Compute
The chips, memory, networking, systems, capital, and energized capacity that set the ceiling on intelligence production.
exmxc is the research institution building information assets on the AI economy. We follow Compute, Interface, Alignment and Energy through two assets: Strategic Signal, the record of deals, and AI Lattice, the map of who depends on whom. The AI Power Index is our view.
A connected framework for understanding capability ceilings, power concentration, and long-horizon capital flows in the AI economy.
The chips, memory, networking, systems, capital, and energized capacity that set the ceiling on intelligence production.
The distribution surfaces, agents, workflows, and physical actuation layers through which intelligence reaches users and institutions.
The control capacity that determines what increasingly capable systems may know, do, authorize, disclose, and recover from.
The firm generation, grid access, fuel, permitting, and power infrastructure that determine whether AI capacity can actually scale.
The daily record of disclosed deals across the AI economy.
Explore the record ↗ The mapThe map of who depends on whom: 92 entities, 169 sourced dependencies, and 12 chokepoints.
v0.5.7 ·
Explore the map ↗ The viewexmxc’s assessment of where power sits across the Four Forces and where it is headed.
Explore the Index →Every thesis keeps its original falsifier and status. Open work stays open; rejected ideas are not counted as wins; failed theses stay visible.
| Thesis | What would prove it wrong | What happened | Status |
|---|---|---|---|
| AI Infrastructure Scarcity | Scarcity fails to persist or earnings revisions fail to support the valuation thesis. | The published 88-day recheck found all five highest-conviction sub-0.60 sPEG names delivered double-digit returns; the network layer, flagged as scarce but priced for perfection, declined. | Supported · May 2026 |
| System of Record Scarcity / PTC | Embedded product and workflow authority fails to translate into strategic value or buyer willingness to pay. | PTC subsequently entered a strategic transaction at $205 per share. The transaction is evidence for the scarcity mechanism, not proof that every system of record is investable. | Supported · 2026 |
| Nuclear Uprating / Vistra | Federal financing fails, uprate economics deteriorate, or incremental firm-power demand does not support the investment. | The market has repriced the signal, but price action is not treated as validation. Financing terms, delivered megawatts, and cash-flow capture remain the test. | Open |
| Convergence Monitor | The six exit signals prove too noisy, correlated, or late to improve de-risking decisions. | The monitor is live and time-stamped weekly. No final verdict is claimed before a full exit-side cycle is observed. | Live test |
| Workday system-of-record candidate | Backlog deceleration, stock-based compensation, or AI-layer risk remains too large relative to valuation. | The candidate did not clear the decision rule. The rejection stays visible rather than being erased from the hunt board. | Rejected |
Miss discipline: a published thesis that hits its predeclared falsifier will remain here as Falsified. exmxc will not manufacture a “miss” for optics or relabel an unresolved thesis as a win.
The Four Forces generate the questions. Evidence creates conviction. Capital tests the conviction. Outcomes sharpen the next research cycle.
Observe → Research → Conviction → Deploy → Learn.
Scan Compute, Interface, Alignment, and Energy for structural shifts, bottlenecks, and emerging scarcity.
Build original datasets, frameworks, calculations, decision rules, and falsifiable theses around the strongest signals.
Commit capital only when structural conviction, valuation, and downside discipline align.
Track what the market, counterparties, and operating data confirm or refute, then feed that evidence back into the model.
Capital scope: exmxc deploys capital on its research. Research only. This is not investment advice.
Primary research is organized around the Four Forces. Capital deployment follows conviction, valuation discipline, and explicit falsifiers.
Frameworks and Signal Briefs interpret structural shifts across Compute, Interface, Alignment, and Energy.
Explore the research ↗ Capital deploymentResearch-backed capital notes translate structural scarcity, valuation, and decision rules into real allocation discipline.
Explore exmxc Capital ↗ Longitudinal observatoryEntity Clarity reads who lets AI in (Alignment) and who AI answers surface (Interface). AI-search retrieval remains a longitudinal project.
Explore the longitudinal work ↗sPEG and Power Lens support capital research. Entity Clarity and AI Retrieval track these signals over time.
A governed research index of companies with durable scarcity, explicit admission gates, evidence confidence, exclusions, and a separate valuation discipline.
Explore sPEG →The index reads two forces: who lets AI in (Alignment) and who AI answers surface (Interface).
Explore ECI →First-party longitudinal research tracking what generative search retrieves from exmxc and controlled web properties, with preserved baselines and methodology changes.
Explore the Observatory →exmxc studies structural shifts across the Four Forces, converts evidence into explicit conviction, deploys capital behind the strongest theses, and uses outcomes to refine the next research cycle.
Research precedes conviction.
Conviction precedes capital.
Outcomes update the research.