A packed year starts here, and it starts, as it usually does, with a deliberate slow-down. January is a hard month in the northern hemisphere, so I spend part of it with our team in Cape Town before turning north into the year. This quarter we are leaning hard into face-to-face work, with Simplifying AI for Family Offices workshops in Stockholm, London and Copenhagen, and our first larger family office gathering in Copenhagen.
We also launched Simple Select this month, built to cut the work and cost of choosing technology by up to 80%. If picking the right tool has ever eaten a quarter of your year, that one is for you.
From Cape Town to the Alps
Davos last week put operators, investors and policymakers in the same overheated rooms to talk AI. The striking thing was how far the conversation had moved. The models and the demos are yesterday's argument. What people wanted to talk about now was ownership, integration, and physical limits. A few things stuck.
Live data is the dividing line
The edge is going to systems that run on live, messy, real-world inputs, video, speech, sensors, while staying compliant. That quietly favours whoever already owns the distribution and the constant exhaust of data that comes with it. Everyone else is renting.
Ownership beats the model
The line of the week. Who owns the rails now matters more than who has the best model. Content, identity, devices, cloud, payments, the default surfaces people open out of habit, all of it compounds, while a third-party dependency gets commoditised the moment someone cheaper shows up. It is why so many people in the room saw Google as structurally advantaged, with YouTube, Search, Android and Workspace forming a closed loop that runs from data to product and back again.
The bottleneck is physical
Here is the reality check nobody at a tech conference enjoys hearing. Progress is now running into energy, grids, data-centre capacity, permits and skilled labour. Almost nobody argued AI is overhyped. The argument was that it is a huge, uneven, deeply physical build-out, with rising capex and hard limits you cannot prompt your way around.
The ROI reckoning
We are in the show-me phase now. The pressure to prove results is real, and plenty of pilots still fail to scale, tripped up by integration, workflow redesign and change management. The interesting money has stopped asking whether AI works and started asking whether it works here, in this office, on this process.
How families are playing it
Pragmatically, and I mean that as a compliment. Less about winning some abstract race, more about defensible exposure: applied and industrial AI, energy, compute, data rights, often outside the US. Barbell positions are forming, light on models, heavier on infrastructure. The warning that came up again and again was AI everywhere with nothing defensible underneath it. Lose control of distribution, data rights or infrastructure and the value drains out quietly. The AGI debate, for all its noise, felt like a sideshow. These systems are already reshaping labour, energy and trust, which is quite enough to be getting on with.
In December, I gave a strategic take on exactly this at the IMD Family Office Conference with FBN in Lausanne. The video is up for anyone who missed it.
Upcoming events
Family Office Days at Techarena, 10 February, Stockholm: invitation-only, closed-door discussions and Nordic investment insight. The Simple AI workshop follows on 11 February at Techarena, hands-on and applied. And our Family Office Gathering on AI lands on 18 March in Copenhagen, a half-day with leading family offices and speakers.