-
What is a modular data center?
-
A factory-built, containerized compute unit. IT, power, cooling,
fire suppression, and security in a 20-ft or 40-ft ISO container.
Site work and module work happen in parallel; the unit ships
factory-tested at full load and energizes on site within days
of delivery.
-
Is this a single product or a catalog of options?
-
A single named cluster sized 2 to 5 MW, productized end-to-end.
Configured to your GPU mix, density, climate, and power source.
Catalog underneath; productized at the SKU level. One PO, one
schedule, one weld-traceability standard.
-
What is the lead time?
-
24 to 34 weeks from order to factory completion. Site-ready in
6 to 8 months including delivery and commissioning. Compare
to 50 to 100+ weeks for the same equipment from Schneider, Eaton,
ABB, or Siemens direct, and 18 to 24 months for a site-built
data center.
-
What about interconnection? Equipment is fast — utilities aren’t.
-
Lead time on the box doesn’t help if the medium-voltage tie-in
queues for 18 months. We bundle interconnect-engineering services
through partner firms in NoVA, ERCOT, the Permian, and the Pacific
Northwest. Talk to us about your specific site early — the equipment
lead time is durable; the permit lead time is what kills schedules.
-
What is the highest kW per rack you have delivered?
-
50 kW per rack on air-cooled deployments (MDC-500). 150 kW per
rack on direct-to-chip liquid (MDC-500 DLC) using Vertiv CDUs.
Designed for NVIDIA H100, H200, GB200, Blackwell, and AMD MI300
server platforms.
-
What standards does your equipment meet?
-
Switchgear: IEC 61439 TTA with parallel UL 891 listing for US
installations. MV equipment: IEC 62271 with IEEE C37 parallel.
NFPA 70 / 75 / 110 for fire and electrical code compliance.
ASCE 7 for structural.
-
What climates can you deploy in?
-
−35°C to +52°C operating envelope. IP54 sealed. No external water
required for closed-loop air-cooled units. Designed for desert
and Arctic conditions; tested at full load before shipment in
both extremes. Free-cooling and DX-hybrid configurations
available for climate-specific optimization.
-
Can it be redeployed to a different site later?
-
Yes. ISO container form factor; standard road and rail
transport. Anti-vibration mounting available for mobile and
seismic applications. Full disconnect and recommissioning
documented in the SAT package; no proprietary tooling required.
-
Can you deliver hardened spec for federal or critical infrastructure?
-
Hardened configurations are available on request — discuss specific
requirements (shielding, accreditation, compliance lens) on the SE
call.
-
Can I purchase direct, or does this run through a partner?
-
Direct purchase from Teraplex is available for end customers.
A separate channel partner program exists for system integrators
and EPCs delivering modular DC and power-infrastructure scopes
to their own clients. Both motions ship from the same factories
to the same standard.
-
Do you offer pre-development engineering and site selection services?
-
Yes. Teraplex is the US-resident development partner. We orchestrate
pre-development engineering (civil and electrical), site selection,
and development planning through a curated A&E, MEP, and EPC partner
network. Engineering labor sits at the partner firms; Teraplex
coordinates delivery and owns the customer relationship. See
development services
for the full scope.
-
Can you support pipeline-level development across multiple sites?
-
Yes. Our development leadership has delivered across multiple
hyperscale and wholesale colo portfolios. We bring pipeline-level
institutional knowledge — sequencing, go/no-go decision frameworks,
and early-stage cost estimation that integrates equipment lead-time
intelligence. Built for developers scaling beyond a single anchor
site.
-
Are pre-development deliverables professionally stamped?
-
Yes. All utility-facing engineering packages — primary service
configurations, load analyses, one-line diagrams — are stamped
by a licensed Principal Electrical Engineer before they leave
Teraplex. Civil deliverables are stamped by the licensed
Professional Engineer at the responsible A&E partner firm.
-
How does pricing work for productized clusters?
-
Standardized cluster configurations carry transparent published
pricing — roughly 80% of every cluster is the same factory-tested
build. The remaining 20% is configurable (cooling architecture,
density, climate envelope, power source) and carries a defined
delta on the standard price.
Configure your cluster
to see the next factory slot, lead time, and indicative price band.
For custom scope outside the standardized configurations, use the
custom quote form.
-
Can I reserve a factory build slot before issuing a full PO?
-
Yes. After configuring your cluster, you can reserve a factory
slot with a deposit. Reservation locks your delivery window and
your standardized price for the agreed reservation period. Full
PO follows reservation; the deposit is credited against the final
invoice. Reservation flow launches alongside the configurator
in Q3 2026; in the meantime, schedule a call with engineering
to scope and reserve manually.
-
Is reserving the same as a binding order?
-
Reservation locks the factory slot and the standardized price
for the documented reservation period. During the reservation
window we finalize the engineering scope (the 20% configurable),
commercial terms, and the full PO. Reservation deposits are
credited against the PO. Cancellation policies are documented
in the reservation agreement; deposits are refundable on the
conditions defined therein.
-
What is the GB300 module?
-
Our productized 1.2 MW modular data center, optimized for NVIDIA
GB300 NVL72 training workloads. Each module houses approximately
8 racks at 150 kW per rack, direct-to-chip liquid cooled,
factory-tested as a unit before delivery. Substation, gensets,
switchgear, and SCADA ship on the same schedule. Site-ready in
6 to 8 months from PO.
-
Why 1.2 MW per module instead of a larger cluster?
-
The 1.2 MW boundary matches the GB300 NVL72 reference architecture
(approximately 8 racks × 150 kW). Productizing at this size makes
the module factory-testable as a single unit, ships in standard
ISO container envelopes, and lets you phase capacity into a
multi-MW campus by stacking modules on a shared power spine.
Each module is a discrete, factory-validated deployment unit.
-
How do modules combine to make a multi-MW campus?
-
A single substation, transformer bank, and SCADA spine are sized
for full campus capacity on day one. Modules phase in alongside
demand — 4 modules = 4.8 MW, 8 modules = 9.6 MW, up to 32+ modules
for hyperscale campus scale. Each module is independently
commissioned but shares the site-level power and monitoring
infrastructure. GPU generations change every 18 to 24 months —
early-phase modules run current hardware, later-phase modules
spec next-generation hardware.
