---
title: "Data-Centre Real Estate"
slug: datacenter-real-estate
type: data-index
sector: real-estate
canonical_url: https://deepstoryresearch.com/data/datacenter-real-estate
series: [dc_vacancy, preleasing, absorption]
series_count: 3
data_point_count: 7
data_as_of: unreviewed
source_quality: curated_snapshot
tier: free
generated_at: 2026-10-21
---

# Data-Centre Real Estate
_Record-low vacancy, heavy preleasing and record absorption, the opposite end of commercial real estate from the office market._

> Deepstory Research context file · **Free tier** · <https://deepstoryresearch.com/data/datacenter-real-estate>
> Self-contained AI briefing. Drop into an LLM window or RAG pipeline.

## AI research use contract

- Treat this file as source-grounded context, not a live database. Quote `source_ref` and `data_as_of` when using numbers.
- Separate `FACT`, `ESTIMATE`, `FORECAST`, `TARGET`, and `INFERENCE`. If a row basis is unclear, say it is unclear.
- Prefer T1/T2 sources for hard facts; use T3/T4 sources as context that should be checked before money, legal, medical, operational, or policy decisions.
- Keep answer structure clean: first say what the data says, then what you infer, then what would change the conclusion.

## TL;DR: index summary

Data-centre vacancy in North America’s primary markets fell to an all-time low of 1.9% at the end of 2024, then 1.6% in H1 2025 and 1.4% at year-end 2025, and it kept falling while supply grew 36% year on year to 9,432 MW. Demand is outrunning a construction boom. Read this against the us-office-vacancy index: same broad asset class, same quarter, one at a record high of 21% and the other at a record low of 1.4%.

Around 79% of capacity under construction in H1 2024 was already preleased, and 74.3% in H1 2025: overwhelmingly to cloud and AI providers locking in future capacity. Only a handful of 10 MW-plus facilities due for 2025 delivery were still unleased. Buildings are spoken for years before anyone can occupy them.

Net absorption hit a record 2,497.6 MW in 2025, beating the previous record of 1,809.5 MW set in 2024, a 38% increase. Absorption measures space actually taken up rather than built, so it is the cleaner demand signal of the two.

## What this index covers

Record-low vacancy, heavy preleasing and record absorption, the opposite end of commercial real estate from the office market.

This file is generated from the index spec and the seeded rows. Its per-series sections carry the reviewed chart notes verbatim; it has no separate editorial layer (no glossary, entity table, timeline or cross-series synthesis), and no reviewed "data as of" date.

**Series in this index:**

- `dc_vacancy`: Primary-market vacancy rate, %. Successive record lows.
- `preleasing`: Percentage of under-construction capacity already leased, %.
- `absorption`: Net take-up of data-centre capacity in primary markets, MW.

## Evidence basis map

The free tables above show `source_ref` but not the row-level `value_basis`. This section mirrors the visible rows only, so AI tools can separate observed values from estimates, forecasts, targets, and derived values.

| series | row | source_ref | value_basis |
| --- | --- | --- | --- |
| dc_vacancy | checkpoint=End 2024; pct=1.9 | cbre-vacancy-pr | North American primary-market data-centre vacancy 1.9% at end-2024, an all-time low at the time |
| dc_vacancy | checkpoint=H1 2025; pct=1.6 | cbre-h1-2025 | Primary-market vacancy fell to a record-low 1.6% in H1 2025 |
| dc_vacancy | checkpoint=End 2025; pct=1.4 | cbre-h2-2025 | Primary-market vacancy fell further to a record low 1.4% at year-end 2025 |
| preleasing | period=H1 2024; pct=79 | cbre-vacancy-pr | Nearly 80% of the 3,871.8 MW under construction in H1 2024 was preleased (3,056.4 MW) |
| preleasing | period=H1 2025; pct=74.3 | cbre-h1-2025 | 74.3% of capacity under construction preleased in H1 2025, primarily to cloud and AI providers |
| absorption | year=2024; mw=1809.5 | cbre-h2-2025 | Record net absorption of 1,809.5 MW in 2024 (the previous record) |
| absorption | year=2025; mw=2497.6 | cbre-h2-2025 | Record net absorption of 2,497.6 MW in 2025, beating the 2024 record |

## The series (per-chart briefings)

### North American data-centre vacancy: line
**Direct answer:** Primary-market vacancy rate, %. Successive record lows.

| checkpoint | pct | source_ref |
| --- | --- | --- |
| End 2024 | 1.9 | cbre-vacancy-pr |
| H1 2025 | 1.6 | cbre-h1-2025 |
| End 2025 | 1.4 | cbre-h2-2025 |

**Read:** Data-centre vacancy in North America’s primary markets fell to an all-time low of 1.9% at the end of 2024, then 1.6% in H1 2025 and 1.4% at year-end 2025, and it kept falling while supply grew 36% year on year to 9,432 MW. Demand is outrunning a construction boom. Read this against the us-office-vacancy index: same broad asset class, same quarter, one at a record high of 21% and the other at a record low of 1.4%.

