Hardware-Verified · Cardano · Hex Node Network

Reality is measured. Not estimated. Hardware-signed truth.

Mālama Labs is the hardware-signed trust anchor for physical-world data. Every environmental market, every emissions disclosure, every ESG claim, every parametric insurance payout, and every supply-chain attestation depends on one thing: knowing what is actually happening in the physical world.

For thirty years, that knowledge came from estimates, models, self-reports, and annual audits. We replaced estimation with cryptographically signed sensors. Each reading is signed at the device by a key burned into hardware at manufacture, anchored on-chain, and verifiable by anyone. Carbon proved the pipeline works. AI compute scales it. Water, supply chain, biodiversity, and grid follow.

Join our waitlist → Schedule a Call →
One Protocol · One Hex Network | On-Chain Proof
07 · AI Energy Data
30 models tracked live
Estimation tier today · hardware tier next
aipower.fyi
Video Gen Cost
944Wh
GPT-o3 Per Prompt
39.2Wh
Water Per Video
1.0L
Efficiency Gap
1.88M×

The AI Energy dashboard at aipower.fyi is an open-source project of the Mālama Foundation. Explore there, or visit MalamaFoundation.org for our non-profit work.

Estimation tier Sensors next
06 · Live Sensor Data · Dallas Pilot Node #1 Pilot stream
Soil Temperature
°C
Air Temperature
°C
Humidity
%
Battery
%
Hardware signed · ECDSA
Orientation · 60 seconds →
What is Mālama Labs?

What is Mālama Labs?

Carbon markets lost buyer trust because credits are verified once a year by humans walking transects. AI data-center emissions are reported via spreadsheets and renewable energy certificates. Parametric insurance triggers on weather stations hundreds of miles away. Supply-chain provenance still lives on PDFs.

Every one of these markets has the same missing input: a trusted, continuously monitored, cryptographically signed physical data source. That is what Mālama builds. Not a methodology. Not a registry. The hardware-signed measurement layer downstream markets have been trying to work around.

See how the pipeline works →
01 · One Pipeline. Any Physical Measurement.

Carbon proved it. AI compute scales it.

The Dallas pilot has emitted 2,786 signed SaveCards from a soil sensor since June 2024. The same signing architecture, same Cardano anchor, and same Hex Node validators now extend to rack-level power sensing inside AI data centers. The same architecture generalizes to any sensor that can be housed in a device with a secure enclave.

PROVEN · LIVE SINCE JUNE 2024

Carbon dMRV

Genesis 300 Pilot Node #1 in Dallas has operated continuously since June 2024 on Cardano preprod. Full credit lineage (sensor → oracle → event → credit) is operational with 2,786+ records and zero gaps. Mainnet migration targets Q2 2026 post-audit.

  • · Soil, atmospheric, and enhanced rock weathering telemetry
  • · LCO2 pre-finance and VCO2 verified credits
  • · Registry engagement underway with Puro.earth and Isometric
  • · Same signing architecture used across all verticals
Carbon dMRV detail →
SCALING · Q2 2026 PILOTS

AI Compute Monitoring

Rack-level power sensors measure exact electrical load per inference and training cycle. Same hardware-signed pipeline. Same Cardano anchor. Same Hex Node validators. Verified scope 1 data the data center cannot omit.

  • · Direct power sensing per inference and training cycle
  • · Water cooling attribution and evaporation tracking
  • · Real-time grid carbon intensity sync
  • · Built for SEC, EU CSRD, and SBTi-aligned reporting
AI compute detail →
DESIGNED · IN ACTIVE ARCHITECTURE

Adjacent Physical Markets

Parametric insurance triggers, EUDR supply-chain provenance, grid and microgrid intelligence, water rights, and permit compliance. Each uses the same signing pipeline. None require a new core trust layer.

Parametric insurance·Supply chain provenance·Grid intelligence·Water compliance
Carbon dMRV detail → AI Compute detail → Platform roadmap →
02 · The Stack

Four layers visible. Six layers in the protocol.

