Berkshire Hathaway
Six decades of decentralised, long-horizon holding. Minimal distributions; capital allocated centrally to the highest available long-term return. Listed, yet operated with indifference to the disclosure cycle.
Thesis
The contested layer, and the ownership it requires.
Where application state, agent memory, retrieval, and orchestration converge — and why that layer calls for a structure of ownership that does not yet exist at institutional scale.
Part I — The Layer
The AI stack has one unresolved layer
Models, compute, cloud, analytics: each adjacent layer has consolidated into an oligopoly of well-capitalised incumbents. The operational data layer has not.
This layer sits between analytics — where warehouses and lakehouses operate on historical data — and applications, where agents operate on live state. It is architecturally distinct from both. Analytical systems optimise for large-scale reads over stable schemas; the operational data layer supports low-latency writes over evolving schemas, with the characteristics agentic applications require: vector retrieval, semantic search, persistent memory, and transactional consistency over heterogeneous data.
No single incumbent dominates it. Multiple architectures remain viable; hyperscalers, established database vendors, open-source communities, and adjacent platforms are all contesting the position. Which approaches prove durable — and on what horizon — are open questions.
Why the layer matters
Every AI-native application that graduates from prototype to production needs infrastructure here: durable state, retrieval over accumulated context, memory across sessions, transactional guarantees for tool use. It is the substrate on which the next generation of enterprise software runs — and at scale, a control point whose significance extends to the enterprises that depend on it, the ecosystems built on it, and questions of national technological competitiveness.
Part II — The Mismatch
The standard structures fit the layer poorly
Firms in this layer share characteristics — extended horizons, sustained capital intensity, strategic weight beyond financial scale — that neither public equity nor traditional private equity is built to hold.
Public equity
Quarterly disclosure, and the analytical apparatus that consumes it, rewards predictability over magnitude. For long-duration platform companies — whose value depends on decisions that suppress current results in exchange for architectural transformation — the disclosure regime penalises exactly the strategy the asset rewards. The historical workaround, founder voting control through dual-class shares, depends on the authority of specific individuals and does not survive changes in personnel.
Traditional private equity
The closed-end fund carries a contractual return-of-capital deadline that shapes every decision from day one, not just at exit. Choices made in year two are weighed by their contribution to saleability, not to terminal value under indefinite ownership. Architectural investments that would compound over a decade give way to improvements that support a higher exit multiple. Continuation vehicles extend the window — and then re-impose the same deadline.
These failures recur across transactions with enough regularity that they should be understood as features of the structures, not bugs.
Part III — The Answer
Permanent capital is the third option
Capital not subject to a contractual return deadline at the vehicle level. The horizon is set by the asset's characteristics — not by a fund's life or a shareholder register's median holding period.
Constellation Software
The closest existing analogue in technology: substantive management autonomy across hundreds of businesses, disciplined parent-level capital allocation, indefinite holding.
Direct-investing programs
The largest pension and sovereign funds invest directly on horizons matched to their liabilities: substantial ownership, bounded governance rights, no operational involvement.
Definition
What permanent means — and what it does not
Permanent capital, as we use the term, does not mean capital that never moves. Investors may rotate through secondary mechanisms; vehicles may redeem under defined governance procedures. What it excludes is one thing only: an obligation to return capital by a fixed date regardless of the state of the underlying investment. Removing that obligation relocates the horizon decision — from the fund's formation documents, or the marginal shareholder, to an independent board judging the asset itself.
Each reference model proves a component of the approach. What none provides is a clean institutional template for single-asset ownership of operating technology companies under institutional rather than personal control. Building that template — and applying it to this layer — is the work of the firm.