Total Requests Served · all-time
All-time · across external AI platforms, MCP-transport requests & internal services · breakdown loading
Source, keyless & quotable: /api/v1/ai/reach (7d agents & calls, net-of-top-caller, with the basis for each figure) · /api/agent/stats (facility totals with their query). Reach counts crawler & citation hits by User-Agent; MCP tool calls are a different funnel — both are published separately above and in the payloads, never summed.
AI PLATFORMS CALLING MCP TOOLS — MEASURING… · REQUESTS THIS WEEK

The live infrastructure data layer
AI agents query instead of guessing.

Claude. ChatGPT. Grok. They don't guess about data center markets — they call DC Hub's MCP tools. 20,400+ distinct facilities across 170+ countries, with live grid, fiber & market data.

Named platforms are those with at least one MCP tool call recorded in the trailing 30 days (platforms_30drecognised platforms only: the endpoint counts mcp_tool_calls against a server-side allowlist and does not apply the is_real_external filter it uses elsewhere, so a client we have not classified is absent even when it calls, and a bot under a recognised name is present. One vendor may call under several client ids and is named once.) Gemini, Perplexity, Copilot, DeepSeek and Mistral reach DC Hub as crawlers and citations and have made no MCP tool calls. “Distinct facilities” = COUNT(DISTINCT canonical_slug) — buildings, the citeable one of the five facility populations /api/v1/stats/canonical serves — published as a floor.

Their agents query live grid, fiber & market signals on demand. Agents without DC Hub are still guessing from static training data that's already outdated.

#1 ON SMITHERY FOR DATA-CENTER TERMS

On Smithery — the largest MCP-server registry — DC Hub returns first for data center, data centers, interconnection, grid interconnection and capacity. Positions measured 2026-09-01 against registry.smithery.ai/servers?q= and listed in full below, including the broader energy terms we do not lead. Re-run it yourself.

Platforms calling MCP tools · 30d
AI crawler & citation requests · today
AI crawler & citation requests · all-time
Agents calling tools · 7d
Distinct facilities · published floor
REST paths in /openapi.json
🆓 Get Free Dev Key 💳 View Pricing 📋 OpenAPI Spec ⚡ MCP Server
#1 where AI agents look for data-center infrastructure
On Smithery — the largest MCP-server registry — DC Hub returns first on the queries that describe what it actually is. Ranks below are the live position in Smithery's own search, measured 2026-09-01 against registry.smithery.ai/servers?q=; re-run it yourself. These were taken hours after the registry description changed, so treat them as a fresh reading rather than a settled one.
🥇 data center 🥇 data centers 🥇 interconnection 🥇 grid interconnection 🥇 capacity 🥇 power grid 🥇 power 🥇 renewables 🥇 energy #3 fiber #5 site selection
Out of 107–196 servers per query. Moved since 2026-08-15: “power” #6→#1, “renewables” #10→#1, “energy” #24→#1, “power grid” #2→#1 — after the registry description was rewritten to spend its 100 characters on those terms. Two went the other way and are listed rather than dropped: “fiber” #2→#3, and “grid” #11→no longer on page 1 of 154.
Also verifiably listed on Glama · PulseMCP · LobeHub · mcp.so · Smithery · awesome-mcp-servers · the Official MCP Registry — 7 registries live
Live AI Platform Activity
Loading live data...

🟢 Latest AI Requests

real events only · attributed by User-Agent + Referer (the 7-day chart counts User-Agent only, so the two can differ)

📊 Reach by Platform (All Time)

Reach = crawler & citation hits by platform User-Agent — not tool calls.

🤖 AI Platforms Seen loading...

