COMPUTE ATLAS Supercomputer supply-chain graph
1345 systems 511 sites Sourced data

Supercomputers/Systems/OpenAI Project Camellia

USA · planned · ai training

OpenAI Project Camellia

Also known as Effingham County Georgia campus

Operated by OpenAI at Project Camellia Site (Effingham) .

Phase 1 seed dataset, compiled by hand. These rows were built from public operator, laboratory and vendor sources. A mechanical second-reader pass has since fetched every cited source: 1263 of 1345 systems have a readable citation that names them, and 56 are genuinely weakly sourced. Every claim carries its source and a confidence tier. Treat anything below verified as a lead, not a citation.

Measured performance

Rmax
-
Rpeak
-
Rmax ÷ Rpeak
-
Cores
-
Power
-
Per watt
- GF/W

Figures are as last publicly reported for the configuration described below, not a live measurement. Where a system was upgraded in place, the post-upgrade configuration is shown and the earlier one appears in the timeline.

What this machine is made of

One row per supplier relationship. “Supplier at build” is the company that shipped the part at the time; where that company has since been acquired, the parent it rolls up to today is shown beside it. That distinction is what makes ticker-level aggregation possible across a thirty-year dataset.

Role Part Supplier at build Quantity Confidence Source
Integrator -
OpenAI designs and develops the campus itself; a build and operate partner is still to be selected
OpenAI
- Reported news.constructconnect.comeweek.com
Power -
Georgia Power supplies 3.2 GW under a 25-year agreement with up to 1,000 MW of flexible demand response
Georgia Power
- Reported eweek.com

AI datacenter metrics

Figures beyond an accelerator count, each with who stated it. What can and cannot be compared across AI datacenters.

MetricValueAs ofBasisNoteSource
Stated investment 20 USD bn 2026-07-23 Design target Effingham County valuation of the project in a joint county announcement; Bloomberg quotes an OpenAI executive at more than 30 billion at full build-out. The 20 billion is the county figure, not an OpenAI-audited spend. news.constructconnect.comeweek.com
Facility power 3,200 MW 2026-07-23 Design target OpenAI: 3.2 GW contracted with Georgia Power, phased 2028 to 2032, with up to 1,000 MW of flexible demand response. news.constructconnect.comeweek.com

How it is powered

SourceProviderMWStatusAs ofBasisNoteSource link
Grid Georgia Power 3,200 contracted 2026-07 Operator stated OpenAI: 3.2 GW from Georgia Power under a 25-year agreement, phased 2028 to 2032, plus up to 1,000 MW of flexible demand response. eweek.comnews.constructconnect.com

How AI campuses are powered: capacities are what the operator, utility or filing states, not what has been delivered.

The read

OpenAI's first campus that it designs and develops itself, announced on 23 July 2026: four buildings on about 1,400 acres inside the Savannah Gateway Industrial Hub in Effingham County, Georgia. OpenAI says it is contracting 3.2 GW from Georgia Power under a 25-year agreement, delivered in phases from 2028 to 2032, with up to 1,000 MW of flexible demand response and a closed-loop water system. Effingham County officials value the project at 20 billion dollars; Bloomberg quotes an OpenAI executive at more than 30 billion at full build-out. A build and operate partner had not yet been selected and no accelerator type or count is stated, so the status is planned and the tier is reported (both pages relay OpenAI's release).

Timeline

Site & facility

Project Camellia Site (Effingham)

Location
Effingham County, GA, USA
Utility
Georgia Power
Cooling
Closed-loop water system

Change history

Source check

We fetched this system's own citations and recorded whether each page actually mentions it. This is a corroboration signal, not a fact check, and it is published so you can see how well the sourcing holds up rather than take it on trust.

Cited sourceResultFound on the page
news.constructconnect.comnames system only-
eweek.comnames system only-

Checked 2026-10-13 by pnpm verify. Re-run it and the table changes with the web.

Further reading