The four-layer summary below collapses Mālama's full six-layer architecture (Physical, Verification, Oracle, Blockchain, Reward Scaling, Redemption) into what teams need day-to-day. Carbon SaveCards, AI compute packets, and future sensor types all flow through the same trust pipeline. Full architecture: technology page →

01
Hardware
The device
Two hardware classes, one signing architecture. Genesis 300 outdoor nodes for environmental sites. Rack-mount AI Power Sensors for data center deployment. Both carry ATECC608B secure enclaves with private keys burned in at manufacture.
02
Trust
The signature
ECDSA/SHA-256 device signing. Every reading, whether soil temperature or rack-level wattage, is signed by the device private key before it leaves hardware. Tamper-evidence is cryptographic, not procedural.
03
Blockchain
The anchor
Dual-chain by design. Cardano anchors SaveCards via CIP-25/CIP-68 datum-metadata for archival custody and registry-grade auditability. Base handles execution, rewards, claims, and EVM market activity. LayerZero OApp coordinates state across both. Cardano mainnet migration post-audit Q2 2026.
04
Hex Nodes
The validators
Decentralized validator network running Proof-of-Truth consensus. Each Hex Node operates within an H3 geographic cell governed by an NFT-HEX rights object. Nodes validate both carbon SaveCards and AI compute packets. Phase 1 Hex Node launchpad coming soon.
03 · Who It's For

Three audiences today. The list grows as the sensor catalog grows.

01 / CARBON PROJECT DEVELOPERS

Deploy Genesis 300 Nodes

Deploy a Genesis 300 node and move from intermittent audit assumptions to continuous hardware-signed verification. Sub-acre deployments become viable when field telemetry is continuously signed and queryable.

Cost framing: Traditional verification often carries $20,000–$50,000+ per-project audit overhead. Mālama targets single-digit dollar MRV cost per tCO2e at scale, depending on project size and sensor density.

02 / AI OPERATORS

AI Compute & Data Center Teams

Hardware-verified scope 1 data ready for SEC climate disclosure, EU CSRD, and SBTi reporting. Scope 3 support for businesses consuming AI compute services. Stop estimating at the meter; start measuring at the rack.

Scope 1: Direct measurement of electrical load per inference.
Scope 3 support: Hardware data supports upstream emissions calculations for AI buyers.

03 / NODE OPERATORS

Hex Node Validators

Run a Hex Node in an H3 cell and validate carbon SaveCards, AI compute packets, and future physical-world data types. Earn MLMA verification yield, uptime multipliers, and governance weight in the Mālama DAO.

04 · Live Proof

Carbon dMRV is the production reference. AI compute is next.

Sensor stream live today on Cardano preprod. Mainnet SaveCard minting is paused pending audit completion and migration (target Q2 2026). Recent transactions below are the latest involving the Mālama preprod wallet from our indexers.

SaveCards Minted
2,786
Hardware-signed · Yes
Field Events
3
Biochar · Soil Test
Genesis Date
JUN '24
Continuous since June 2024
Latest Block
4,561,635
Updated live

Recent Transactions · Preprod

● Auto-refresh 30s

Sensor stream live today on preprod. Cardano mainnet SaveCard minting paused pending audit completion and mainnet migration (target Q2 2026). Recent transactions below are the latest involving the Mālama preprod wallet from our indexers. Each row opens in Cardanoscan. If the feed is empty or indexers are slow, we show the last list loaded in your browser as a fallback.

Loading recent transactions…
05 · The Measurement Crisis (AI Compute)

Why the next vertical after carbon is AI.

The AI industry accelerates while its environmental footprint remains obscured by voluntary disclosure and market-based accounting. A Federation of American Scientists analysis indicates Meta's actual emissions may be materially higher than market-based reports suggest. Software reporting cannot close the gap between what is disclosed and what is real. Hardware-signed direct measurement can.

DIRECT POWER SENSING

Per-Inference Wattage

Rack-level sensors measure exact electrical load per inference and training cycle. Not market-based estimates. Not vendor self-reports. Direct measurement at the source.

WATER ATTRIBUTION

Cooling and Evaporation

Hardware-linked tracking of cooling energy and evaporation rates. A single video generation prompt can consume up to 1 liter of water for cooling alone.

GRID INTENSITY SYNC

Real Carbon, Not Average

Cross-references real-time grid carbon intensity with sensor-verified energy use. The same kilowatt-hour at different times and locations carries radically different emissions weight.

AI Compute Product → Live Dashboard ↗

Phase 1 Hex Node launchpad

Own the infrastructure of truth.

Each Hex Node validates carbon SaveCards and AI compute packets across the Mālama network. Get updates—learn first when we open in the coming days.

Join our waitlist ↗
Schedule a Call