### Share of capacity preleased before completion: bar
**Direct answer:** Percentage of under-construction capacity already leased, %.

| period | pct | source_ref |
| --- | --- | --- |
| H1 2024 | 79 | cbre-vacancy-pr |
| H1 2025 | 74.3 | cbre-h1-2025 |

**Read:** Around 79% of capacity under construction in H1 2024 was already preleased, and 74.3% in H1 2025: overwhelmingly to cloud and AI providers locking in future capacity. Only a handful of 10 MW-plus facilities due for 2025 delivery were still unleased. Buildings are spoken for years before anyone can occupy them.

### Net absorption: bar
**Direct answer:** Net take-up of data-centre capacity in primary markets, MW.

| year | mw | source_ref |
| --- | --- | --- |
| 2024 | 1809.5 | cbre-h2-2025 |
| 2025 | 2497.6 | cbre-h2-2025 |

**Read:** Net absorption hit a record 2,497.6 MW in 2025, beating the previous record of 1,809.5 MW set in 2024, a 38% increase. Absorption measures space actually taken up rather than built, so it is the cleaner demand signal of the two.

## How to use this with AI

Paste this file into an LLM context window (or a RAG store) and ask cross-series questions. The tables carry a `source_ref` per row; the source registry below maps each ref to a named source and a trust tier.

### Suggested prompts (multi-series)

```text
Using the tables in this file, describe what has changed across `dc_vacancy`, `preleasing`, `absorption` and cite the source_ref for every figure you quote.
```

```text
Which claims in this file rest on a single source? List them with the ref, and say what a second source would need to show to confirm each one.
```

## Sources & trust

| ref | source | trust tier | url |
| --- | --- | --- | --- |
| cbre-h2-2025 | CBRE: North America Data Center Trends H2 2025 | T3 · Reputable press / research org / OWID | https://www.cbre.com/insights/books/north-america-data-center-trends-h2-2025 |
| cbre-h1-2025 | CBRE: North America Data Center Trends H1 2025: record-low vacancy | T3 · Reputable press / research org / OWID | https://www.cbre.com/insights/briefs/north-america-data-center-trends-h1-2025-ai-and-hyperscaler-demand-lead-to-record-low-vacancy |
| cbre-vacancy-pr | CBRE: North American data center vacancy rates hit record low | T3 · Reputable press / research org / OWID | https://www.cbre.com/press-releases/north-american-data-center-vacancy-rates-hit-record-low-despite-significant-supply-growth |

## Caveats & what this index cannot answer

- This file has no reviewed "data as of" date: check `last_updated` and the source URLs before treating any figure as current.
- It carries no glossary or entity table, so terms are used as the sources use them.
- Estimates and forward targets are labelled in the working dataset; never read a labelled estimate or forecast as a settled figure.
- This is an observational data index, not investment, legal, or medical advice.

## Data freshness & methodology

- **Last updated:** 2026-10-21
- **Data as of:** unreviewed

**Methodology (as seeded):**

> Covers CBRE’s North American PRIMARY data-centre markets only, > Northern Virginia, Dallas, Chicago, Phoenix, Silicon Valley and
> similar. Secondary and tertiary markets are excluded, and they behave
> differently, so this is not a national vacancy rate.
>
> Three measures that answer different questions. VACANCY is what is
> available now. PRELEASING is what is spoken for before it is built.
> NET ABSORPTION is capacity actually taken up over a period, which is
> the cleanest demand signal of the three because it nets out both new
> supply and move-outs.
>
> The striking fact is the combination: vacancy fell to successive
> record lows (1.9% → 1.6% → 1.4%) WHILE supply grew 36% year on year to
> 9,432 MW. Demand is outrunning a construction boom, not filling slack.
>
> CAVEAT: the small dip in preleasing (79% → 74.3%) should not be read as
> cooling demand. It is a ratio against a much larger construction
> denominator, so a lower percentage can still represent more absolute
> preleased megawatts.
>
> Deliberate cross-reference with us-office-vacancy: 21% office vacancy
> at a record high against 1.4% data-centre vacancy at a record low, in
> the same market and the same quarter.
>
> Re-verified 2026-08-11.

---

_Free tier. The full row-level dataset and per-source detail live on the live page and in the working file. Canonical: <https://deepstoryresearch.com/data/datacenter-real-estate>. Deepstory Research · https://deepstoryresearch.com_
