Loading…
Training cluster spanning Memphis, Tennessee and Southaven, Mississippi, generating power on site rather than waiting for utility interconnection.
What is established
Under construction. 2 GW announced (gross site power); 440,000 accelerators (NVIDIA GB200/GB300); expected Dec 2027; owned or developed by xAI; power source: On-site generation.
What is not disclosed
Project capex, named tenant, cooling technology — not disclosed by any source on this page. Recorded as unknown, never as zero.
Analyst note
On-site generation is the defining feature: the site sidesteps the interconnection queue, which is why its timeline is faster than grid-dependent peers. CORRECTION 2026-08-01: the 2 GW figure is Musk's claimed total site capacity after the third-building purchase (Dec 2025), not current capacity — Epoch estimates 946 MW IT operating (Aug 2026), projecting 1,531 MW by Q1 2027. The prior 555k GPU figure was an Introl estimate; Musk's own statement was 550k GB200/GB300s as a first batch, and Epoch estimates ~440k Blackwell chips installed, which is now recorded.
Corroborated by more than one source, but not yet backed by physical evidence.
Power readiness
Capacity basis stated
The source distinguishes IT load from total site power or on-site generation.
Derived from the evidence on this page by a published rule, not a model. How this is calculated.
NVIDIA GB200/GB300
35.0651, -90.0673
No companies linked to this project yet.
| Metric | Value | Confidence | As of | Basis / methodology | Source |
|---|---|---|---|---|---|
| Power | 2,000 MW | Medium | Not disclosed | Basis: site power. Musk-claimed total after third building; ~946 MW IT operating per Epoch (Aug 2026), 1,531 MW projected Q1 2027. | NextBigFuture |
| GPU count | 555,000 GPUs | Low | Not disclosed | Reported approximate accelerator count at full build. | NextBigFuture |
NextBigFuture · Not disclosed · accessed Jul 26, 2026
Trade-press compilation of announced capacity. Figures require primary-source confirmation.
Reliability
55/100
Epoch AI · Aug 1, 2026 · accessed Aug 1, 2026
1,112k H100-equivalents of compute, supported by 946 MW of IT power
Reliability
85/100
On-site generation is the defining feature: the site sidesteps the interconnection queue, which is why its timeline is faster than grid-dependent peers. CORRECTION 2026-08-01: the 2 GW figure is Musk's claimed total site capacity after the third-building purchase (Dec 2025), not current capacity — Epoch estimates 946 MW IT operating (Aug 2026), projecting 1,531 MW by Q1 2027. The prior 555k GPU figure was an Introl estimate; Musk's own statement was 550k GB200/GB300s as a first batch, and Epoch estimates ~440k Blackwell chips installed, which is now recorded.
Correction: GPU estimate 555k→440k installed (Epoch); 2 GW clarified as Musk's post-third-building target, ~946 MW IT operating
Aug 1, 2026
System
Created from NextBigFuture reporting.
Jul 26, 2026
System
Oracle · Abilene, Texas, United States
Galaxy Digital · Dickens, Texas (Dickens County), United States
Nebius Group · Butler Township, Pennsylvania, United States
Applied Digital · Ellendale, North Dakota (Dickey County), United States
Meta · New Albany, Ohio, United States
Related Digital · Saline Township, Michigan, United States