---
id: "concept-ai-industrial-economics"
type: "concept"
source_timestamps: ["¶2"]
tags: ["ai-economics", "infrastructure"]
related: ["concept-great-value-loop", "claim-ai-bottleneck-electricity", "contrarian-ai-is-industrial", "quote-model-is-chips-cooling", "quote-energy-not-renegotiated"]
definition: "The reality that AI models are not just code, but are deeply tied to physical, industrial assets like chips, cooling, land, and power contracts."
sources: ["futures"]
sourceVaultSlug: "hbr-seg-futures"
originDay: 2
articleStem: "hbr-nm-101-energy-strategy-ai"
sourceUrl: "https://hbr.org/2026/06/your-company-needs-an-energy-strategy-for-ais-next-phase"
sourceTitle: "Your Company Needs an Energy Strategy for AI’s Next Phase"
---
# Industrial Economics of AI

## Definition
The reality that AI models are not just code, but are deeply tied to physical, industrial assets like chips, cooling, land, and power contracts.

## The Paradigm Shift
AI is no longer just software or code; its underlying economics are **fundamentally industrial**. An AI model is inextricably linked to physical assets: chips, cooling systems, land, interconnection rights, and power contracts. As the source puts it directly: [[quote-model-is-chips-cooling]].

Because of this, scaling AI is subject to the constraints of the physical world — local grid capacities, permitting delays, and the slow build times of power-generation facilities — rather than just the marginal cost of software distribution. Energy in particular behaves nothing like a renegotiable annual input: [[quote-energy-not-renegotiated]].

## Connections
- Drives [[claim-ai-bottleneck-electricity]] — physical constraints are why the bottleneck is migrating to electricity.
- Is the operative mechanism behind the current phase of [[concept-great-value-loop]].
- Directly challenges the conventional SaaS mental model — see [[contrarian-ai-is-industrial]].

## Enrichment (external validation)
External analyses reinforce that AI is gated by industrial hardware: the World Economic Forum frames grid connectivity as *"the binding constraint"*; Morgan Stanley forecasts U.S. data-center demand of ~74 GW with a ~49 GW power shortfall by 2028; industry analysis identifies high-voltage transformers, switchgear, and grid-tie batteries as *"100% of the bottleneck,"* with transformer lead times stretching to ~5 years; and Brookings notes training a single frontier model may require ~5 GW.


## Related across articles
- [[concept-new-ai-triad]]
- [[claim-capex-obsolescence]]
- [[contrarian-physical-limits]]