One protocol. Three primitives. Two chains.
The Mālama platform is a hardware-verified evidence network. SaveCards instantiate the data layer. NFT-HEX instantiates geographic rights and reward logic. veMLMA instantiates governance and redemption. Cardano provides archival custody. Base hosts the MLMA token, liquidity, and execution. The chains operate independently; the protocol does not require bridging them.
Carbon proved it. AI compute scales it.
The Mālama platform serves two upstream data streams through one verification pipeline. Carbon dMRV is the proven product, operational on Cardano preprod since June 2024 with 2,786+ on-chain SaveCards. AI compute monitoring is the scaling product, bringing rack-level power sensing to data centers in 2026. Same hardware-signing architecture. Same Hex Node validators. Same on-chain anchor.
Carbon dMRV
Hardware-signed environmental telemetry from biochar and enhanced rock weathering sites. LCO₂ pre-finance instrument issued against verified live data streams. VCO₂ verified credit upon registry approval. Issuance and conversion timelines are protocol targets, not current realized cycles.
dMRV Engine detail ↓AI Compute Monitoring
Rack-level power sensors inside AI compute facilities. Direct wattage per workload. Cooling water attribution. Real-time grid carbon intensity. Built for SEC climate disclosure, EU CSRD, and SBTi-aligned reporting.
Continuous verification, not periodic audit.
A unified MRV pipeline that ingests hardware-signed sensor streams, runs anti-spoofing and anomaly detection, batches by region and methodology into Merkle trees, and emits registry-compatible verification records hourly. MVP methodologies: enhanced rock weathering and biochar. Roadmap: forestry, soil carbon, and regenerative agriculture on the same engine.
Hardware-Signed Streams
Every sensor reading arrives cryptographically signed at the device identity. Signatures are validated against the on-chain device registry before ingestion. Anything unsigned or invalid is rejected at the gate.
Layered Validator Stack
Cryptographic, protocol, physical, spatial, temporal, and methodology validators each contribute to a quality score. Statistical anomaly detection flags spoofed and physically implausible readings. Validators participate under a PONO credential and face credential revocation plus forfeiture of unvested rewards for falsification. The penalty structure makes dishonest reporting economically irrational once milestone vesting is factored in.
Merkle Roots On-Chain
Validated readings are batched by region and methodology, bundled into Merkle trees, and persisted in immutable off-chain storage. Only the Merkle root is anchored on Cardano. The on-chain footprint stays constant regardless of telemetry volume; auditors retrieve inclusion proofs for any individual record from the storage layer.
Registry-Compatible Output
Verra VCS, Gold Standard, Puro.earth, Isometric, ACR, and Article 6.4 output formats supported. Continuous monitoring records map directly into the methodology fields each registry requires.
Three objects. Six layers.
Each primitive bundles the outputs of multiple architectural layers into a unit operators, validators, and institutional counterparties can transact with directly.
The Evidence Object
Hardware signing, validator review, and oracle aggregation collapsed into a portable cryptographic record. Each SaveCard points to hardware-signed data with retrievable Merkle inclusion proofs. The standard evidence primitive, anchored on Cardano via CIP-68.
The Geographic Right
An H3 Resolution 5 cell rights and policy object. Encodes non-exclusive geographic operating rights, capacity constraints, reward weighting, and acquisition policy. Not a collectible. The unit through which capital deploys infrastructure into specific climate-relevant regions.
The Governance Lock
Vote-escrowed MLMA staking on Base. Aligns long-term protocol commitment with governance weight. veMLMA combined with a PONO credential drives operator claims, treasury distribution, methodology approval, validator parameter changes, and revenue distribution through tiered governance thresholds.
Cardano for archival. Base for execution.
A climate data network needs two different blockchain property profiles. Cardano gives deterministic finality, predictable fees, and registry-grade auditability for long-horizon archival custody of SaveCards. Base gives EVM-native liquidity, mature audit ecosystem, and broader institutional access for the MLMA token, staking, and governance. The chains are not bridged at TGE. They do separate jobs, and the token does not need to move between them for the protocol to function. A Cardano-side MLMA mirror is a Phase 2 question contingent on bridge selection, security audit, and market readiness.
