Investment Thesis
Website presentation prepared 21 October 2026. This page explains the existing investment framework in website form; it is not a new strategy announcement. Read the original thesis published 10 October 2026.
The Economics of Chokepoints
Owning the businesses that can turn structural demand into lasting earning power
I invest in the physical and digital bottlenecks behind the modern economy: businesses whose technology, scale or place in their customers’ operations makes them difficult to replace.
My central conviction is that the expansion of artificial intelligence and digital infrastructure can create substantial value for the companies supplying its hardest-to-replicate capabilities. I prefer to own those capabilities rather than make the portfolio depend on predicting which new application wins.
That is what I mean by the Economics of Chokepoints. The opportunity is not simply to find a growing industry. It is to identify where that growth can translate into durable profits, and buy those profits at a price that leaves room for a worthwhile return.
Follow the spending, then follow the profits
Capital expenditure is a useful starting point. Spending by cloud platforms and other customers becomes revenue for suppliers of equipment, chips, memory, networks and infrastructure. But revenue is only the beginning of the investment case.
A supplier may win orders while accepting lower margins. It may need to spend heavily on new capacity, or face a competitor by the time that capacity arrives. A platform may attract customers but struggle to earn an adequate return on its infrastructure.
I therefore ask two separate questions: Who receives the spending? And who can retain an attractive share of it after operating costs and reinvestment?
The “toll booth” analogy describes bargaining power, not a guaranteed payment. In chipmaking, precision and production complexity impose real constraints; in data centres, electricity supply remains a physical requirement. My investment question is whether a particular company can turn such constraints into a lasting commercial advantage. Chip manufacturing · Data-centre infrastructure
Four connected areas of conviction
These areas share a search for hard-to-replace capabilities, not identical business models or sources of demand.
1. Silicon and tool builders
The first area covers semiconductor design, manufacturing and the equipment used to produce chips. ASML, TSMC, NVIDIA and AMD illustrate different positions in that chain, not interchangeable businesses. My interest is in the accumulated expertise, technical performance and customer dependence that can make a capability difficult to reproduce. How the chipmaking chain works
The test is not whether a company participates in AI. It is whether its particular role supports attractive economics as technology and customer requirements change.
2. Memory and high-speed interconnects
The second area focuses on moving and supplying data, alongside processing it. I look for businesses that can address constraints in memory and connectivity, then retain value through differentiated products and disciplined capacity investment.
Scarcity alone is not enough. If competitors add supply, customers change designs or pricing weakens, a strong industry outlook may still produce a disappointing investment.
3. Hyperscale and orbital infrastructure
The third area covers large-scale computing platforms and orbital infrastructure. Here, I look for an advantage built through scale, engineering capability, customer relationships and the cost of reproducing a network of assets.
These businesses are not just beneficiaries of infrastructure spending; they may also fund it. The investment case must therefore explain how their own capital commitments become customer demand, cash generation and returns on invested capital. Size is a starting point for analysis, not proof of an unbeatable position.
4. Enterprise data, security and biological platforms
The fourth area extends the thesis to enterprise data, cybersecurity and platform-based biotechnology. The common question is whether a capability becomes deeply embedded in how customers operate, or provides a repeatable foundation for developing products.
These are different businesses with different tests. In software, I examine customer adoption, retention and the cost of switching. In biotechnology, a platform does not remove the need for clinical evidence, regulatory approval and reliable manufacturing. Its promise must be tested product by product. Drug development and approval
A strong business is not attractive at every price
Across these areas, I look for a defensible competitive position, evidence of pricing power, sound finances and a credible path from revenue to cash available to shareholders.
Valuation determines how much of that success I am paying for in advance. I consider what the price assumes about growth, margins and future investment, and whether the case still works if results are less favourable. An excellent company can be a poor investment when expectations leave too little room for disappointment.
Concentration is deliberate: I want meaningful exposure to the businesses in which I have the strongest conviction. It also means that several holdings can be affected by the same customers, spending cycle or change in market expectations.
What would change my mind?
I would reassess an investment if customers found credible substitutes, pricing power weakened, capacity outgrew demand, or growth repeatedly failed to produce the expected cash returns. Funding pressure, regulation, geopolitical restrictions and execution failures can also change the economics.
A falling share price alone does not disprove the thesis. Equally, a rising price does not prove it. The question is whether the business evidence still supports the original case at the valuation available.
My aim is to own hard-to-replace capabilities, not to assume that any company, technology or investment case is beyond challenge.
Capital at risk. Past performance does not guarantee future results.
How I approach research and risk → · Dated portfolio snapshot →