The verified record for the planet's health — and what we cost it.
An independent measure of biodiversity, climate, water, and food — and of the energy, water, and carbon behind the technology now racing to understand them. AI and its datacenters are the front line today. Every number carries how it was made — its method, its sources, and how strongly it can be trusted.
New to the vocabulary? Hover any dotted term on this site for a plain-language explainer.
Naturecode is a research preview. Built on real public records — species and habitat data, grid carbon intensity, datacenter reporting, watershed sensors — with every measurement carrying the proof of how it was verified.
A global platform for measuring, verifying, and publishing the state of the planet — and what we cost it.
Naturecode is where the health of living systems is recorded — biodiversity, climate, water, and food — alongside the true footprint of the technology now racing to understand them. Measured at the source and kept honest for anyone who needs to act on it.
It opens with a working record on day one — species and habitat signals, grid carbon intensity, datacenter reporting, water draw. AI and its datacenters are the fastest-moving pressure on that system right now, so they're front and centre. You don't arrive to an empty page; you arrive to a live, auditable picture you can use immediately, and add to as you measure more.
Naturecode treats every number as something to be proven, not asserted. Each measurement carries a clear record of where it came from, how it was captured, and how strongly it can be trusted — so the footprint is never something you take on faith.
Four fronts. One foundation.
Biodiversity
Species, habitats, and the pressures on them — observed from space, sensors, and the people on the ground.
ClimateClimate
Carbon stocks, emissions, and climate stress — measured where they happen and held to account.
FoodFood Security
Crop conditions, hunger early-warning, and farm advisory — from satellite to smallholder.
WaterWater Security
Floods, drought, groundwater, and water quality — forecast early and monitored by communities.
Nobody can prove what the planet is actually paying.
The numbers come from the same operators and agencies being measured, with no way to check them.
From a single AI inference to a whole watershed, there's still no standard way to count what's really drawn.
Footprints and habitat loss are modelled from averages, not read from the ground where they happen.
Where a workload ran, on what grid, cooled by whose water — or which forest was cleared — almost never disclosed.
Meanwhile, the pressure is compounding:
- Datacenter electricity use is set to roughly double by 2030, driven by AI.
- Water is now a hard constraint — on new compute, on farms, and on the ecosystems both depend on.
- Biodiversity and climate signals are shifting faster than the reports meant to track them.
There is no trusted, independent record of any of it.
Every footprint measured — with the method shown and the sources named.
Most facilities don't publish their draw in real time. Where they do, we carry the operator's figure and link the disclosure. Where they don't, we estimate it from public inputs — capacity, cooling, local grid — and record every input so the number can be replayed. Either way, each figure shows exactly how far it can be trusted.
estimated or operator-reported
the method + inputs recorded
the number can be re-derived
trust tier on every number
A network already exists.
Trees, fungi, and the wider community of life trade nutrients, water, and signals through underground networks. No single tree is the source. The forest itself is the system.
A grid's carbon reading dates a datacenter's energy figure. A water-draw sensor grounds a cooling estimate. A method for measuring one inference is adopted and sharpened by the next team — and every reading strengthens the record the others rely on.
When a footprint moves — a workload shifting to a dirtier grid, a site drawing more water, a model's inference cost climbing — the signal reaches researchers, regulators, and operators in the moment it matters, backed by proof rather than a press release.
Not a dashboard. A shared, verifiable record of what technology costs the planet.
Distinct measurements above ground. Mutual reinforcement below. Proof travelling with every number. Signals moving at the speed of the grid, not the speed of a sustainability report. The pattern that keeps forests honest, applied to the machines drawing on them.
Five layers. One living system.
Open, continuously updated measurements of technology's footprint — grid carbon intensity, datacenter energy and water, AI compute impact — ready to read from day one and growing as more sources are measured.
- — Impact measurements and live streams
- — Day-one, real-world baseline
- — Open, restricted, or paid access
Operators, researchers, and monitors contribute the readings closest to them — a datacenter's own meters, a grid feed, an independent sensor — and each contribution is attributed to its source.
- — Operators and independent monitors
- — Clear attribution
- — Rules for access and use
Energy meters, water draw, grid carbon feeds, cooling telemetry, per-inference compute figures, and the retained nature signals — different sources, different speeds, all flowing into the same record, with the source clearly traced.
