in particular VC) are not comfortable with e.g. professional services, advertising, tools etc.

Demographics

A lot of new spaces we look at combine all the above factors with a generational gap (this isn’t just age, it’s also about definition of quality) that make it hard for investors to relate to the founders or vice versa.

Roblox (and to a lesser extent, Fortnite’s UEFN platform) is a good example of this. In the global gaming landscape (which has slowed down to a crawl), it is by far the fastest growing consumer platform. In fact, two of the biggest games in 2025 (Grow a Garden and Steal a Brainrot) were both built in Roblox. But it’s extremely difficult for funds to invest here: most game developers are under 25, entirely bootstrapped and aren’t building for an exit. The successful investors in this space have not been the VC funds but instead the (mostly) native developers who have built hybrid publisher/investor models focused on owning minority positions in a large portfolio of titles. Obviously you can buy the public stock for an aggregate position but that’s not giving you individual game exposure: If you had bought RBLX and a position in Steal a Brainrot six months ago your platform position would be down 25% but your game position would be up 300-400%.

It turns out that it is possible to know the future and yet be unable to find an easy way to invest in it. Business historians might argue that this was a similar landscape to that which Art Rock encountered in the 1970s. Perhaps. All we can say for certain is that, in several growth areas, there is a considerable amount of value out of reach for the conventionally-shaped vehicles for capital. As a result VC and PE are sitting on the sidelines waiting for scale players to emerge.

The importance of professional services in AI success

We haven’t spoken publicly about our 10xHumans strategy yet (I’d expect to be a bit more public about this in Q1) so this is something of a preview.

For the second half of last year, Max and I made ourselves quite unpopular in certain circles by asking a lot of questions (both to companies and AI sales teams) about just how easy or hard it was for business to meaningfully adopt AI tools. It has informed my opinion that a) AI will be long-term transformative but also b) that it will require a huge effort in training, consulting and implementation to help companies get there. Dwarkesh Patel made the point more elegantly in his December essay:

Models keep getting more impressive at the rate the short timelines people predict, but more useful at the rate the long timelines people predict.

AI maximalists tend to dismiss the role of human-powered enablement and argue that agents will simply make this aspect of professional services (literally) redundant. And yes, it may sound like a strange thing to suggest our AI strategy is to invest in everything except the technology. At this point we’ve built what we believe is the biggest market map of bootstrapped AI professional services firms across Europe and the US which has included at least a hundred founder conversations (more every day). The evidence on the ground is consistent: enterprise adoption of AI requires leadership buy-in, planning, change management, learning, implementation and that’s assuming their assumptions on underlying models remain consistent. Think of a time when you switched from Windows to Mac operating system (or maybe the opposite). Now imagine a company doing the equivalent of that and you’re beginning to grasp the challenge.

One of the sentiments we’ve heard quite frequently in our conversations is that the incumbent management consulting firms (Accenture, Bain etc.) are simply not sufficiently AI-native to handle this challenge. Accenture’s ~$1b acquisition of Faculty (25x revenues!) last week suggests that’s a real issue. But it isn’t just the Fortune 500 level of companies who will need this support, every company is going to need some equivalent of a fractional Chief AI Officer or some kind of AI consultancy of record.

This led us to a high conviction thesis around AI Enablement professional services, and specifically around the newer, AI-native firms. This is very much our kind of unconventional landscape, which is challenging for VC (bootstrapped founders, services business models) and PE (mostly small players) yet driven by extremely strong tailwinds.

Unconventional weapons for unconventional wars

There have always been investment opportunities in places where most people weren’t prepared to do the work. One way of describing what we invest in is PE types of risk/return operating in VC types of spaces i.e. too small for PE and too sensible for VC. This sounds like we’re driven simply by the opportunity space but it really comes down to the nature of founders we like working with. Ultimately it’s the intersection of the profitable and the interesting.

In 2025 it was surprising that we saw so many of these opportunities in spaces which have such strong fundamentals. A lot of that is AI-driven. Ultimately AI is lowering the barrier for the creation of sustainable entrepreneurship which is incredibly exciting. The counter-factual on what companies were not started because someone just couldn’t afford a developer is intriguing.

I am not here to declare the death of VC (or PE) or anything like it. But I am quite certain that we’re entering into a period where there are increasing opportunities for non-linear investors. This fracking-style model will continue to evolve. We set up 10xHumans to focus on the AI Enablement thesis and it builds on our learnings from LFG Holdings. Firstly we took some dedicated capital into the main vehicle to add to our capability. Secondly we have started to create a more platform approach to building (with Daria joining). I’ll write more about this (publicly) in a few weeks.

You can think of our approach to unconventional investing showing up in three ways for 2026.

Firstly, when we find a theme we like (rare), we look at it from an ecosystem perspective. Admittedly ‘ecosystem’ is now on par with ‘fly-wheel’ in terms of general abuse in corporate terminology but it’s still a useful model to think about the inter-connectivity of things. Within an ecosystem, we try to identify the most valuable parts and own as much of it as we can. By definition this will often mean we own interoperating parts of the same ecosystem. This is a feature, not a bug and it provides us with superb levels of data to base other investments on. For example, LFG Holdings is focused on behaviors emerging from Gen Alpha/Gen Z (theme) and we have filtered this to three ecosystems (Discord, Roblox and Fortnite). Our Discord ecosystem investments include Levellr (enterprise Discord community tools for major companies), Wildfire (Discord community activation for brands) and Discord itself (the platform). Triangulating signals from all three gives us a very well formed opinion on further investment opportunities we see in the Discord ecosystem.

Secondly, our vehicles for investing are deliberately not funds but instead holding companies. Even though it sounds very on trend for former founders to have a ‘holding company’, I think that one of the best edges to have in these new growth spaces is flexibility on how you capture value. Our north star is founder quality and that comes in a lot of different investment permutations.

Thirdly, I want to talk about creativity, which I really feel is underrated in business. Part of our edge comes from authenticity (read: scar tissue) as founders and experienced builders, the other part comes from the creative thinking which leverages that. Being unconstrained in deal structure and also growth strategy opens up lots of interesting ways to think about both investing models and scaling.

Thinking about 2026

There are lots of other people’s predictions which you can read so I won’t add to that bucket of noise here. A few themes I continue to ponder as 2026 starts to move:

• The unbundling of wealth management: I’ve written about this before but it feels like wealth management is bar-belled as either major bank or wealth management service (start-up or otherwise) and I would love a way to be able to think about this in a much more componentised basis.

• The upside in European economies that can be unlocked by AI: Although Europe’s economic situation remains pretty terrible relative to a lot of other areas, I increasingly wonder if this offers a huge upside in what can be unlocked by AI over time.

• The concept of special purpose collaboration vehicles: a concept I keep coming back to over time is the need for scale to unlock valuation step-change in markets where you have a collection of sub-scale VC-backed players. In the context of a lot of investments stuck in VC portfolios, I wonder if we’re going to see some industry-level M&A paradigm like this emerge.

It’s going to be an intriguing year.