All research sources
Sources & Citations

The Two Most Valuable Assets in the AI Race Are Not What You Think

62 primary sources 10 categories We publish the receipts
The essay The Two Most Valuable Assets in the AI Race Are Not What You Think
A

The clock on every layer

Epoch AI · Stanford HAI · Andreessen Horowitz

Open Models Report

Epoch AI View source

The gap between the best open and closed models (ECI)

Epoch AI View source

2026 AI Index Report — Technical Performance

Stanford HAI View source

LLMflation: LLM inference cost is going down fast

Andreessen Horowitz View source

LLM inference price trends

Epoch AI View source

H100 GPU rental pricing report

Chosun Biz View source

The Great GPU Shortage: Rental Capacity

SemiAnalysis View source

Inside the AI Bottleneck: CoWoS, HBM & 2/3nm capacity through 2027

Fusion Worldwide View source

The AI Wrapper Trap: When Your Moat Is Someone Else’s API Call

Tian Pan View source
B

The energy objection

Lawrence Berkeley · Ryan Wiser · CBRE

Interconnection queue data

Lawrence Berkeley National Lab / Blizzard Power View source

The latest interconnection queue data

Ryan Wiser, LBNL View source

Global Data Center Trends 2026

North American data center market set records in 2025

North America Data Centers — Market Dynamics

Powering Intelligence — Executive Summary

Energy and AI — Energy demand from AI

International Energy Agency View source
C

Distillation and copyability

Beck Reed Riden LLP · Reuters · Dropbox Engineering

Understanding AI distillation in the trade-secret context

Beck Reed Riden LLP View source

OpenAI accuses DeepSeek of distilling US models to gain advantage

Logits distillation to smaller models

Dropbox Engineering View source

Distillation from frontier open models

arXiv 2410.18588 View source
D

Proprietary data

Hagiu & Wright · Casado & Lauten · Reuters

Data-enabled learning, network effects and competitive advantage

Hagiu & Wright View source

The empty promise of data moats

Casado & Lauten, a16z View source

US judge approves Anthropic’s $1.5 billion copyright settlement

Why Anthropic’s copyright settlement changes the rules for AI training

Jones Walker LLP View source

Reddit says it’s made $203M so far licensing its data

TechCrunch View source

Reddit AI content-licensing deal with Google

OpenAI, News Corp strike content deal

The Wall Street Journal View source

AI models collapse when trained on recursively generated data

Shumailov et al., Nature View source

A critique of the model-collapse mechanism

arXiv 2410.12954 View source
E

Critical thinking and the skill gap

Brynjolfsson · Shin · Huang

Generative AI at Work, Quarterly Journal of Economics

Brynjolfsson, Li & Raymond View source

LLMs in problem reframing

Shin, Polyanskaya, Lucero & Oulasvirta (Aalto) View source

Generative AI, human capital and the locus of creative advantage

Huang, Jin & Li View source

The Superforecasters’ track record

Good Judgment View source
F

What framing looks like

George Pólya (1945) · Michael Erard · Kees Dorst

How to Solve It

George Pólya (1945) View source

How to build a metaphor to change people’s minds

Michael Erard, Aeon View source

The core of ‘design thinking’ and its application

Kees Dorst View source
G

The gigawatt problem

BuilderMuse · Moduledge · Data Center Dynamics

Data center construction timeline, phase by phase

BuilderMuse View source

NVIDIA Blackwell — per-rack power

Moduledge View source

NVIDIA prepares industry for 1 MW racks and 800-volt DC

Data Center Dynamics View source

800 V HVDC architecture for the next generation of AI factories

NVIDIA Developer View source

The 1 MW AI IT rack is coming — and it needs 800 VDC

Schneider Electric View source

The 800V DC rack transition: Rubin Ultra and the last 50 feet

Supercomputing News View source

Modular data centers for AI — stranded-asset economics

Moduledge View source

Corporate America cracks down on AI spending

Bloomberg View source

UBS finds enterprises throttling AI spending

UBS via Let’s Data Science View source

Chinese open-weight models and enterprise policy risk

Copilot model availability — Kimi K2.7, Grok 4.5

GitHub Changelog View source

How to build a data center without going over budget in 2025

Ingenious.Build View source
H

Why framing is scarce — and trainable

Gentner · Loewenstein · Gick & Holyoak

The roles of similarity in transfer

Gentner, Rattermann & Forbus View source

Analogical encoding facilitates knowledge transfer in negotiation

Loewenstein, Thompson & Gentner View source

Schema induction and analogical transfer

Gick & Holyoak (1983) View source

How strategists really think: tapping the power of analogy

Gavetti & Rivkin, HBR View source

Analogical distance in engineering design (near/far)

Fu, Chan et al. View source
I

The two assets are one

Brynjolfsson · US Census Bureau · Management Science

Strength in Numbers: how data-driven decisionmaking affects firm performance

Brynjolfsson, Hitt & Kim View source

Data-driven decision making across manufacturing plants

US Census Bureau (CES WP 16-06) View source

Big-data investment and productivity growth

Management Science (2014) View source

The value of data and algorithms with decentralized skills

Prasanna Tambe View source

Management in the analytics age

McElheran et al., Wharton Mack Institute View source

Netflix and the conservation of attractive profits

Ben Thompson, Stratechery View source
J

What would make me wrong

Lewis & Mitchell · Nicholas Carr · Harvard Business

Evaluating the robustness of analogical reasoning in LLMs

Lewis & Mitchell View source

IT Doesn’t Matter

Nicholas Carr View source

Why IT Does Matter (rebuttal)

Harvard Business School View source
These are the primary sources behind our research. Have a better one, or spot an error? Tell us — we’ll correct it.