Eric Flanigam — Generative Value - A Primer on AI Data Centers - October 2024
FreeA deep dive into the AI data center buildout
FreeFree tier
About Eric Flanigam — Generative Value - A Primer on AI Data Centers - October 2024
An in-depth article by Eric Flaningam from his newsletter Generative Value, published in October 2024. It provides an overview of the history, technology, and companies involved in the AI data center buildout. The piece covers the scale of modern AI data centers (up to 1 GW energy consumption), the data center value chain (construction, industrial equipment, computing infrastructure, energy), and an abridged history from early computers like Colossus and ENIAC to today's hyperscaler facilities. It is the first part of a series breaking down AI data center infrastructure and investment opportunities.
Key Features
Overview of AI data center scale and energy consumption
Breakdown of data center value chain: construction, equipment, computing, energy
Historical context from early computing (Colossus, ENIAC) to modern hyperscalers
Discussion of investment opportunities and exposed companies
Part of a multi-part series on AI data center infrastructure
Pros & Cons
Pros
- Well-researched with historical and current context
- Provides a clear value chain framework
- Includes specific examples and comparisons (e.g., NYC energy consumption)
- Free and accessible on Substack
- Part of an ongoing series for deeper dives
Cons
- Limited to a single article; full series not yet completed
- Focuses primarily on US and hyperscaler perspective
- Does not cover specific company financials or valuations
- Could be too introductory for experienced data center professionals
Best For
Investors seeking to understand the AI data center ecosystemTechnology professionals wanting a primer on data center infrastructureStudents of technology history and infrastructure buildoutsAnyone interested in the energy implications of AI data centers
FAQ
What is the main focus of this article?
The article provides a primer on the history, technology, and companies exposed to the AI data center buildout, comparing it to the historical electric grid buildout.
Who is the author?
Eric Flaningam, an investor across public and private markets, writes the Generative Value newsletter.
How large are the largest AI data centers?
The largest hyperscaler data centers consume up to 1 GW of energy capacity, compared to New York City's ~5.5 GW.
What is the data center value chain covered?
The article breaks it into initial construction, industrial equipment, computing infrastructure, energy, and the companies that own/lease data centers.
Is the article part of a series?
Yes, it is the first part of a several-part series on AI data center infrastructure.