A working-paper series on the theory of the AI-native firm
Classical theory of the firm carries four implicit axioms: value grows with accumulation; assets depreciate and liabilities are repaid to zero; firm value equals the present value of cash flow; marginal cost is positive and hard to attribute per unit.
This series argues — and proves, where proof is possible — that all four are rewritten for a class of firms whose capability base accumulates almost without bound but whose value delivery is bandwidth-capped. The claim is not that these firms are better. It is that they are not obtainable as a parameter limit of the traditional firm, and so require a distinct analytical object.
All papers are pure theory. Numbers are illustrative calibrations, not empirics. Each paper states an explicit refutable interface — the conditions under which it would be overturned.
Paper I ──► Is value finite when accumulation is unbounded?
Yes — saturation decoupling. Value's finiteness is
independent of the growth rate.
│
▼
Paper II ──► Granting finiteness, how should such firms be valued?
Cash-flow anchoring incurs a strictly positive,
closed-form misspecification cost, monotone in
distance to saturation. FACE re-anchors on net worth.
│
▼
Paper III ──► What do I and II jointly imply?
Four axioms of the traditional firm are rewritten,
interlockingly, and irreducibly. A new species.
| # | Title | Status | EN | 中文 | Notes |
|---|---|---|---|---|---|
| I | Saturation Decoupling: Finite Firm Value under Unbounded Capital Accumulation | Working paper · v2 · Jun 2026 | → | ||
| II | Saturation Room and Firm Value: A Valuation Framework for AI-Native Firms | Working paper · v1 · Jun 2026 | → | ||
| III | The AI-Native Firm as a New Species: A Research Programme | Research programme · v1 · Jun 2026 | → |
→ Full index with abstracts and open objections: /papers
Analytical tools extracted from the papers, usable independently.
| Framework | One-liner | From |
|---|---|---|
| Saturation Room | Value lives in the gap S = x̄ − x, not in current cash flow |
Paper II |
| FACE Valuation | Re-anchor valuation from cash-flow history to net-worth accumulation | Paper II |
| Non-Monetary Balance Sheet | Five assets, five liabilities, direction set by meta-capability | Paper II |
| Amdahl Bottleneck | Why accelerating execution lengthens the value–cash-flow lead | Papers I & II |
The programme is unfinished by design. What is genuinely unsolved:
- Operationalizing the species criterion — how is "axiom rewritten" measured in data?
- Measuring the non-monetary balance sheet — the A/C taxonomy is proposed, not validated
- Identification of (𝒩, h, κ) in FACE — consistency is proved; uniqueness is not
- A Coasean boundary model where judgment bandwidth, not transaction cost, sets the boundary
- Panel-data empirics for every testable prediction in Papers I and II
The exogenous-shock design in Paper II §6.1 — frontier-model releases as date-stamped, firm-exogenous shocks — is the most readily identified and the natural first empirical target.
Work in progress: half-formed arguments, reading notes, counterexamples.
Less polished, more frequent. → /notes
@misc{mei_ainative_2026,
author = {Mei, Junjie},
title = {AI-Native Research: Working Papers on the Theory of the AI-Native Firm},
year = {2026},
url = {https://github.com/MedocMay/AI-Native-research},
note = {ORCID: 0009-0002-8909-8229}
}Individual papers carry their own citation blocks in their subdirectories.
The point of working in the open is to be corrected faster. The strongest objections to each paper are listed in its own README under "Why it might be wrong." If you have a better one — or a counterexample — open an issue.
Collaboration on the empirical work is actively sought.
Contact: maymedoc@gmail.com
CC BY 4.0 · Junjie Mei (Dan) · FrontierFirm.Fund