


Every dollar of AI demand passes through four layers — the compute that gets built, operated, monetised, and owned before it lists. We hold conviction across all four.
The compute supply — GPUs, accelerators and high-bandwidth memory. A handful of designers and foundries (NVIDIA, TSMC and the HBM makers) form the structural bottleneck the entire industry depends on, at ~75% gross margins.
Explore the AI strategy
The compute operators — hyperscalers funding the >$500B 2026 buildout and turning raw silicon into contracted, rented inference revenue. We own the operators converting today's capital expenditure into durable cloud income.
Explore the AI strategy
The compute monetisers — foundation models, copilots and vertical software that convert raw compute into recurring enterprise revenue. The youngest and fastest-moving layer, which we size for asymmetry and manage actively.
Explore the AI strategy
The compute owners — before they list. Direct pre-IPO and secondary-market stakes in the frontier private AI labs, where the largest value is still created off the public market — Anthropic near $965B, OpenAI filing to list above $850B — late-stage compounding we access before the exchange does.
Explore the AI strategy






