01 / 10
Working session · 19 August 2026

Hardware-signed measurement
across the data center lifecycle

Power, water, air, and sound. Baseline before ground disturbance, continuous through operations. Every reading signed inside the sensor and anchored publicly, so a third party can verify it without trusting the operator and without trusting us.

Sensing domains
4
One trust architecture across all four
Lifecycle phases
6
Baseline through decommissioning
Trust stack status
Live
Carbon dMRV nodes, Cardano preprod
DC hardware status
Pre-deployment
No data center reference customer yet
Truth is infrastructure. Measured, not estimated. The claim we defend is narrow: a specific piece of silicon produced a specific reading at a specific time, and it has not been altered since.
Framing

The real issue is not carbon accounting. It is license to operate.

Data centers cannot prove their carbon numbers. That is true, and it matters. It is not the argument that gets a project built.

Water, noise, and air are what stop projects, void rezonings, and generate lawsuits. In 2026 that stopped being a forecast and became a docket.

The problem in every one of those disputes is the same, and it is a measurement problem. When a neighbor's well goes bad, when the hum starts, when the air smells different, the operator has no defensible record of what conditions were like before the facility existed.

Without a before, there is no after. Causation becomes a battle of retained experts, and the operator, who chose the site and built the thing, loses the narrative by default.

The position
Deploy the sensor network before you break ground.
Sign every reading in hardware. Anchor it publicly. Then keep measuring for the life of the asset.
Why this is a category shift
A consultant's sound study is a snapshot an opposing expert can attack. A continuously anchored record is a different category of evidence. The baseline is not a study you commission and file. It is the first phase of one continuous instrument.
Market condition

The risk surface moved in 2026. These are filed cases.

MatterLocationDateClaim
Haley v. X.AI Corp.Southaven, MSJun 2026Nuisance. Gas turbine noise: high-pitched squealing, continuous roaring, low-frequency rumbling.
NAACP v. X.AI Corp.MississippiApr 2026Clean Air Act citizen suit over unpermitted turbines.
Valenzuela v. Alliance Cloud ServicesDowagiac, MIMay 2026Class action, continuous cooling system noise.
Montgomery v. DataOne USA LLCVineland, NJMay 2026Cooling and generator noise. Voluntarily dismissed 19 June 2026.
Demers v. Digi Power X Inc.N. Tonawanda, NYApr 2026Noise and vibration.
Pearson v. Port of MorrowOregonSettled Mar 2026Nitrate groundwater contamination. $20.5M settlement.
Bd. of County Supervisors v. Oak Valley HOAPrince William Cty, VAMar 2026Rezoning voided for inadequate public notice.
Citizens Concerned re Wolf Hollow v. Marathon DigitalHood County, TXOct 2024Private nuisance. Survived motion to dismiss at the pleading stage.
Observation one
The plaintiffs' bar has found the sector. Old tools, nuisance and negligence and trespass, pointed at new infrastructure. Crypto mining litigation is the precedent, and Texas courts have already declined to dismiss noise nuisance claims at the pleading stage. The theories work.
Observation two
The defense bar's own recommended mitigation is our product. Counsel advising developers list baseline and post-construction sound studies, permitted on-site generation with monitored emissions controls, and transparent resource disclosure. That is not our marketing claim. The gap is that the recommended practice is currently executed as a pair of consultant snapshots.
The structural failure

A snapshot is attackable. A continuous anchored record is not the same object.

Status quo · the consultant pair
A consultant visits for a few days, measures under whatever conditions prevailed, issues a report. The facility is built. A complaint arrives. A second consultant measures again.
  • Sampling too short for seasonal, diurnal, weather variation
  • Equipment class often insufficient
  • Locations chosen by the developer
  • Report commissioned by the party it exonerates
  • Two studies, two methods, no continuity
  • Nothing tamper-evident, so integrity is attackable too
Continuously anchored record
  • Duration. Twelve months pre-construction across every season, weather condition, and hour of day
  • Continuity. Same instruments, same locations, before, during, and after. No methodology discontinuity to attack
  • Independence. Signed at the point of measurement by silicon the operator does not control and cannot alter, with a public timestamp
  • Falsifiability. Cuts both ways, and that is the point
The honest framing for a community meeting is not "our data proves we are clean." It is "we instrumented the site before we built, we cannot alter the record, and you can check it yourself." Evidence you could not have faked is worth more than evidence you merely say you did not fake.
Product surface

Four sensing domains. One trust architecture.