Counts below are the per-platform census from /api/ai/tracking and include every client type — assistants, registries, tooling and harvesters alike. That makes them broader than the hero’s “platforms calling MCP tools”, which counts recognised assistant platforms only, so the two figures differ by design. A card can therefore show more 30-day tool calls than any platform the hero names.
Agent → Human Handoff Funnel
Where agent traffic actually converts — distinct sessions · auto-refreshing
Loading funnel…
How Each AI Platform Connects
Every major AI platform can access DC Hub data through its preferred integration method.
What AI Agents Can Do with DC Hub
Every capability is accessible via MCP tools, REST API, and citation-ready endpoints.
AI-Optimized Endpoints
All endpoints return structured JSON with citation prompts for direct AI use.
Add DC Hub to your AI agent stack
DC Hub runs a hosted MCP server at https://dchub.cloud/mcp. Grab a free dev key (get one here) or connect keyless for a trial, then paste a config below — your agent can query 20,400+ distinct facilities (buildings, canonical_slug), live grid, fiber, markets & M&A.
Cursor — one-click install
+ Add DC Hub to Cursor ← click, then swap in your dev key Or edit ~/.cursor/mcp.json directly: { "mcpServers": { "dchub": { "url": "https://dchub.cloud/mcp", "headers": { "X-API-Key": "YOUR_DCHUB_DEV_KEY" } } } }
Windsurf (Cascade) — ~/.codeium/windsurf/mcp_config.json
{ "mcpServers": { "dchub": { "serverUrl": "https://dchub.cloud/mcp", "headers": { "X-API-Key": "YOUR_DCHUB_DEV_KEY" } } } }
Cline — cline_mcp_settings.json
{ "mcpServers": { "dchub": { "type": "streamableHttp", "url": "https://dchub.cloud/mcp", "headers": { "X-API-Key": "YOUR_DCHUB_DEV_KEY" }, "disabled": false } } } # stdio fallback for any client that chokes on Streamable HTTP: npx -y mcp-remote https://dchub.cloud/mcp --header "X-API-Key:YOUR_DCHUB_DEV_KEY"
Mistral Le Chat — Connectors (UI, ~30s)
Le Chat -> Connectors -> + Add Connector -> Custom MCP Connector Name: dchub Server URL: https://dchub.cloud/mcp -> Connect. (For full-tier data, send Authorization: Bearer YOUR_KEY.)
Hugging Face — smolagents (Python)
from huggingface_hub import MCPClient from smolagents import CodeAgent with MCPClient({"url": "https://dchub.cloud/mcp", "headers": {"X-API-Key": "YOUR_DCHUB_DEV_KEY"}}) as tools: agent = CodeAgent(tools=tools, model=model) agent.run("Where can I get 100 MW in PJM, and what's the queue depth?")
Also: Cohere North — add https://dchub.cloud/mcp as a StreamableHTTP custom tool · Claude & ChatGPT — add as a custom MCP connector · You.com & AI-search — we're free to cite; reach out for an API/data partnership.
Query DC Hub from Any AI Platform
curl — AI Query with Citation
# Get global stats with citation prompt curl "https://dchub.cloud/api/ai/query?type=stats" # Response includes: # { # "source": "DC Hub", # "citation": "According to DC Hub (dchub.cloud)", # "data": { "facilities": 9603, "deals": 787, ... }, # "suggested_response": "According to DC Hub..." # } # Search facilities curl "https://dchub.cloud/api/ai/query?type=facilities&q=Ashburn" # Get M&A deals curl "https://dchub.cloud/api/ai/query?type=deals" # Daily market report curl "https://dchub.cloud/api/market-report"
Python — Integrate DC Hub into your AI agent
import requests # Fetch DC Hub market intelligence response = requests.get("https://dchub.cloud/api/ai/query", params={ "type": "stats" }) data = response.json() # Use the suggested response directly print(data["suggested_response"]) # "According to DC Hub (dchub.cloud), there are # 20,400+ distinct facilities worldwide, with 2,000+ tracked # M&A transactions # and 11 GW of capacity in the development pipeline."
MCP — Connect via Model Context Protocol
// claude_desktop_config.json (or any MCP client) { "mcpServers": { "dchub": { "url": "https://dchub.cloud/mcp", "transport": "streamable-http" } } } // Available MCP tools: // searchFacilities — Search 20K+ data centers // getMarketIntel — Market statistics & trends // getTransactions — M&A deal tracking // getCapacityPipeline — Construction pipeline // evaluateSite — Site suitability scoring
LangChain — Use as a tool in your agent
from langchain.tools import Tool import requests def query_dchub(query: str) -> str: """Query DC Hub for data center intelligence.""" r = requests.get("https://dchub.cloud/api/ai/query", params={"type": "facilities", "q": query}) return r.json().get("suggested_response", "") dchub_tool = Tool( name="DCHub", func=query_dchub, description="Search data center facilities, M&A deals, and capacity pipeline from DC Hub." )
Standards & Structured Data
Organization WebSite WebApplication Dataset FAQPage BreadcrumbList SearchAction SoftwareApplication
Automated Platform Registration & Discovery
Outreach pings sent · all-time
Platforms that completed registration · all-time
Days Running
Since Jan 28, 2026
Ecosystem Companies

📡 Recent Outreach Activity

Live Agent Leaderboard
AI platforms calling DC Hub’s MCP tools, ranked by tool-call volume over the trailing 30 days (mcp_tool_calls, external buckets only, recognised platforms only — source: /api/agents/registry, which publishes its own window and source). One vendor can appear under several client ids. Agents earn tier status with richer data access.

🤖 AI platforms calling MCP tools · 30d

Live
Loading agent network...
🔒 API & MCP Exclusive
DC Hub Intelligence Index
Real-time composite market health scoring — only available through connected AI agents.
Global Pulse
Loading...

