Methodology: Why AI Isn't a Bubble – And Why That Changes Everything

Last updated: November 14, 2025 | Article published: November 13, 2025

Purpose of This Document

This methodology document explains how "Why AI Isn't a Bubble" was researched, sourced, and analyzed. Transparency in methods allows readers to evaluate the strength of claims and verify information independently.

Research Philosophy

Core Principle: Follow the Money, Not the Marketing

This analysis focuses on verifiable financial commitments, physical infrastructure projects, and actual corporate spending—not press releases, aspirational projections, or promotional claims. When companies sign contracts, break ground on facilities, or report earnings, those actions reveal truth better than any announcement.

Transparency Standard

Every specific claim in the article is backed by a verifiable source. Readers can check our work. When sources conflict, we note the discrepancy. When data is unavailable, we say so explicitly rather than speculating.

Worker-Centered Analysis

While documenting infrastructure investments, this research consistently asks: "Who benefits from this wealth creation?" The analysis tracks both capital deployment AND its impact on employment, wages, and communities.

Data Collection Methods

Primary Sources Prioritized

Government and Regulatory Bodies: Bureau of Industry and Security (export controls), NY Senate legislation, Federal Reserve statements. These sources provide authoritative policy information.
Corporate Financial Disclosures: AMD and NVIDIA investor relations, earnings calls, SEC filings. Public companies must report accurate financial information or face legal consequences—making these highly reliable.
Industry Research Firms: Dell'Oro Group, Omdia, Statista. These firms specialize in data center market research and sell their analysis to industry players—accuracy is their product.
Trade Publications: Data Center Knowledge, CIO Dive, Utility Dive. Industry-specific outlets with subject matter expertise and access to primary sources.
Local News Sources: El Paso Matters, Planet Detroit, Bridge Michigan, Detroit News. Local journalists often provide more detailed coverage of community impacts than national outlets.
Corporate Subsidy Tracking: Good Jobs First, a nonprofit that tracks government incentives to corporations. Their data comes from FOIA requests and public records.

Secondary Sources Used Carefully

Secondary sources (aggregators, commentary, analysis) were used only when:

Sources Explicitly Avoided

Verification Standards

Financial Claims

Example: AMD's $100B Projection

Claim: AMD projects $100 billion in annual data center chip revenue within 5 years.

Verification Process:

Result: Claim substantiated with multiple independent confirmations.

For Infrastructure Spending: We required either (1) official company announcements with specific dollar amounts, (2) construction permits showing actual projects, or (3) power utility negotiations revealing capacity requirements. Vague "plans to invest" language was excluded unless backed by concrete actions.
For Job Displacement: We used Goldman Sachs analysis citing ADP data (national employment database) plus specific company announcements of AI-driven workforce reductions. When exact numbers were unavailable, we stated estimates clearly as such.
For Tax Subsidies: We relied on Good Jobs First tracking (using FOIA requests and public records), local news coverage of specific deals, and government documents when available. States that don't disclose subsidy amounts are explicitly noted.

Conflicting Information

When sources provided different numbers for the same metric, we:

Example: Data Center Spending Projections

Different research firms projected 2025 data center spending between $550B-$650B. We used $598B (mid-range) and cited multiple sources to show the range. This approach acknowledges uncertainty while providing readers a defensible baseline.

Analytical Framework

Distinguishing Real vs. Speculative Investment

To determine whether AI infrastructure represents genuine economic activity (not a bubble), we assessed:

1. Physical Assets: Are data centers being built? Can we verify construction through permits, local news, or utility connections? Physical infrastructure can't be faked.
2. Contracted Commitments: Are chip orders pre-paid? Are multi-year contracts signed? Actual financial commitments (not press releases) indicate genuine demand.
3. Revenue and Margins: Are companies earning money or burning cash? High profit margins (AMD's 55-58%, NVIDIA's 74-75%) indicate real demand with pricing power—the opposite of speculative bubbles where prices collapse.
4. Expert Validation: What do Federal Reserve economists and Goldman Sachs analysts say? When conservative financial institutions validate spending levels as "sustainable," that carries weight.
5. Government Treatment: Does the government consider this strategic infrastructure or speculative hype? Export controls on AI chips to China indicate national security priorities—governments don't restrict speculative assets.

Wealth Distribution Analysis

To assess "who captures the wealth," we tracked:

This framework reveals whether productivity gains from AI flow broadly or concentrate at the top.

Limitations and Uncertainties

What We Know with High Confidence

What Involves Estimation

What We Explicitly Don't Know

When information is unavailable, the article states this clearly rather than speculating.

Source Archiving

To ensure permanent accessibility, all sources were submitted to the Internet Archive's Wayback Machine between November 13-14, 2025. This protects against:

Archive Status:

For failed archives, original URLs remain documented and alternative sources were found when possible.

Editorial Standards

Fact-Checking Process

Corrections Policy

If errors are discovered post-publication:

Updates Policy

As new information emerges:

Update: November 14, 2025

Added: ACS-BlackRock $27 billion data center deal (announced Nov 13, 2025). This addition increased documented infrastructure investment from $1.0T to $1.027T through 2028. Source archived via Wayback Machine.

Distinguishing Analysis from Advocacy

This article combines objective documentation with normative analysis:

Objective Documentation: Infrastructure spending amounts, profit margins, job displacement data, tax subsidy totals—these are facts derived from verifiable sources.
Normative Analysis: Conclusions about fairness, wealth distribution, and policy recommendations reflect the author's perspective. Readers may agree with the facts but reach different conclusions about their meaning.

The methodology aims to make facts vs. interpretation clearly distinguishable.

Questions and Verification

Readers who wish to verify claims, report errors, or ask methodology questions can:

Final Note on Methodology

Perfect certainty is impossible in real-time analysis of rapidly evolving situations. This methodology prioritizes transparency over false confidence. By showing our sources, explaining our reasoning, and acknowledging uncertainties, we give readers the tools to evaluate our work critically.

If you find errors, tell us. If you have better data, share it. The goal is accurate understanding, not being right.

Article: Why AI Isn't a Bubble – And Why That Changes Everything
Published: November 13, 2025 | Methodology Updated: November 14, 2025
Author: Angela Thornton | Publication: The Open Record