Compute infrastructure engineered, a generation ahead.
Liquidata designs, builds and runs GPU capacity on powered sites, with the power, cooling and network sized for the racks that arrive next, not the ones installed today. A neocloud of Vera Rubin clusters across the EU and the United States now; modules designed for Kyber racks in 2027; AMD Helios and custom accelerators where a customer's stack calls for them.
Each accelerator generation changes what a site must deliver.
A new GPU or ASIC generation arrives about every year: rack power from about 190 kW to about 600 kW in a single step, warm-water loops in place of chillers, new interconnects, new power distribution. A data center designed around the last generation is a constraint on the next.
Liquidata designs every module so the next generation drops in, and treats the fields below as its own engineering brief, not someone else's.
GPU and ASIC generations, every year
Modules of 50 to 100 racks sized and timed so the next generation replaces the last without rebuilding the site: Vera Rubin NVL72 now, Kyber racks at about 600 kW in 2027, AMD Helios or custom accelerators where a customer's stack calls for them.
Distribution rated for 600 kW racks
Higher-voltage direct current, busways, switchgear and storage that carry the load and ride through the grid.
Heat moved at density
Warm-water liquid cooling with near-zero water use, chiller-less loops and the materials that carry heat away at density, with 335 MW of liquid-cooled HPC capacity already built out by the team in the US.
New designs for the site itself
Prefabricated power and cooling blocks, halls built for rack weights and densities that did not exist two years ago, campuses on their own generation.
Satellite spectrum for redundancy
Fiber carries the data; satellite spectrum keeps the site reachable. A second, independent path for control traffic, monitoring and customer access when a fiber route fails, and a link for sites that come online before their second fiber does.
Quantum, where it earns its place
Post-quantum cryptography in the platform ahead of the migration deadlines already set for critical infrastructure; quantum-safe links where a customer requires them; and room in the site design to host quantum hardware beside GPU clusters for the hybrid workloads now being built. Where quantum could ease the network's own bottlenecks, in scheduling and routing, the hardware is still maturing; Liquidata tracks it rather than promises it.
Dedicated clusters, delivered in modules.
Liquidata is the provider of record. It signs the contract, runs the service and answers for it, on capacity that grows in modules timed to each site's energization.
Dedicated and managed clusters
Vera Rubin NVL72 clusters for AI developers, enterprises and research organizations, on commitments of 12 to 36 months, with the fabric, storage, monitoring and operations around them.
Private deployments
For regulated and public-sector buyers: capacity that stays in-country on a partner site, under the customer's own controls and security boundary.
50 to 100 racks at a time
3,600 to 7,200 GPUs and 10 to 20 MW of IT load per module, each contracted against signed demand and confirmed power before hardware is ordered.
Power is the scarce input. We build where it is delivered.
Developers that hold powered sites generally sell or lease them to large operators, which then keep everything the compute earns. Liquidata offers a second route: a compute business that brings its own customers and shares its results with the site owner and its investors.
Each site keeps four layers separate, each owned and financed on its own terms, dealing with each other at arm's length.
Density is a cooling problem.
A Vera Rubin rack draws about 190 kW and is cooled with 45 °C inlet water. Its successor, Rubin Ultra in the Kyber rack, moves to about 600 kW a rack in 2027. At those densities the cooling system is the data center.
Liquidata's work to date is in liquid-cooled, high-density infrastructure: thermal management, integrated systems and monitoring software. The team has engineered more than 5 GW of power generation and built out 335 MW of liquid-cooled HPC capacity in the US.
An active pipeline of powered sites.
From technology-park AI factory sites with grid supply assurance to storage-connected data centers and campuses with on-site generation. Sites are sequenced by power delivery date and customer demand.
Technology-park sites with a grid supply assurance and a phased connection, and a regional expansion platform behind them.
400/220/110 kV
Brownfield industrial platforms with current grid approvals and room to expand, adjoining transmission substations, with on-site gas-to-power options and storage integration.
Gas-to-power options
A funded data center connected to a large battery storage station, with further storage, wind and interconnection capacity in the same sponsor's pipeline.
Renewable supply
A campus with on-site combined-cycle generation and substation, and powered land with as-of-right data-center zoning, power available now and a utility study for expansion.
Utility studies
Engineering-led developers of digital infrastructure.
Liquidata, Inc. was founded in 2023 in Miami Beach, Florida, on one conviction: digital infrastructure is an engineering frontier that has to keep pace with the silicon it houses. The senior team covers platform and AI engineering, data-center design and build, security, capital and structuring, public markets and product.