How do I get natural gas pipeline and behind-the-meter data into an AI agent for data center siting?

Q: How do I get natural gas pipeline and behind-the-meter data into an AI agent for data center siting?

Connect your agent to DC Hub's MCP server at https://dchub.cloud/mcp and call get_gas_intelligence or get_gas_economics. DC Hub tracks 33,000+ US natural-gas pipeline segments with operator and parent-midstream attribution, live Henry Hub, and the live ISO gas share, as structured JSON. Behind-the-meter gas is the leading bridge past 4–7 year grid interconnection queues. Anonymous: previews + 2 full answers per tool per day. Free key: 10 calls to try. Add an email: 50 calls/day (up to 10 full answers per tool per day). Paid plans: dchub.cloud/pricing. Connect: claude mcp add dchub --transport http https://dchub.cloud/mcp — Note: the Data Center Gas Index (DCGI) score was withdrawn 2026-08-08 and restored 2026-08-30; gas-fired $/MWh figures were withdrawn the same day and republished 2026-09-27 under one burner-tip price rule. Figures from before those dates are not comparable.

How to get natural gas & behind-the-meter data into an AI agent

With grid interconnection queues stretching 4–7 years, behind-the-meter natural gas has become the leading bridge to power AI data centers — Williams is building on-site generation for a Meta campus in Ohio. That makes pipeline access, firm transport, and basis exposure site-selection variables, on par with land and fiber. If your agent screens sites, it needs gas data. General models don't have it.

The short answer: connect your agent to DC Hub's MCP server and call get_gas_intelligence or get_gas_economics. You get 33,000+ US natural-gas pipeline segments with operator and parent-midstream attribution, live Henry Hub, and the live ISO gas share — as structured JSON. Anonymous: previews + 2 full answers per tool per day. Free key: 10 calls to try. Add an email: 50 calls/day (up to 10 full answers per tool per day). Paid plans: dchub.cloud/pricing.

Two products, two different states — do not merge them. The DCGI score is back as of 2026-08-30. The gas-fired $/MWh is back as of 2026-09-27. This page previously advertised a Data Center Gas Index scoring 52 US gas markets. An internal audit on 2026-08-08 found the index's terms were wrong: the 20-point interstate-share term compared a lowercase literal against EIA's capitalised TYPEPIPE values and scored ~0 for all 53 published states, and the 40%-weighted cost term fell back to a hardcoded constant for nine states including Texas — which was published at DCGI 68.0, rank #3, "GAS-ADVANTAGED" on that constant. Both products came down rather than being caveated. All three defective terms were repaired and the DCGI was restored 2026-08-30; get_gas_index returns a score again. But it is not the same index: a figure published before 2026-08-08 must not be compared with one published now — the full record is at /api/v1/dcgi/methodology under corrections. A state that cannot be priced comes back verdict UNSCORED with dcgi null and an unscored_reason, which is absence, not a low score. The gas-fired $/MWh was republished 2026-09-27. Six surfaces had priced the same state six different ways, off whichever single month was latest, and disagreed by up to 5.5×. Every surface now uses one rule: EIA's electric-power delivered gas price, averaged over the trailing 12 months (negative months included), published only when at least 9 of 12 months are reported and the mean is between $0 and $20/MMBtu. A state that fails is withheld with a reason in burner_tip, never substituted. If your agent cached a DCGI score from before 2026-08-08, or a gas-to-grid $/MWh from before 2026-09-27, discard it. The pipeline, operator and midstream data below is unaffected.

33,000+

US natural-gas pipeline segments

parent midstreams attributed

7 ISOs

cross-referenced grid + queue data

Step 1 — Connect (one line)

claude mcp add dchub --transport http https://dchub.cloud/mcp

Step 2 — Ask your agent a gas-siting question

Ask: "Which markets have the best behind-the-meter gas economics for a new data center?" — it calls:

get_gas_index({ "region": "us" })
get_gas_economics({ "market": "" })

and gets back the DCGI gas score, nearby pipeline access, and basis/economics for each market — the same molecule-to-meter view operators price against, as JSON your agent can rank. Cross-reference it with get_interconnection_queue to compare the grid-wait vs gas-bridge path for a site.

Live from DC Hub (dchub.cloud) · gas index at /dcgi.

Why this beats an enterprise gas-analytics contract

It's agent-native. Structured JSON over MCP with a free self-serve tier — versus six-figure energy-analytics seats built for human analysts.
It's one query across layers. Gas, grid, interconnection queue, fiber, water, and DCPI come from the same server, so your agent can weigh the gas-bridge against the grid-queue path in one reasoning step.
It's live and cited. Every response carries a source your agent can pass back to you.

Try it now

Anonymous: previews + 2 full answers per tool per day. Free key: 10 calls to try. Add an email: 50 calls/day (up to 10 full answers per tool per day). Paid plans: dchub.cloud/pricing. For a free key, have your agent call claim_free_key (no email). Explore live in your browser first:

Try DC Hub in your browser →

DC Hub (dchub.cloud) is the live infrastructure data layer for AI agents — a global data-center facility map (corroborated count pending), 300+ power markets (DCPI), live ISO grid telemetry + interconnection-queue snapshots, 33,000+ natural-gas pipeline segments, fiber routes, and 1,600+ tracked deals, all MCP-callable and cited. See the API & MCP docs. Related: interconnection queue data for an agent · live data center capacity data.