Bring your racks. We bring the power.
Lease powered, cooled, connected space for the GPU, CPU, and storage hardware you already own — on a self-powered data center platform in the Gulf of America. No utility interconnection queue. No permits. No neighbors. And a power rate that doesn't move with the grid.
Rack · Cage · Dedicated GenPod · Fixed-rate term leases · Cross-connect to GulfGenAI GPU Cloud
The hardest part of owning GPUs isn't buying them. It's finding somewhere to plug them in.
Power that's already there
Onshore, a new high-density site waits years for a utility interconnect. A GenPod generates its own electricity on the platform, so the power for your racks is there when your hardware is — not when the grid gets around to it.
Cooling that scales with density
The ocean is the heat sink. A closed-loop seawater system delivers liquid cooling to your racks — at densities air-cooled colo can't reach — with zero freshwater draw and zero evaporation.
No permits. No neighbors. No rate hikes.
GenPods operate 100+ miles offshore under existing federal regulation (BSEE, ABS, U.S. Coast Guard, BOEM, EPA). There's no zoning fight, no community moratorium, and no utility tariff to re-price your power mid-contract.
Everything below the rack rails. Delivered as a service.
Power
Self-generated, on-platform electricity delivered to your rack at a committed per-kW rate. Redundant distribution, metered per tenant, with UPS ride-through and generation redundancy at the platform level.
Cooling
Closed-loop seawater cooling with liquid-to-chip (direct-to-chip) manifolds and rear-door heat exchangers for air-cooled gear. Facility water loop maintained and monitored by GulfGenAI.
Network
Diverse, carrier-grade uplinks from the platform to mainland landing points, with on-platform cross-connects to the GulfGenAI GPU cloud and to other tenants. Bring your own carriers or ride ours.
Space
Purpose-built data halls on modified jack-up platforms, engineered for high-density rows. Lease by the rack, by the cage, or take a dedicated hall — with room to grow into adjacent positions.
Security & Compliance
An offshore platform is one of the most access-controlled environments there is. Badge- and escort-controlled halls, locked cages, video coverage, and a U.S.-flagged, federally regulated operation.
Remote Hands & Logistics
Our on-platform crew installs, racks, cables, swaps, and reboots on your behalf, around the clock. We coordinate marine logistics for inbound hardware and spares so you never have to think about a supply boat.
From a single rack to a dedicated GenPod.
Rack Colocation
Individual high-density rack positions in a shared, access-controlled hall. Metered power, liquid cooling, and network uplinks included; you supply and own the servers.
Best for: teams with a handful to a few dozen GPU or CPU racks, pilots ahead of a larger commitment, and storage or CPU tiers adjacent to GPU cloud workloads.
Private Suite / Cage
A dedicated, locked footprint for an entire cluster: contiguous rows, your own network fabric and PDUs, and a committed power block sized to your build with room reserved to grow.
Best for: neoclouds and enterprises deploying a full training or inference cluster who need physical isolation and a multi-year, predictable power cost.
Dedicated GenPod
An entire platform — generation, cooling, halls, and network — provisioned for a single tenant. Effectively a private, self-powered data center with GulfGenAI operating the plant.
Best for: hyperscale and sovereign deployments that want the economics of owning a powered site without building, permitting, or operating one.
| Responsibility | GPU Cloud | Colocation | Hybrid (Colo + Cloud burst) |
|---|---|---|---|
| Fuel, generation, cooling, platform | GulfGenAI Owned and operated end to end. | GulfGenAI Owned and operated end to end. | GulfGenAI Owned and operated end to end. |
| Servers, GPUs, storage | GulfGenAI Our clusters; you consume capacity. | You Your capital equipment, your depreciation schedule. | Shared Your baseline hardware; our GPUs for peaks. |
| Network & cross-connects | GulfGenAI Included in the platform. | Shared We provide uplinks and cross-connects; you run your fabric. | Shared Local cross-connect between your cage and our cloud. |
| Orchestration, OS, workloads | Shared Our control plane; your models and data. | You Full control, full custody. | You One environment spanning both. |
| Pricing basis | GPU-hour or reserved capacity, locked to our cost of power. | Committed $/kW + space, locked to our cost of power. | $/kW for your racks + GPU-hours at your locked rate. |
| Time to first workload | Days, on reserved capacity. | Scheduled at scoping around your hardware delivery date. | Cloud immediately; your hardware as it lands. |
From signed order to powered racks, without a single site visit.