DomainMeasuresPrimary risk addressedInstrument tier
Power and carbonPer-rack kWh, measured PUE, hourly Scope 2, hourly carbon-free energy gapReporting, assurance, procurement exposureRevenue grade · IEC 62053-22 Class 0.5S
WaterWithdrawal, discharge, consumption, quality, groundwater level and chemistryPermitting, contamination claims, community trustFlow ±0.5% · probes are screening tier
AirPM, NO₂, O₃, CO, SO₂, VOC at fenceline and stacks, plus generator runtimeClean Air Act compliance, citizen suits, nuisanceSupplemental tier + reference anchor
AcousticBroadband and one-third octave SPL, tonality, low-frequency and infrasoundNuisance litigation, zoning conditions, setbacksIEC 61672 Class 1
Shared economics
The same secure element signs all four. The same pipeline anchors all four. The incremental cost of adding a domain is the sensor, not the infrastructure.
The met station is not optional
Air and acoustic data without co-located synchronized wind data cannot support or refute source attribution. It is the cheapest component and the one that makes the rest admissible.
Acoustic is where we win
An operator can comply with a 55 dBA property-line ordinance and still generate a viable nuisance claim, because nuisance turns on unreasonable interference. Tonality and low-frequency content are what the 2026 plaintiffs actually describe.
Architecture

The trust chain, identical across all four domains

01
Provision. Keypair generated inside the ATECC608B at manufacture. Private key non-exportable, unreadable by anyone including us. Public key becomes the device DID, bound to serial, manufacturing attestation, and calibration record.
02
Measure and sign. Payload of measurement plus monotonic counter plus timestamp is hashed and signed by the secure element before touching any network interface or filesystem.
03
Transport. Outbound mutually authenticated TLS. Transport is not the trust boundary. A compromised network can drop or delay a reading, which is detectable, but cannot alter one.
04
Anchor. Merkle root written to Cardano every 60 seconds. This removes Mālama from the trust path.
05
Verify and export. Any third party with a reading and its inclusion proof verifies independently, with no access to Mālama systems and no access to operator systems.
The precise claim, stated narrowly
This proves a specific piece of silicon produced a specific reading at a specific time and that it has not been altered since. It does not prove the sensor was correctly sited, correctly calibrated, or appropriate for the measurement. Those are handled by commissioning attestations, reference anchoring, calibration records, and cross-domain reconciliation. Anyone presenting the cryptography as a complete answer to data quality is overselling, and a competent opposing expert will say so.
Security posture, in three lines
It cannot change anything. It cannot see tenant data. It cannot be reached from outside. Outbound-only mTLS, no inbound connectivity, no vendor VPN, no control authority over facility systems. Full integration brief is written and available.
Lead product

MSB-01 · the site baseline package

Ships firstNo operator network integrationNo security review in the hard senseNaming is a placeholder
Typical single-campus configuration
Node typeCountPlacement logic
MWN-Acoustic6 to 10Boundary at each compass aspect, nearest residential receptors, two off-site controls
MWN-Air4 to 6Prevailing upwind and downwind, nearest receptors, planned generator yard
MWN-Water, surface2 to 4Planned intake and discharge, upstream and downstream of receiving water
MWN-Water, groundwater4 to 8Monitoring wells on site plus at least two off-site controls
Met mast110 m, feeding both air and acoustic interpretation
Gateway1 to 2LoRaWAN concentrator, cellular backhaul, no operator network dependency
Timeline
PhaseDurationOutput
Siting and design2 to 4 wkNode placement plan reviewed by acoustic and hydrogeologic consultants. Placement determines admissibility.
Well installation4 to 8 wkDrilling, screening, development. Longest lead item, typically the critical path.
Deploy and commission1 to 2 wkInstall, calibrate, verify against reference, sign commissioning attestations
Baseline collectionMin 12 monthsFull seasonal cycle. Anything shorter is attackable on the ground that it missed a season.
Baseline report2 wkStatistical characterization with the signed dataset and verification instructions attached
The twelve-month requirement is the hard part of this product. It has to be sold roughly a year before groundbreaking, which is earlier than most developers are thinking about environmental monitoring and often before land is fully controlled. This is a go-to-market constraint, not a detail.
Six phases

Same network, same keys, same chain. Continuity is the product.