- — Energy, water, carbon, and compute
- — One-time reports or live streams
- — Source and history always visible
A shared shelf of agents and models anyone can run against the record — to validate a reported figure, detect a footprint shifting, enrich a reading with grid or weather context, and flag anomalies. Contributors can publish their own methods.
- — Validation, detection, and enrichment
- — Use what's on the shelf, or publish your own
- — Every result traceable to its source
Every figure carries where it came from — the method, the inputs, and the sources behind it — and a trust tier that says how strongly it can be relied on. As live meters and signed feeds come online, readings gain a source you can trace all the way back.
- — The method and sources behind every figure
- — A trust tier on every number
- — Signed, traceable readings as they come online
A deep tour of each layer — including how figures are gathered or estimated, and how trust tiers are assigned — lives on the dedicated pages.
Full system overviewOne protocol. Different doors in.
See what's being built next to you.
Neighbourhoods, water districts, and local organisers living beside datacenters and the grids and reservoirs that feed them — the people who carry the cost first.
- See the energy and water a nearby site actually draws, not what a brochure claims
- Track how much of it comes from the local grid and watershed
- Get alerted when a site's draw climbs or shifts
- Every figure carries the proof of how it was verified
Govern AI's footprint with numbers you can trust.
Local and national governments, energy regulators, and public bodies setting rules for datacenters, grids, and AI infrastructure at the territorial scale.
- An independent record you didn't have to take from the operators being regulated
- Energy, water, and grid-carbon figures with the trust tier shown on each
- Publish open observatory dashboards constituents can audit
- Ground siting, permitting, and disclosure rules in measured reality
Query the footprint. Cite the method.
Universities, energy and climate labs, and independent monitors — anyone who needs open, well-sourced data on what AI and infrastructure cost.
- Open figures across energy, water, grid carbon, and AI compute
- Cite individual figures with their method and sources; cohort exports for replication
- Save queries; receive alerts when thresholds are crossed
- Retained nature signals — biodiversity, climate, water, food — on the same record
Cut your footprint on an independent baseline.
Datacenter and cloud operators, AI teams, and sustainability leads who don't just want to report a number — they want to bring it down, and show their working.
- Start from an open, independent baseline instead of a self-graded one
- Reduction and optimization tools live on naturecode.ai — the commercial platform
- Shift workloads to cleaner grids and cooler hours, then verify the saving on this record
- Publish before-and-after with proof, not just a pledge
Bring the verified footprint into your system.
Engineering teams, climate-tech startups, and reporting-software vendors integrating trustworthy impact data into a product.
- Read measurements, submit readings, subscribe to streams — one SDK
- Invoke agents for validation, detection, and enrichment
- Plug into any LLM client; agents can talk to other agents on your behalf
- Every response carries its source and trust tier
Contribute your readings, show your working.
Datacenter, grid, and cloud operators who want to report a footprint that stands up — from their own meters rather than modelled from averages.
- Submit readings straight from your meters and telemetry
- Each figure is graded so its trust tier is visible to everyone
- Move up the tiers as you add independent, instrumented sources
- A record you can point auditors and regulators to
Plug your devices into a trusted record.
Meter, sensor, and telemetry vendors — anyone running instrumentation on grids, water systems, and datacenter floors.
- Stream readings straight from the source — attributed back to you
- Naturecode handles routing, identity, and attribution back to your customers
- First-class integrations for grid feeds, water sensors, and facility telemetry
- Get listed so operators can install with one click
See what the machines are drawing.
Water utilities, watershed authorities, and grid operators who need to know what new compute demand is actually pulling from their systems.
- Track datacenter water draw and grid load against your own capacity
- Cross-check operator figures with independent readings on the same record
- Flag sites whose draw outpaces what was permitted
- Retained water and climate signals sit alongside the infrastructure data
Talk to us — we'll route you.
Onboarding is hands-on. We help map your existing setup — meters, feeds, reporting — onto Naturecode, and point you to whichever side fits: the public record here, or reduction tools on naturecode.ai.
- 30-min discovery call with the Naturecode team
- Mapping of your data sources and measurement needs
- Recommended starting point — read the record, contribute, or reduce
- A path from first reading to a fully verified footprint
Every number carries its own proof.
Shown in the open, traceable to its sources.
The same record-keeping backs every footprint figure and every agent result the network produces.
Where an operator discloses a figure, we carry it and link the disclosure. Where none exists, we estimate it from public inputs and record every input, so anyone can re-derive the number by hand. Estimates are cross-referenced against independent sources — grid carbon feeds, water-stress data, satellite passes, and the emerging standard for measuring AI compute. Open method, named sources — never a black box.
