Question 1

What is AIDC-AI.IO?

Accepted Answer

AIDC-AI.IO is an AI data center automation tool, or AI 데이터센터 자동화 툴, that generates and validates data center building, rack, electrical, cooling, network, PUE/WUE, cost, commissioning, and operations evidence from one EngineSession-backed 3D node-edge graph.

Question 2

Which NVIDIA GPU generations does the AIDC-AI.IO engine support?

Accepted Answer

The active physical rack-planning basis is NVIDIA Vera Rubin NVL72 / VR200 with selectable workload ratio profiles and explicit source/RFI status. Hopper and Blackwell-era catalog entries remain available as separate reference records; larger scale-domain labels are treated as planning metadata unless source-backed for the selected design.

Question 3

Is the engine usable outside Korea, or only on the 22.9 kV grid?

Accepted Answer

The engine is region-configurable. Korea is the launch deployment market, and other markets require local utility, code, AHJ, climate, water, seismic, security, and operations inputs. Future 800 VDC architectures are tracked as source-backed reference evidence where applicable, but rack-level design is validated against the selected site and utility basis rather than assumed identical worldwide.

Question 4

Can an AI agent call the design engine directly?

Accepted Answer

Yes. Site Finder remains the save gate for real-parcel projects, and 3D Rack Viewer is the primary human visual session. AI agents can call /api/engine/design, /api/engine/validate, and /api/engine/layout for graph/API projections; /api/agent/* wrappers include citation metadata. OpenAPI 3.1 spec at /api/openapi.json, MCP server card at /.well-known/mcp/server.json, plugin manifest at /.well-known/ai-plugin.json, full LLM context at /llms-full.txt. MCP-capable clients can connect with: npx -y aidc-mcp-server.

Question 5

What design PUE can a fully liquid-cooled (DLC) configuration achieve?

Accepted Answer

PUE is a design and operating metric derived from the selected IT load, electrical topology, cooling topology, climate, water strategy, and operations basis. The engine keeps PUE and WUE as explicit OPR/BOD values and validation outputs rather than treating a single vendor rack claim as universally applicable.

Question 6

Which engineering tools and standards does AIDC-AI.IO use?

Accepted Answer

Planning PowerFlow, SLD, and Sankey are projections of the same EngineSession. Configured external runs can attach pandapower.runpp 3.5.4, Modelica/FMU, and OpenFOAM-compatible result evidence. NetBox-compatible rack metadata supports planned DCIM handoff, not live synchronization. ASHRAE and OCP materials are reference inputs and do not imply certification, endorsement, or affiliation.

Question 7

What is 800 VDC in-rack distribution and why does it matter for worldwide deployment?

Accepted Answer

800 VDC is tracked as an emerging high-density power architecture and source-register item. The engine may use it as a candidate power architecture when selected and supported by evidence, while upstream MV, transformer, UPS, protection, grounding, AHJ, and commissioning requirements remain project-specific validation work.

설계 기준을 보내주시면 모델 기반으로 검토합니다

설계 기준을 보내주시면 모델 기반으로 검토합니다