# GPUQuant AI Infrastructure

Wires an agent into GPUQuant's records on data centres and companies: search sites, read facility facts, trace company links. Read-only.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI
- Author: gpuquant
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/gpuquant/mcp
- Updated: 2026-09-13
- HTML: https://affpapa.org/en/nevibecode/mcp/gpuquant-ai-infrastructure

## What it does

Three tools come with the server. `search_infrastructure` looks up records by name or location and pages through results; `get_facility` returns a site's public facts alongside its sources; `get_company` covers a company and the relationships documented for it. Pull the exact input schemas with `tools/list`.

The endpoint sits at `https://gpuquant.com/api/mcp` and speaks Streamable HTTP. Nothing to authenticate — this service is open to anyone. Add that URL in your client's remote-server settings; each client formats its config differently. Both the current MCP protocol and older clients using the 2025 initialization handshake are accepted.

What lives in the repository is connection docs, a registry manifest and an HTTP smoke test. The database itself stays on GPUQuant's side, so cloning gets you no local copy. Running the smoke test needs Node.js 20 or newer; it pulls in no third-party packages and asks for no credentials. This first release covers infrastructure records and company relationships — licensed Ornn and Silicon Data feeds, trading accounts, orders, imagery and building geometry are outside it. Dates attached to answers describe the record or its source, not a fresh check of the fact, and anything missing stays unknown.

## Requirements
- Remote server: `https://gpuquant.com/api/mcp`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http infrastructure https://gpuquant.com/api/mcp
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add gpuquant-ai-infrastructure --url https://gpuquant.com/api/mcp
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "gpuquant-ai-infrastructure": {
            "url": "https://gpuquant.com/api/mcp"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "gpuquant-ai-infrastructure": {
            "type": "http",
            "url": "https://gpuquant.com/api/mcp"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http gpuquant-ai-infrastructure https://gpuquant.com/api/mcp
```

## When to use it
- find infrastructure records for a city and pull their sources
- check who operates a facility, its status and when it was verified
- separate documented supply-chain links from merely reported ones
- flag stale or unknown facts before writing a company summary
- gather data-centre context for a research note

## Example prompts
- Search GPUQuant for infrastructure records near Copenhagen and list the sources for each.
- Show this company's supply-chain relationships and mark which ones are only reported.

## Similar
- [extension-core-infrastructure](https://affpapa.org/en/nevibecode/skills/extension-core-infrastructure.md) — skill · 27900 installs
- [Azure Terraform Infrastructure Planning](https://affpapa.org/en/nevibecode/agents/azure-terraform-infrastructure-planning.md) — subagent · ★ 40451
- [cloud-infrastructure](https://affpapa.org/en/nevibecode/plugins/cloud-infrastructure.md) — plugin · ★ 41823
- [terraform-data-engineering-infrastructure](https://affpapa.org/en/nevibecode/skills/terraform-data-engineering-infrastructure.md) — skill · 1800 installs
- [infrastructure-cicd-data-engineering](https://affpapa.org/en/nevibecode/skills/infrastructure-cicd-data-engineering.md) — skill · 1600 installs
- [logfire-infrastructure](https://affpapa.org/en/nevibecode/skills/logfire-infrastructure.md) — skill · 532 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md