Dataset · Power · Updated 2 Nov 2026
- Sourced facts
- 52Each with source, quote and date
- Markets
- 17Countries and states
- Sites and companies
- 25Linked to the facility and operator records
- Distinct sources
- 48Primary and government first
Every value carries its source, a verbatim quote and the date it was true. No price is published: each request is scoped and quoted.
What is in the power deals dataset?
Data center power deals are the contracts and connections that supply a campus: long-term power purchase agreements, on-site generation such as gas turbines or fuel cells, and positions in the utility's connection queue.
Power is now the first constraint on new capacity. Who has secured power, from what source and when, tells you which projects will be built first.
What is inside the dataset
One row per fact, linked by id to the facility, the operator and the market, so it joins with every other DC Market Insights dataset.
| Field | What it holds | Facts held |
|---|---|---|
| Power purchase agreement | Buyer, seller, MW, technology and term. | 29 |
| On-site generation | Technology and MW at or next to the campus. | 23 |
| Grid queue position | Requested MW and status with the utility or grid operator. | 0 |
| Nuclear agreement | Restart, uprate or small reactor deals, as stated. | 0 |
| Source, quote and date | The link, the exact words of the source and the date the value was true, on every row. | All |
Formats: Excel or CSV, one row per deal or queue entry.
Preview: five rows from the dataset
The five most recently checked facts. The full dataset holds 52 facts.
| Site or company | Market | Field | Value | As of | Source |
|---|---|---|---|---|---|
| VIRTUS Data Centres (UK data centres) | United Kingdom | Power purchase agreement | 31 MW, signed: tri-party corporate PPA (with energy consultant Inspired) for 31 MW of wind power from the Lynn and Inner Dowsing offshore windfarms (about 16% of their output), commencement 1 Oct 2025, long-term; sites not named | 2 Sep 2025 | Company source, on file |
| Novva Data Centers, West Jordan campus | Utah, United States | On-site generation | 3,271.8 kW per engine, approved equipment: Seventy-two (72) Jenbacher JGS 620 J715 natural gas-fired generator engines, Rating: 3271.8 kW (4,601 hp) each engine, as written | 19 Dec 2024 | Government record |
| Global Switch | United Kingdom | Power purchase agreement | 70 GWh annually, First CPPA sleeved by Drax Energy Solutions (one of two large UK CPPA transactions Drax supported in 2025), accounting for more than 35% of Global Switch's electricity needs | 12 Feb 2026 | Company source, on file |
| Global Switch London Docklands data centre | United Kingdom | Power purchase agreement | 70 gigawatt hours per year, Eight-year PPA signed with RWE for power from the Brechfa Forest West onshore wind farm, supplying the London Docklands data centre from January 2026 until 2033 | 22 Jan 2026 | Company source, on file |
| Kao Data, Green Data Solar Farm PPA for Harlow campus | United Kingdom | Power purchase agreement | 40 MW, PPA, agreement signed: 40MW solar PV farm (Green Data Solar Farm) to be developed, built and operated by Downing Renewable Developments, private wire supply to Kao Data's Harlow campus | 18 Nov 2025 | Company source, on file |
Where: coverage by market
| Market | Facts | Sites or companies | Last checked |
|---|---|---|---|
| United States | 22 | 9 | 30 Oct 2026 |
| United Kingdom | 8 | 6 | 2 Nov 2026 |
| Spain | 2 | 1 | 29 Oct 2026 |
| Ireland | 2 | 2 | 29 Oct 2026 |
| Pennsylvania, United States | 2 | 2 | 29 Oct 2026 |
| Netherlands | 2 | 2 | 1 Nov 2026 |
| New Jersey, United States | 2 | 1 | 1 Nov 2026 |
| Germany | 2 | 2 | 1 Nov 2026 |
| Iceland | 2 | 1 | 30 Oct 2026 |
| Italy | 1 | 1 | 29 Oct 2026 |
| Georgia, United States | 1 | 1 | 30 Oct 2026 |
| Utah, United States | 1 | 1 | 2 Nov 2026 |
| Finland | 1 | 1 | 29 Oct 2026 |
| Colorado, United States | 1 | 1 | 1 Nov 2026 |
| Texas, United States | 1 | 1 | 30 Oct 2026 |
When: how often it is updated
Update cycle: Updated with the Power and Policy line, twice a day, one country at a time. Earlier values are never overwritten, so you can see when a date moved or a figure changed.
Who uses it and for what
Utilities and power developers
Who is buying and how much.
Operators
What peers have secured, at what scale.
Investors
Which projects have power and which do not.
Why you can rely on it
- Primary sources first. We never copy data center directories, research publishers or press-release wires. This dataset is built from:
- Utility and grid operator filings
- Energy regulator dockets
- Company releases and filings
- The words and the date. Each value is stored with the source link, the exact sentence it comes from and the date it was true.
- A second, independent check. Every value is checked again against its source before it is stored. Values that cannot be confirmed are dropped.
- History, not overwrites. A new value is a new record, so changes in dates, MW and status stay visible.
- Our estimates are labelled. Anything we compute, such as a value per MW, is marked derived or DC Market Insights estimate.
How to get it
- 1Tell us the scopeMarkets, companies, sites or your own list, and the fields you need.
- 2Get the quoteScope, delivery date and price, usually within one business day.
- 3Receive the dataExcel or CSV with sources, quotes and dates, plus a gap list. Updates on request.
Tell us the markets, companies or projects you need and the fields that matter to you. We reply with the scope, the delivery date and a quote, usually within one business day.
Request a quoteFrequently asked questions
How do data centers get their power?
From the grid under a utility tariff, through power purchase agreements with generators, and increasingly from on-site generation. Our tracker records each route per campus with its source.
Do you track grid queues?
Yes, where the utility or grid operator publishes them, tied to named projects.
Related datasets
Custom data
Need this for your own list of companies or sites?
Send the list. We match every name to the database, fill the fields you need with a source for each value, and return the gaps.