exmxc Capital is the deployment layer of the institution's research flywheel. Capital follows research—not headlines—and is committed only when a Four Forces thesis reaches a defined conviction threshold.

Research begins with structural shifts, bottlenecks, and scarcity formation across the Four Forces of AI Power:

• Compute
• Interface
• Alignment
• Energy

The core loop is research → conviction → deployment → feedback. Frameworks, datasets, Signal Briefs, calculations, and decision rules create the thesis; valuation and falsifiers determine whether conviction is strong enough to deploy; market and operating outcomes then feed back into the research.

The primary capital question is not whether a company is exposed to AI. It is which Four Forces constraint is binding, who controls the scarce asset, how durable that scarcity is, and whether the valuation still leaves room for repricing.

1. Structural scarcity can precede repricing.
2. Research must identify the mechanism that converts scarcity into earnings or cash flow.
3. Capital deploys only when conviction and valuation align.
4. Outcomes are evidence: every position can confirm, refine, or falsify the thesis.

Capital positioning is informed by:

• Four Forces of AI Power
• Original datasets, calculations, and knowledge objects
• Dependency vs. Leverage Mapping
• Scarcity-Adjusted PEG (sPEG)
• Scarcity-Durability Framework (SDF)
• Explicit falsifiers and decision rules
• Cross-Force Cycle Compression

Allocation decisions emphasize structural alignment rather than tactical reaction. Longitudinal entity and retrieval research can contribute supporting evidence when relevant.

The sPEG framework formalizes how structural scarcity modifies valuation logic across AI infrastructure layers, linking constraint concentration to capital repricing dynamics. Where sPEG governs valuation — what a structurally scarce asset is worth — the Scarcity-Durability Framework (SDF) governs allocation: whether that scarcity is durable enough, and the rent retainable enough, to size into.

Supplemental indices monitor constraint concentration in:

• AI Infrastructure
• Energy & Power
• Memory & Data Density

These instruments function as structural reference layers for capital allocation — not as investment products.

Conviction is demonstrated through structural calls, not through returns. exmxc Capital does not publish performance figures or track records. It publishes dated, falsifiable scarcity-adjusted reads — and lets the market confirm or refute them over time.

The sPEG Index Series functions as the validation layer. Each publication time-stamps a structural position before the market reprices it. The record below is drawn from the AI Infrastructure Scarcity Index 88-day rerating check (Feb 13 → May 11, 2026), applying a uniform methodology across both observation dates.

• The framework's highest-conviction quartile — the five names below 0.60 sPEG (SK Hynix, Micron, TSMC, ASML, NVIDIA) — each delivered double-digit returns over the 88-day window.
• The memory layer, flagged as the most structurally underpriced, repriced hardest: SK Hynix +218%, Micron +93%.
• The network layer, read as scarce but priced for perfection at 0.93 sPEG, was the only name to decline — validating the framework's caution as precisely as its conviction.
• Index-average sPEG fell from 0.50 to 0.47 despite a 47% average price gain — earnings revisions outpaced price, and the rerating gap widened rather than closed.

The discipline is symmetric. A framework that only records its winners is marketing; a framework that records where it said scarce but fully priced — and was right to hold back — is architecture. Validation is methodological, not promotional: the sPEG series is a structural reference layer, not an investment product, a performance claim, or a solicitation.

The same apparatus is operating on the current cycle. Alphabet's June 2026 $80B capital raise — and Berkshire Hathaway's $10B placement within it — drives 2026 capital expenditure toward $180–190B, roughly double the prior year. That capex acceleration is read directly through the AI Energy & Power Constraint Index, where the binding constraint is no longer compute but the physics, permitting, and capital intensity of power. The framework's task is not to react to the raise, but to price the scarcity it confirms.

Validation runs the full length of the position. Where the sPEG Index Series validates the entry — pricing scarcity before the market reprices it — the Convergence Monitor validates the exit, applying the AI Infrastructure Convergence Framework to time-stamp the state of six independent signals each week, so de-risking is governed by dated signal rather than sentiment. Entry and exit are held to the same standard: dated, falsifiable, and confirmed or refuted in public. A scarcity thesis with an entry and no exit is conviction, not architecture.

1. Observe — Identify structural shifts across Compute, Interface, Alignment, and Energy.
2. Research — Build source-linked evidence, original knowledge objects, valuation work, falsifiers, and a decision rule.
3. Conviction — Require a clear mechanism from structural scarcity to earnings, cash flow, or strategic value.
4. Deploy — Commit capital only when conviction, valuation, sizing, and downside discipline align.
5. Feedback — Compare subsequent market and operating evidence with the original thesis and update the framework.

Aug 05, 2026 · SPCX SPCX Lockup Architecture: Supply Arrives Before Index Demand Maps approximately 5.33 billion shares of potential year-end supply against delayed, provider-specific passive demand.