← Overview

EAD Research Programme

The EAD Research Programme is a connected sequence of five working papers by triNetra Research on External AI Dependence. It covers the financial survivability of independent businesses (EAD-2026-01), why enterprises fail despite capable systems (EAD-2026-02), how asset ownership and infrastructure participation build resilience (EAD-2026-03), field evidence from AI builder communities (EAD-2026-04), and how AI hardware is accounted for (EAD-2026-05).

The question

How does dependence on externally owned AI infrastructure shape the survivability, governance and economics of independent businesses and institutions?

5 hypothesesJuly 2026 to September 2026Start with EAD-2026-01

Shubham Agarwal, Founder and Chief Executive Officer, triNetra Research

Published individually on SSRN. View the SSRN author page

Hypotheses

Hypothesis 1EAD-2026-01

External AI Dependence and Startup Financial Survivability

Examines how dependence on externally owned AI infrastructure is associated with independent-business survivability. The income-adjusted cost layer is roughly 12 to 20 percent of average annual formal income in India and Brazil, against about 1.5 percent in the United States.

Working PaperJuly 202616 pages

Open PDF ↓
Cite this

Agarwal, S. (2026). External AI Dependence and Startup Financial Survivability. triNetra Research, EAD-2026-01.

Hypothesis 2EAD-2026-02

The Judgment Layer

Proposes why large enterprises fail repeatedly despite sophisticated operational systems: the absence of a judgment layer that turns system outputs into documented reasoning at consequential decisions. Grounded in fifteen organisations and more than USD 80 billion in documented consequences.

Working PaperJuly 202640 pages

Open PDF ↓
Cite this

Agarwal, S. (2026). The Judgment Layer. triNetra Research, EAD-2026-02.

Hypothesis 3EAD-2026-03

The Infrastructure Loop

Asks what changes in the AI economy if Specialised Digital Asset revenue is directed back into infrastructure ownership and community knowledge transfer. Builds on EAD-2026-01.

Working PaperJuly 202611 pages

Open PDF ↓
Cite this

Agarwal, S. (2026). The Infrastructure Loop. triNetra Research, EAD-2026-03.

Hypothesis 4EAD-2026-04

PaaF in the Field

Applies PaaF to field observation of independent AI builder communities, June to July 2026. Eighteen observations across four layers point to one condition: no judgment layer at any level. The field evidence base for EAD-2026-01, 02 and 03.

Working PaperJuly 202633 pages

Open PDF ↓
Cite this

Agarwal, S. (2026). PaaF in the Field. triNetra Research, EAD-2026-04.

Hypothesis 5EAD-2026-05

The Effective Life Problem

Argues AI hardware is mis-accounted as a single-use capital input: units stay functional after they stop being competitive for their original workload, then are resold, redeployed or recovered.

Working Paper, Version 2.0September 202620 pages

Open PDF ↓
Cite this

Agarwal, S. (2026). The Effective Life Problem. triNetra Research, EAD-2026-05.

Frequently asked questions

What is the EAD Research Programme?
The EAD Research Programme is a connected sequence of five working papers by triNetra Research on External AI Dependence. It covers the financial survivability of independent businesses (EAD-2026-01), why enterprises fail despite capable systems (EAD-2026-02), how asset ownership and infrastructure participation build resilience (EAD-2026-03), field evidence from AI builder communities (EAD-2026-04), and how AI hardware is accounted for (EAD-2026-05).
What question does the EAD Research Programme ask?
How does dependence on externally owned AI infrastructure shape the survivability, governance and economics of independent businesses and institutions?
How many hypotheses does the EAD Research Programme contain?
5 hypotheses, coded EAD-2026-01 to EAD-2026-05 and published July 2026 to September 2026. Each hypothesis is one PDF.
Which hypothesis should I read first?
Start with EAD-2026-01, External AI Dependence and Startup Financial Survivability. Examines how dependence on externally owned AI infrastructure is associated with independent-business survivability. The income-adjusted cost layer is roughly 12 to 20 percent of average annual formal income in India and Brazil, against about 1.5 percent in the United States.
Who wrote the EAD Research Programme?
Shubham Agarwal, Founder and Chief Executive Officer, triNetra Research.
Are the EAD papers on SSRN?
Yes. Each paper is published individually on SSRN, and the SSRN author page for Shubham Agarwal lists them all: https://ssrn.com/author=12335668