You never have to step on a boat. Ship your hardware to our Gulf Coast staging facility; our crew handles marine transport, installation, cabling, and burn-in, then hands you the keys remotely.
Size the footprint
Share your rack count, per-rack power, cooling method, and network needs. We map it to available GenPod capacity and return a committed $/kW quote.
Lock power and space
Sign a term lease. Your rack positions, power block, and cooling capacity are reserved on the capacity calendar — ahead of your hardware delivery.
Ship to us. We rack it.
Deliver to our Gulf Coast staging facility. We handle marine logistics, racking, power and liquid-cooling hookups, structured cabling, and burn-in per your runbook.
Run it your way
Out-of-band access, per-rack power and thermal telemetry, and 24/7 remote hands from the on-platform crew. Burst into the GulfGenAI GPU cloud over a local cross-connect whenever you need more.
Built for teams who own their compute — or want to.
Neoclouds & GPU Operators
You've secured allocation from the vendor; now you need a powered, liquid-cooled home for those racks this year — at a power rate that doesn't erode your margin every time a utility files a tariff.
Enterprises with Owned or Sovereign Hardware
Regulated industries, defense-adjacent work, and data-custody requirements that rule out a shared cloud. Keep your own hardware, on U.S.-flagged infrastructure, without building a data center.
HPC, Research & Long-Running Jobs
Simulation, rendering, scientific computing, and training runs measured in months. When the job runs 24/7, the power bill is the budget — and ours doesn't move.
The questions every infrastructure team asks first.
How does my hardware get to an offshore platform?
You ship to our Gulf Coast staging facility exactly as you would to any colocation site. From there GulfGenAI manages marine transport, crane-to-hall handling, and installation using the same logistics chain our backer, Enterprise Offshore Drilling, has run in the Gulf for decades. You never need to send staff offshore.
What about latency and connectivity back to the mainland?
GenPods connect to mainland landing points over diverse network paths, and we match tenant port speeds to your fabric during scoping. That link suits training, batch inference, HPC, and most production serving. If your workload is latency-sensitive, share your requirements and we'll walk you through the network topology before you commit.
Who services my hardware if something fails?
Our on-platform crew provides 24/7 remote hands — reboots, cable moves, drive and GPU swaps, and vendor-directed procedures — from a tenant spares locker you stock. For vendor warranty work, we coordinate escorted vendor access or return-to-depot logistics on your behalf.
How is power priced, and what happens if gas prices change?
Colocation power is billed at a committed per-kW rate fixed for the term of your lease. Because the fuel is a dedicated, stranded reservoir we already control — not gas bought on the open market and not electricity bought from a utility — our cost base doesn't track the spot market or a tariff, and neither does your bill.
What cooling do you support for high-density GPU racks?
Direct liquid-to-chip cooling via facility-water manifolds and CDUs, and rear-door heat exchangers for air-cooled equipment. The facility loop is closed-loop seawater — no freshwater, no evaporation, no cooling towers. We'll confirm supply temperature and flow against your server vendor's specifications during scoping.
Is this a regulated, U.S. environment for data residency purposes?
Yes. GenPods are U.S.-flagged platforms operating in U.S. waters under existing federal oversight from BSEE, ABS, the U.S. Coast Guard, BOEM, and the EPA. Your data stays in the United States. Ask us about our compliance roadmap for formal attestations.
Can I combine colocation with your GPU cloud?
That's the hybrid model, and it's the reason to colocate here rather than anywhere else. Run your owned hardware as the baseline and burst into GulfGenAI GPU capacity over an on-platform cross-connect — no egress fees, no second vendor, one contract, and the same locked power economics on both sides.
What happens when a reservoir depletes and the GenPod relocates?
Each reservoir holds decades' worth of energy and is selected to outlast lease terms by a wide margin. Relocation is planned years in advance, coordinated with tenants, and covered contractually: your capacity commitment follows you to the next platform or to another GenPod in the fleet.
Tell us about your racks.
Send us your rack count, per-rack power, cooling method, and target date. We'll come back with available GenPod capacity and a committed per-kW quote.