Cardano
CIP-68 SaveCard issuance. Datum-metadata for richer state semantics. Deterministic Plutus contracts for registry-grade auditability. 60-second Merkle anchoring. Long-horizon integrity at predictable cost. Mainnet migration Q2 2026, post-audit.
Base
EVM L2 for faster, cheaper execution. MLMA ERC-20, veMLMA staking, PONO governance credential, reward distribution, and LCO₂ market activity. EVM liquidity composability with no cross-chain bridge dependency at launch.
From verified data to tradeable instrument.
The Credit Studio is the lifecycle layer that converts dMRV output into pre-finance LCO₂ instruments and verified VCO₂ credits. Project developers monitor issuance, conversion, retirement, and registry status from a single workspace.
Project Onboarding
Site assessment, methodology selection, and Hex Node deployment planning. Onboarding covers hardware deployment and field installation. Continuous signed telemetry makes sub-acre deployments economically viable.
LCO₂ Issuance
Pre-finance instrument minted on Base against the live data stream at a discount to expected verified value. Designed to compress time-to-capital from the 18 to 24 months typical of conventional credit issuance into a continuous-stream pre-finance flow. Specific discount and timing parameters are protocol targets, finalized at methodology integration.
VCO₂ Conversion
Upon registry approval, the LCO₂ instrument converts to a VCO₂ verified credit with cryptographic chain of custody from device signature to ledger record. Audit-ready, retirement-ready, and formatted for the integrated registry.
Methodology-compatible, not methodology-substituting.
The dMRV engine speaks the language of every major durable carbon removal registry. Mālama's role is digital evidence infrastructure for operating conditions, environmental context, logistics traceability, and project-level data integrity, not the sole determinant of issued removal quantity. Methodology rules still govern quantification. Switch registries without switching infrastructure.
Beyond estimation. The AI measurement crisis has a hardware answer.
Mālama Labs is bringing rack-level hardware-signed power attestation to AI compute facilities. The same verification pipeline that produced 2,786+ on-chain SaveCards from the Texas Pilot Node now extends to the largest unverified emissions source in the world: AI compute.
Software telemetry tells you what the software says. Mālama proves it at the hardware line.
The hardware answer to AI's biggest blind spot.
A 60-second walkthrough of why estimation breaks at AI scale and how rack-level attestation with cryptographic signing closes the gap. Same proven pipeline that anchors the Texas Pilot Node, applied to the largest unmeasured emissions source on the planet.
The measurement gap is not a rounding error.
The AI industry has no standardized methodology for measuring its environmental footprint. Companies disclose what they choose, if they disclose at all. Software telemetry runs on the same machines it is measuring, which means the trust loop is unbroken and unverifiable.
The Federation of American Scientists concluded that Meta's actual emissions may be up to 19,000× higher than market-based reports suggest. That is the difference between climate disclosure as a marketing exercise and climate disclosure as physical reality.
Every model upgrade, every video generation, every reasoning step compounds the gap. A single 5-second video generation consumes ~944 Wh — a day of laptop power for a single clip. GPT-o3 uses ~39.2 Wh per prompt, roughly 2,500× a lightweight text classifier.
Voluntary disclosures will not close this gap. Software telemetry that runs on the systems it is measuring will not close this gap. Hardware-signed measurement at the rack, independent of the host software, will.
What aipower.fyi reveals.
Mālama's AI Energy Impact dashboard tracks 30 AI models with full methodology transparency. Every assumption is published with confidence levels and source citations. The contribution form is open. This is the estimation tier. Hardware sensors are next.
Per 5-second clip. Equivalent to a day of laptop power. Facility-average water allocation of up to 1 L per clip.
Per prompt. ~2,500× more than lightweight classification (0.016 Wh). Frontier reasoning is energy-intensive by design.
Between most- and least-efficient AI tasks. Model choice and workload pattern matter enormously.
Multi-step agent workflows compound cost per task. Next-generation AI is more agentic by default.
Directly measured. Cryptographically signed. Independently verifiable.
