Sirte Basin · Libya · Est. 2026
Every new oil well releases gas that is usually burned off into the sky. Gas4AI captures it at the wellhead, turns it into electricity on site, and uses that power to run AI datacenters in the Libyan desert.
- 01Flare gas
- 02Power
- 03Compute
01 — The problem
When a well is drilled, associated gas rises with the crude. With no pipeline nearby, operators flare it. That is wasted energy, wasted revenue, and millions of tonnes of CO₂.
*Approximate, based on public satellite flaring estimates (World Bank GGFR). Illustrative figures, to be refined.
02 — Capture
Instead of sending gas to the flare stack, a skid-mounted unit at the wellhead scrubs, dries and compresses it, then pipes it to power generation a few hundred metres away.
- Skid-mounted, deployable within weeks
- H₂S scrubbing & liquids recovery
- Zero routine flaring at the pad
03 — Convert
Modular gas turbines and gensets turn the captured gas into steady baseload electricity. It is an off-grid microgrid that needs no national transmission lines.
04 — Compute
Containerised, liquid-cooled AI racks sit beside the generators. Each DataServerNode trains and serves models around the clock, cooled by dry coolers that use the cold desert nights.
- GPUs per site
- ~5,000
- IT load
- 8.3 MW
- PUE target
- 1.2
- Cooling
- Direct-to-chip liquid
- Backhaul
- Fiber + LEO satellite
- Build time
- < 9 months
05 — Why Libya
Libya holds Africa's largest proven oil reserves and wide stretches of open desert, and sits milliseconds from Southern Europe through Mediterranean subsea cables. The gas is already here, and so is the demand for compute.
Candidate zones (illustrative)
06 — Roadmap
- 2027
Pilot
One well pad, 5 MW, first AI cluster online.
- 2028
Cluster
10 sites across the Sirte Basin, 100 MW.
- 2030
Grid
500 MW+ of flare-powered compute, linked by fiber.
07 — Join us
We partner with operators, investors and AI companies. Bring a well, capital, or demand for compute.