01
Pre-construction baseline. Twelve months minimum. Establishes the prior condition across all four domains, with off-site controls that separate site effects from regional trends.
02
Construction. The noisiest, dustiest phase, and the one that generates the first complaints. Same instruments, no methodology break.
03
Commissioning. Generator and turbine testing against permit hour limits, with a signed evidence trail rather than a maintenance log that was never designed to be evidence. MRAA-01 and MGM enter here.
04
Operations. Continuous compliance evidence, complaint response within hours instead of weeks, hourly carbon-free energy gap, assurable disclosure data.
05
Expansion. A multi-year unbroken record walking into the second rezoning hearing. Probably the largest value in the model and the hardest to put a figure on.
06
Decommissioning. Closure conditions, site restoration evidence, and the record that ends the tail liability.
Power and carbon, why it lands
The GHG Protocol Scope 2 revision is moving the market-based method toward hourly matching and deliverability, with a revised standard anticipated late 2027. Hourly matching is a data problem before it is a procurement problem. Estimated, the carbon-free energy gap is a liability. Measured, it is a hedgeable position.
Water, why it lands
Roughly 80% of withdrawn freshwater evaporates rather than returning to the local system. Using withdrawal rather than consumption in community materials reads as evasion. Groundwater monitoring wells are the item most operators skip and the one that decides a drawdown claim.
Air, why it lands
The issue is combustion, not the data hall. Generators permitted as emergency-only carry hour limits. Run them for peak shaving or extended commissioning and the permit basis can fail. Most operators track runtime in a system that was never built to be evidence.
Credibility discipline

What we will not claim, said before anyone else says it

Instrument tiers carry different evidentiary weight
TierDomainsStanding
Revenue and reference gradePower, acoustic Class 1, flowBilling, permit compliance, expert testimony
Regulatory adjacentContinuous water quality probesScreening and trend. Permit compliance usually still requires lab analysis of grab samples on a defined schedule.
Supplemental and informationalFenceline air sensorsTrend detection, complaint response, community transparency. Not federal reference or equivalent methods.
Every reported figure carries its uncertainty. Power at roughly ±0.5%. Acoustic at Class 1 tolerances. Grid carbon intensity at roughly ±5% locational, which dominates any location-based Scope 2 figure. Supplemental air sensors carry substantially wider bounds and are always reported as ranges.
The falsifiability conversation
The record cuts both ways. If the facility causes a change, the data shows it. That property is exactly why the data carries weight, and it is not negotiable. An operator who wants deniability is not a customer. Discovery posture should be discussed with the customer's counsel early, not glossed.
Where we hold the line
  • Class 1 acoustic, never value-engineered to Class 2. Reduce node count before reducing class.
  • Supplemental tier stated explicitly in every air report, anchored by reference instruments.
  • No audio retention or transmission. Computed metrics only, boundary placement, event clips off by default.
  • Per-rack load data aggregated before egress, or an on-premise pipeline where only the Merkle root leaves.
Stage honesty
The environmental package and MRAA-01 are pre-deployment. There is no data center reference customer. The trust stack underneath them is live in the field on carbon dMRV nodes with a public record. A pilot operator would be first, and we say so.
Economics

Price it against the risk budget, not the facilities budget

Illustrative mid-size campus, MSB-01 baseline
LineRangeNote
Hardware and installation, excluding wells$200K to $300K8 acoustic, 5 air plus 1 reference anchor, 3 surface water, 6 groundwater, 1 met mast, 2 gateways
Well installation, six wells$90K to $240KThird-party pass-through. Frequently the largest single line and the critical path.
Twelve-month baseline SaaS$12K to $18K
Total pre-construction$300K to $550K
Operations, 1,000-rack facility~$1.5M + $200K/yrMRAA-01 at scale, on top of roughly $15K to $20K/yr environmental recurring
Environmental node pricing is a first-pass internal estimate and is not validated. It needs a real bill of materials against instrument quotes before any customer sees a number. Power pricing carries forward from the existing model.
The wrong benchmark
Do not benchmark against a consultant's baseline study at $30K to $80K. That comparison loses, and it is the wrong frame anyway.
The right benchmark
  • A single nuisance class action. Defense costs alone run well into seven figures before any judgment. The Oregon groundwater matter settled at $20.5M.
  • A rezoning delay. A voided or delayed approval costs more per month than the entire monitoring program.
  • A permit condition triggered by absent data. Regulators facing uncertainty impose conservative conditions on cooling capacity or generator hours, permanently.
  • A stalled expansion. The compounding value, and the hardest to quantify.
This is siting insurance with an operational dividend, not an environmental monitoring line item.