We deploy a dedicated attestation appliance between the rack power feed and the compute infrastructure. It continuously measures electrical load, pairs it with GPU and system telemetry, and signs every record in a tamper-resistant signing element before release. The signed record is anchored to a public ledger. Any auditor, today or a decade from now, can verify any record without trusting the operator.
Rack-level, per-workload attribution
High-frequency current and voltage measurement at the rack power feed, correlated with per-GPU telemetry and request traces emitted by standard inference servers. The result is per-workload energy attribution with published uncertainty bounds. Not per-inference point estimates, but defensible envelopes bound to a hardware-signed record.
Measured at the facility, allocated to the workload
Cooling water and grid carbon are measured at facility meters and grid interconnects, not at the workload. The attestation appliance binds facility meters into the same signed envelope as rack power, producing a transparent allocation from hardware-measured facility metrics to workload-level reports. Allocation methodology is published with each record. What is measured is labeled measured. What is allocated is labeled allocated.
Real-time locational, not annual averages
Every signed energy record is paired with the grid's carbon intensity at the time and location of consumption, not an annual average. The same kilowatt-hour carries radically different emissions weight depending on when and where it was drawn. Market-based reporting is supported with cryptographic receipts for power purchase agreements and guarantees of origin.
One trust architecture. Two upstream data streams.
The AI compute product line is not a parallel stack. It is the same six-layer architecture that runs in Texas, with a rack-form-factor signing device feeding the same Hex Node validators as the environmental hex nodes.
AI compute pilot deployment: 2026.
Three buyers. One verified data stream.
Differentiated disclosure, shared substrate.
AI Infrastructure & Sustainability Teams
Hardware-verified electricity consumption with workload-level attribution, formatted for EU CSRD ESRS E1, California SB-253 / SB-261, and SBTi-aligned reporting. Replace software-only telemetry with an independent hardware attestation your auditors can verify without trusting your operations team.
Scope 2 (location-based and market-based): rack-level energy consumption with timestamped grid carbon intensity, signed in silicon and anchored on-chain. Supports GHG Protocol Scope 2 Guidance with cryptographic receipts for PPAs and guarantees of origin.
AI Procurement & Corporate Emissions Reporting
Procuring AI compute from hyperscalers and inference platforms. You need defensible Scope 3 Category 1 (purchased goods and services) and Category 11 (use of sold products) attribution per AI workload for corporate emissions reporting. Mālama provides the hardware-signed audit trail your ESG framework and external auditor require.
Scope 3 support: workload-level energy and carbon attribution derived from provider-side hardware measurement, reconciled to your consumption, with selective-disclosure proofs that do not expose provider trade secrets.
Workload-Level Reporting Without Exposing Operations
We understand that your per-model, per-customer, per-cluster load curves are competitive information. Mālama's selective-disclosure architecture anchors cryptographic commitments to workload-level records on-chain while keeping the underlying detail in your control. Auditors verify aggregate claims against Merkle roots; zero-knowledge range proofs answer regulatory queries without revealing operational data.
Academic, Standards, and Policy Communities
Mālama's open methodology, public dashboard, and verifiable data layer are a public good for the field. aipower.fyi is the contribution tier. The hardware layer is the verification tier. Both are documented, both are published.
From dashboard to deployed sensors.
| Milestone | Status |
|---|---|
| aipower.fyi dashboard — 30 AI models tracked, open methodology, contribution form active | LIVE |
| Reference attestation appliance — architecture reference design, hardware-signed methodology note published | H2 2026 |
| Rack sensor pilot — first attestation appliance deployment in a partner facility | H2 2026 · partner LOIs in process |
| Hex Node validator integration — AI compute packets validated by the same Hex Node network that handles Carbon SaveCards | Q4 2026 |
| Selective-disclosure module — zero-knowledge range proofs for hyperscaler aggregate reporting | 2027 |
| CSRD / SB-253 integration — pre-built export schemas for EU ESRS E1 and California CARB reporting | 2027 |
| Multi-site deployment program — hyperscaler and colo operator expansion | 2027 |