Compute is capital.
The world's software infrastructure is beginning to rebuild itself. Autonomous coding agents inspect, rewrite, test, and improve the systems on which modern economies depend. We see the beginning of a global renewal of software capital, an enduring source of demand for compute, and a foundation for intelligent production.
The next expansion takes intelligence into machines. Emberstoke builds for an economy in which computation develops, improves, and operates the systems that produce the world's goods and services.
01 / Software renewal 02 / Economic productivity 03 / Intelligent production
Fifty years of software are becoming available for reconstruction.
Operating systems, databases, banking platforms, industrial controls, enterprise applications, and embedded software embody decades of accumulated engineering work. They also carry decades of compromises: obsolete dependencies, fragile integrations, incomplete tests, security weaknesses, and expensive maintenance. Autonomous coding agents change the cost of improving this inheritance. Projects held back by the scarcity of engineering labour become available for reconstruction.
The return reaches through the economy. More dependable and efficient software improves the financial, industrial, medical, logistical, and public systems that already produce value.
Software renewal becomes an industrial compute workload.
A coding agent works through repeated cycles of understanding a codebase, planning a change, writing and executing code, running tests, inspecting failures, and improving the result. Each cycle consumes compute. Apply that process to the world's accumulated software, then keep those systems under continuous improvement: the workload extends across industries and persists across generations of hardware.
The opportunity is the renewal of the productive capital on which the economy runs.
From software renewal to machine production.
- 01
Rebuild software capital.
Agents modernize, test, secure, and optimize the software on which the economy runs.
- 02
Raise the productivity of existing systems.
Better software improves the efficiency and reliability of financial, industrial, commercial, scientific, and public infrastructure.
- 03
Extend intelligence into machines.
Computation increasingly enables robotics, autonomous equipment, and physical production systems.
All three stages depend on compute infrastructure, energy, and the ability to deliver useful work reliably.
The value of compute reaches across the economy.
AI application revenue captures one part of the transformation. The larger value lies in what computation enables: better software, more efficient infrastructure, and increasingly capable machines. Our investment thesis follows that value through the capital stack, from processors and facilities to energy, grid access, and control of future useful work. The central question is where expanding productive capability becomes durable economic value.
Productive capacity is also economic power.
As computation becomes foundational to production, ownership becomes a strategic question. Control over compute, electricity, infrastructure, and future capacity shapes who can produce and who receives the returns. Sovereign compute, public and cooperative holdings, and transferable rights to future capacity offer ways to distribute access to intelligent production.
Oculan: computation as creative production.
Oculan turns written prompts into fully editable vector graphics. Stickers, contours, and wordmarks are delivered as shapes that can be edited in compatible document applications. It puts computation to work in creative production.
Compute is becoming the capital that improves other capital.
Software renewal is the first great wave. Intelligent physical production is the next. Our thesis follows the connection between them.