The record updates as new figures arrive, not once a year in a report you can't check — and every figure keeps its lineage: which method, which inputs, which sources, laid out for anyone to follow.
The same openness covers the measurement agents. When an agent validates or enriches a figure, it shows which data and which model it used and how the result was produced. Nothing about how a number was reached has to be taken on trust.
- AI computePer-run energy, water, carbon
Aligns with the emerging standard for measuring the footprint of a training run or inference — the calculation exists; the trusted record doesn't.
- DatacenterPower, water, cooling efficiency
Power-use and water-use effectiveness, cooling method, and grid draw — from operator disclosures where they exist, or estimated per facility with the inputs shown.
- GridReal-time carbon intensity
Where and when a workload ran, against the live carbon intensity of the grid feeding it — the difference between a clean hour and a dirty one.
- NatureBiodiversity, climate, water, food
The retained nature signals, on the same verifiable record — one substrate for the impact of technology and the systems it draws on.
Every number shows how far to trust it.
Where an operator publishes nothing, we estimate the figure from capacity, cooling method, and local grid — using a documented method — and record every input so the number can be replayed by hand. Honest about being a model, not a reading.
- Documented method
- Inputs recorded
- Replayable
A figure the operator publishes in a sustainability report or filing. We carry it and link the disclosure. More specific than an estimate, but still self-reported — it's their number, attributable to them, on the record.
- Operator-disclosed
- Source linked
- Attributable
The goal tier: read from a live meter or feed and carried with its source, so the reading can be traced to where it came from. The strongest evidence — and the one most facilities don't yet reach.
- Direct reading
- Source traced
- Highest trust
Three surfaces. Same protocol underneath.
The Naturecode Platform.
The hosted dashboard where contributors, researchers, and developers read the record, submit measurements, and track footprints over time.
- Mode-aware overview (contributor, researcher, developer)
- Measurements, streams, models, agents — full detail with trust tier
- Footprint trends by site, grid, and workload
- Activity log, workflows, alerts inbox
- Operators and monitors submitting readings
- Policymakers and communities tracking a site
- Researchers querying and citing measurements
Naturecode Apps.
Purpose-built apps for specific roles — designed for the device and context in which the measuring actually happens.
- Field Monitor — mobile reading submission, offline-first, photo + sensor capture
- Operator Console — desktop view: site meters, agent control, alert routing
- Grid & Water View — live carbon intensity and draw against local capacity
- Sensor Manager — device pairing, on-device model deploys, network health
- Public Observatory — read-only embeddable dashboards for communities and journalists
- Verifier — independent-monitor app for corroboration rounds
The Developer Platform.
One key, one platform. Bring the verified footprint record and its measurement agents into what you're already building — or white-label the whole surface.
- Node and Python SDKs
- REST and webhooks for every event — measurements, agent actions, alerts
- OpenAPI viewer and managed keys
- Plug Naturecode into any LLM client
- Agents can talk to other agents on your behalf
- Pay-as-you-call billing for agent and API usage
- Drop-in widgets for dashboards — measurements, trends, trust tier
- White-label under your brand for partner programmes
- Shared catalog of sensors, models, and measurement methods
Most users arrive with meters, feeds, dashboards, and reporting workflows already in place. Naturecode integrates as an open measurement layer underneath — your existing UIs keep working, the data underneath becomes sourced, attributable, and trust-tiered. White-label the dashboard for your programme, embed widgets in your own product, or let your AI agents call Naturecode tools through MCP.
- MCP · A2A
- Grid carbon feeds
- Water telemetry
- REST · GraphQL · gRPC
- Webhooks
- Embeddable widgets
- Facility telemetry
- Agent invocations
Not a dashboard. Not a report. Not a rating.
Naturecode is verification infrastructure — the trusted record of the planet's health and the pressure technology puts on it.
A window that opens once.
Datacenter electricity is set to roughly double by 2030 — the fastest-moving new pressure on energy and water.
Compute, agriculture, and ecosystems are now competing for the same stressed watersheds.
Satellite, sensor, and bioacoustic data can track habitats and climate at planetary scale.
The open data and methods to hold these figures to account exist now — the trusted record that puts them in one place doesn't yet.
The trusted record for what technology costs the planet.
The planet is already paying — AI just made it visible.
Naturecode measures it, proves it, and makes it something you can act on.