We De-Risk Before We Break Ground
Greco uses physics-based digital twins — built on OpenUSD and NVIDIA Omniverse technology — to design, validate and de-risk our energy-first campuses before construction. We model the full loop most developers leave to chance: grid interconnection, on-site generation, liquid-cooled compute, and waste-heat export — as one connected system.
Why We Build the Twin First
Risk reduction, not rendering. A digital twin lets us test power, thermal and layout decisions against physics before any concrete is poured — catching the conflicts that, left to construction, become change orders and delays.
Speed to power. Validating the energy and cooling design up front shortens the path to an energized, tenant-ready facility — reinforcing the whole reason we develop energy-first: to beat the five-to-seven-year interconnection queue.
Shared source of truth. Architects, engineers, utilities and equipment partners can work against one consistent model of the site — so reviews start from the same facts.
What the Twin Models
Specific systems, validated against physics — not a vague “live dashboard.” Toggle a layer to see what’s inside.
Toggle the layers we model
16 systems in the twinOne model, every system — aggregated on OpenUSD so architects and engineers in their own tools work against the same source of truth.
The Twin That Doesn’t Stop at the Chip
Every other data-centre twin models power in and cooling out, and ends at the chip. Greco’s goes one leg further — capturing the heat and exporting it to the city. Explore the full grid-to-chip-to-heat-export loop, then size the heat itself.
The loop the twin models
auto-playing · tap a stagePower-and-cooling twins stop at the chip. Greco’s scope runs one leg further — capturing the heat and exporting it. That full grid-to-chip-to-heat-export loop is the part no off-the-shelf twin puts at the centre.
Heat as a revenue stream
illustrative · modeled at design stageThe leg no other data-centre twin models: how much of that compute heat becomes a neighbourhood asset.
Liquid & immersion cooling recovers ~85% of IT load as usable heat at 50 °C.
Illustrative figures from a design-stage model (≈85% load factor, ~12 MWh of heat per home-equivalent, vs natural-gas heating). Actual export depends on the local district-energy network and offtake terms.
One Template, Every Site
We build the model once as a reusable, standards-based template, then adapt it per site rather than starting from scratch. Site dimensions, available power, cooling capacity and tenant requirements change; the engineering backbone is reused — so each new development is faster, more consistent, and easier to review.
Because the template is built on OpenUSD, models from architects and engineers in their own tools can be aggregated into one shared environment — the same open foundation NVIDIA’s Omniverse blueprints for AI-factory design are built on.
Designed to grow into live operations
Each twin is designed to ingest operational telemetry once a site is built — utility meters, switchgear, CDUs, pumps and rack conditions — so the same model that de-risked the design can become a living operations view. We scope this as a roadmap, commissioned per site as facilities come online.