data center
A giant warehouse full of computing equipment running around the clock; for AI, these must be packed with NVIDIA GPUs to train and run models.
As defined in The Man Who Built NVIDIA (with Stephen Witt)
Other definitions
Large facility containing computing infrastructure (servers, GPUs, cooling systems, power distribution) that runs AI models and applications; measured in megawatts (MW) or gigawatts (GW) of power consumption
The Geopolitics of AI Infrastructure - Dylan Patel, SemiAnalysisA facility containing large numbers of networked computers and servers that process, store, and distribute massive amounts of digital data, requiring substantial electrical power
Doomberg Stops by the Energy News Beat Channel
Where it comes up · 21
- The Geopolitics of AI Infrastructure - Dylan Patel, SemiAnalysis11 claims
- AI Trends 2025: Mary Meeker Deck Deep Dive Part 19 claims
- Leopold Aschenbrenner says "No More Stocks!"6 claims
- Comics that ask "what if?" | Randall Munroe6 claims
- Eric Schmidt and Craig Mundie with David Rubenstein: Genesis6 claims
- Professor Chris Miller - “Chip War: The Fight for the World's Most Critical Technology”5 claims
- Ex-Google CEO: What Artificial Superintelligence Will Actually Look Like w/ Eric Schmidt & Dave B5 claims
- Nuclear Power's Facelift: Small Modular Reactors || Peter Zeihan4 claims
and 13 more
Most important claims · 10 of 77Search all claims for “data center” →
- Google's 2017 'transformer' architecture distills intelligence from massive data, turning AI into heavy industry requiring giant data centers full of NVIDIA equipment running 24/7—like an oil refinery—which multiplied demand for NVIDIA chips by 100x to 1000x and rocketed its valuation from billions to trillions.The Man Who Built NVIDIA (with Stephen Witt)
- Big Tech companies like Microsoft, Facebook, and Alphabet have become significantly more capital-intensive, with Facebook/Alphabet/Microsoft combined capex increasing from $28 billion in 2017 to an estimated $152 billion in 2024 (over 5x increase), driven by cloud infrastructure and AI investments, creating uncertainty about future Returns on Capital that is obscured by accounting lags where depreciation trails capex.Value Investing Fundamentals & Current Market Conditions w/ John Huber (TIP634)
- When partial information is available for some data centers (e.g., both spending and power contracted with local government), the ratios between these quantities can be used to infer missing information for data centers where only one data point is known.Comics that ask "what if?" | Randall Munroe
- Optical fiber providers (Corning and Light) represent the next evolution in data center networking because copper wires become inefficient and generate excessive heat when used to connect very large numbers of GPUs, making optical fiber the necessary next-generation solution.Leopold Aschenbrenner says "No More Stocks!"
- Fermi estimation of Google's data infrastructure can be constrained using spending data, as large data centers require substantial capital investment and the total capital expenditure limits the number of centers that could be built.Comics that ask "what if?" | Randall Munroe
- Data center power consumption has become more efficient by 50,000x in terms of tokens generated per megawatt-year, scaling from 1.3 billion tokens/megawatt-year to 65 trillion tokens/megawatt-year, which is rarely reported in discussions of AI energy consumption.AI Trends 2025: Mary Meeker Deck Deep Dive Part 1
- Google and Amazon are sponsoring small companies and startups to develop small modular reactors (SMRs) that can fit on the back of a truck and plug into a transformer network, enabling rapid conversion of old coal plants or installation at data center sites.Nuclear Power's Facelift: Small Modular Reactors || Peter Zeihan
- Additional data sources for estimating Google's data center locations include Google recruitment messages that indicate where they employ people, unauthorized cell phone photos of data center hardware posted by visitors, and pizza delivery driver knowledge of facility locations.Comics that ask "what if?" | Randall Munroe
- Over the last decade, data center engineering and construction breakthroughs at Meta and Google were largely taken for granted in the background, but data center engineering is now becoming a primary source of competitive advantage in the AI era.Ep19. State of Venture, AI Scaling, Elections | BG2 w/ Bill Gurley, Brad Gerstner, & Jamin Ball
- Data center power use has decreased 43% over 8 years while per-unit energy efficiency has improved 50,000x, and tokens per megawatt per year have scaled from 1.3 billion to 65 trillion, demonstrating dramatic improvements in compute efficiency that are underreported in energy consumption headlines.AI Trends 2025: Mary Meeker Deck Deep Dive Part 1