Where AI Infrastructure Meets Thermodynamic Reality.
The AI buildout is bounded by thermodynamics, power distribution, and physical infrastructure. Global compute expansion is no longer constrained by software or capital, but by multi-year grid congestion and extreme thermal loads. Value has concentrated in the mission-critical manufacturing and technical service layers that solve these bottlenecks — businesses the buildout cannot proceed without, where demand is outrunning installed capacity.
In the mid-market, these constraints are not yet fully priced. Assessing these targets means looking past headline backlog growth to what decides enterprise value: order-book substance, lead-time moats, component availability, raw material volatility, customer unit economics, and long-term margin scalability.
Our practice pairs ex-MBB strategic rigour with ground-truth domain authority — industrial engineering executives, power systems experts, and supply chain operators who have run manufacturing operations in these sectors. We work with private capital sponsors buy-side and sell-side, portfolio management teams, and corporate leadership. Every engagement rests on primary research across supply chains and customer networks, capacity modelling, and field-validated unit economics rather than desktop synthesis or fragmented expert-network calls.
Transformers, switchgear, substations, and the power equipment enabling high-density delivery. Interconnection queues and utility equipment lead times have created multi-year backlogs. We evaluate whether order books represent non-cancellable, high-margin commitments or speculative double-ordering by developers hedging against delivery delays, pressure-testing pricing power against raw material volatility and manufacturing capacity limits.
Direct-to-chip liquid cooling, heat exchangers, cooling loops, and the specialised water processing required for high-density compute clusters. Thermal density is forcing a rapid transition away from traditional air cooling. We analyse proprietary IP differentiation, long-term reliability track records, and the scalability of the field-service and maintenance networks required to keep these systems operational.
Smart grid automation, power distribution monitoring, and process controls embedded in physical assets. Value sits with specialised equipment and software that optimise energy efficiency, reduce downtime, and monitor thermal risk. We evaluate customer switching friction, integration depth, and the sustainability of gross margins against low-cost hardware commoditisation.
Commoditised components, volume automotive manufacturing, consumer electronics, and medtech sit outside our scope by design.