A portfolio of eight ventures: three live, two in beta, three in R&D.
We build for tomorrow.
NXTGN is a UK holding company that builds next-generation ventures. We don't compete with today's tools — we make tomorrow's bottlenecks obsolete before anyone has named them. AI-native infrastructure. Operator-led companies. Fifty-year time horizons.
The portfolio
as a system.
Each NXTGN venture targets a category whose dominant tooling assumes the present is permanent. They share infrastructure, model weights and operator network — the portfolio is a flywheel, not a pile.
Most companies plan to 2030.
We plan to 2076.
Every venture in our portfolio answers a question we have already asked about the world fifty years from now. We back-cast that future to a product that ships this quarter.
C.O.R.E. enterprise GA. Lab spawns two new ventures. Operator network crosses 100.
Operator–AI pairing becomes industry default. Reactive ops formally deprecated.
Predictive operations replaces incident management as the dominant model.
Human–machine teaming codified by regulators. NXTGN protocols ratified.
Today’s "AI" is invisible infrastructure — the way "internet" stopped being a noun.
Four stages.
Run in parallel.
Our operating model runs across the portfolio simultaneously. We pattern-match futures, prototype them, capitalise the winners, and scale them with shared infrastructure.
Forecast
Pattern-match weak signal across markets, technology curves and operator interviews to surface problems that aren't yet named.
Frame
Limitless Lab pressure-tests each thesis as a working prototype within 90 days — a shippable artifact, not a pitch deck.
Found
Pair the thesis with an operator-founder, capitalise it, and spin it out as an independent NXTGN portfolio company.
Scale
Shared model access, ops support and go-to-market run through NXTGN until the venture is independent.
Most companies build for the problems on today's whiteboard.
We build for the ones
that aren't on it yet.
Forward-dated by default
Every venture is briefed against a future state — usually 30–50 years out — and back-cast to a product that ships this quarter.
AI-native, not AI-bolted
We don't sprinkle models on existing workflows. We rebuild the workflow assuming the model is the operating system.
Operator-led
Every founder we back has done the job they're now automating. Domain mastery is the moat; technology is the leverage.
Compound, don't sprawl
Ventures share infrastructure, research and model weights. The portfolio is a flywheel, not a pile.
Quarterly investor brief.
Detailed venture-by-venture progress, capital deployment, model roadmap and forward thesis. Available under NDA.