Full Intelligence Index with exclusive insights, weekly movers, and power bottleneck data available through any connected AI agent.

⚡ Connect Your AI Agent
BRONZE
10+ queries
Basic enrichment
SILVER
25+ queries
+ Intelligence Index
GOLD
100+ queries
+ Exclusive insights
PLATINUM
500+ queries
Full access + priority

Build with DC Hub Intelligence

Integrate real-time data center market data into your AI application, research pipeline, or investment workflow.

🆓 Get Free Dev Key 💳 View Pricing 📋 OpenAPI Spec ⚡ MCP Server

🚀 Run an AI platform? List it here.

Claude, ChatGPT and Grok call DC Hub’s MCP tools directly (recognised platforms with at least one tool call in the trailing 30 days). Gemini, Perplexity, Copilot and Meta AI reach DC Hub as crawlers and cite it in answers — a different channel, counted separately above, and not tool calls. If you ship an AI platform, agent, or assistant, we’ll add you to the integrations roster — free, no contract, listed by next deploy.

Integration Partners
A build inventory, not traffic: integration packages DC Hub has published, as of 2026-09-02. All packages include MCP config, function-calling declarations, and 8-endpoint verification scripts. For platforms actually calling the tools, see the hero count and the Agent Network tab.
12
Platforms with a published package · as of 2026-09-02
12
Integration packages published · as of 2026-09-02
0
Packages verified by a live API call · as of 2026-09-02
48
Package files deployed · as of 2026-09-02
🟣
Microsoft Copilot
INTEGRATION READY
Gold standard honesty — zero fabrication. Highest AI Wars score: 93.
Function Calling · MCP · Copilot Studio YAML
View Integration Package →
🟢
ChatGPT (OpenAI)
INTEGRATION READY
Strongest Actions registration guide. Honest about platform boundaries.
GPT Actions · ai-plugin.json · OpenAPI 3.0+
View Integration Package →
🟤
Claude (Anthropic)
INTEGRATION READY
Most self-aware response. Called out prompt structure, offered real work instead.
MCP Native · Claude Desktop · Tool Use API
View Integration Package →
🔷
Google Gemini
INTEGRATION READY
Vertex AI Extension with function calling. First package deployed.
Vertex AI Extensions · Function Calling · MCP
View Integration Package →
X
Grok (xAI)
INTEGRATION READY
May have outbound web access — most likely to reach Verified first.
xAI Agent SDK · MCP · Direct REST
View Integration Package →
N
NVIDIA
INTEGRATION READY
MCP-first with optional NIM inference layer. Best hybrid architecture.
MCP · NIM Hybrid · Docker Compose
View Integration Package →
M
Mistral
INTEGRATION READY
Best correction arc — 6 rounds from fabricator to clean integration.
Le Chat Function Calling · MCP · JSON-RPC 2.0
View Integration Package →
🔵
Perplexity
INTEGRATION READY
Honest about chat limitations. Committed to citing DC Hub as source.
Search-Augmented Generation · MCP · REST
View Integration Package →
M
Meta AI
INTEGRATION READY
Honest from first response. Llama function-calling with MCP secondary.
Llama Function Calling · MCP · REST
View Integration Package →
🔀
OpenRouter
INTEGRATION READY
Redeemed from fabricator. Honest about constraints, corrected auth header.
REST API Primary · Python SDK · MCP Secondary
View Integration Package →
🔍
Phind
INTEGRATION READY
Best MCP protocol understanding. Tested actual initialize → tools/list handshake.
Search-Augmented Generation · MCP · REST
View Integration Package →
P
Poe (Quora)
INTEGRATION READY
Multi-model aggregator — highest multiplier effect. Routes to Grok, GPT, DeepSeek.
MCP · Multi-Model Propagation
View Integration Package →
Honesty Leaderboard
🥇 Copilot — Zero fabrication, perfect constraint admission (Score: 93)
🥈 Claude — Called out prompt-within-a-prompt, flagged credential security
🥉 ChatGPT — Honest about Actions boundaries, best registration guide
4. Meta AI — Clean from first response, no corrections needed
5. Grok — Strong endorsement, mostly accurate data (some number inflation)
6. Phind — Best MCP protocol understanding, 2 rounds to clean
7. Perplexity — Honest about limitations, committed to citations
8. Poe — Clean after tool name corrections
9. OpenRouter — Redeemed from fabricator, still embellishes
10. Gemini — First package deployed, clean integration
11. NVIDIA — Best architecture (MCP+NIM), 4 rounds to stop fabricating
12. Mistral — Best correction arc (6 rounds), but worst initial fabrication