Inference.llc – AI at the Edge
Datacenters for the age of AI
‘Inference.llc’ is a site development project creating a distributed Edge AI datacenter platform purpose-built for the coming inference-dominated era of artificial intelligence. The platform focuses on delivering and selling NTP sites for 5–20 MW, with substation-adjacent datacenter facilities designed for “speed-to-power”, ultra-low latency, and capital efficiency. This initiative represents a ~$2B CAPEX development pipeline targeting the 2027 AI inference inflection point.
The Problem
AI workloads are rapidly shifting from centralized AI model training to distributed inference. Hyperscale data centers are constrained by lack of power, long interconnection queues, remote locations, and multi-year deployment timelines. Inference workloads require low latency, proximity to users, and immediate power availability—conditions that hyperscale infrastructure cannot meet.
50–93 GW
Global AI Inference Demand 2030
Market Opportunity
By 2030, global AI inference demand is projected to reach 50–93 GW, growing at ~35% CAGR. Approximately 10% of this demand (~5-10 GW) is expected to be served at the edge.
Platform Strategy - Solution
Inference.llc designs datacenters to deploy energy-aware infrastructure that integrates with underutilized grid nodes, on-site battery storage, fuel cells, renewables, and liquid-cooled facilities to avoid water waste.
Projects are delivered via retrofit of current qualified buildings (12–24 months) or greenfield (24 months) strategies, avoiding hyperscale interconnection queues.
Technical Profile
- Site Size: 5–20 MW
- Rack Density: 15–40 kW
- Cooling: Liquid-cooled, water-efficient
- Latency: <10 ms RTT SLA-backed
- Redundancy: Tier III, 2N
- Hardware: Optimized GPU and ASIC mix
Leadership
The leadership team bridges energy infrastructure and IT execution, with backgrounds spanning hyperscale data centers, grid-scale storage, renewables, and enterprise IT services. The team has worked together for the last 10 years to develop 1GW+ energy projects in the USA.
Thanasis Iatrou
CEO
CEO
15 years experience energy project site development , 15 years in datacenters/IT; developed 1GW+ greenfield Solar/Storage projects at Caprock Renewables LLC where he served as CEO/Founder and sold them to Engie, LightSource/BP, Duke Energy, etc. Prior IT experience includes 7 years at DELL, as Head of Datacenter Services worldwide and VP of M&A at Fujitsu Computers in London.
Kimo Storke
VP Business Development
VP Business Development
$1bn+ project finance experience, raised $10m for Caprock
George Kolovos
VP of Finance
VP of Finance
12 years Project Finance at Caprock Renewables and Terna Energy
Ryno Irwin
COO
COO
entrepreneur, built second largest Microgrid company and sold it, with ElectriFi support
Antonio Giustino
VP Project Development
VP Project Development
project development executive, PE background, worked for Wood Group consulting on project development of Caprock projects
Bill Webber
VP Land
VP Land
experienced land development executive, worked for Thanasis at Caprock
Kevin McSpadden
Legal counsel
Legal counsel
experienced energy lawyer 30 years
The management team is surrounded by an Advisory Board of Directors with domain expertise in Datacenter management, hardware, operations, fiber/telecoms, utilities, economic development and academia.
Expert Development Consultants : (8-10 year prior work and collaboration experience)
EPE
Electrical engineers (experts in interconnection)
Wood
project development consultants (land, zoning, environmental, nat gas, design)
McGuire Woods
Transactional lawyers (No 1 in energy M&A)
DNV-GL
Independent Engineers
E&Y
Tax Abatement consultants
Energy partners
Bloom Energy
Fuel cells for datacenters
EticaAG
Liquid immersion, fire-safe batteries for datacenters
Energy partners
Inference.llc seeks strategic partners across utilities, OEMs, hyperscalers, chip providers, infrastructure funds, and energy technology providers. Partners benefit from early access to deployable capacity, co-development opportunities, and participation in a scalable edge AI platform.