AI Data Center Virtual Capacity Control Plane
GridNinja turns site constraints into proof-adjusted virtual capacity through Shadow Mode, runtime assurance, dispatch envelopes, and operator-readable evidence.
Infrastructure X-Ray
GridNinja evaluates power, cooling, storage, workload, policy, and telemetry trust as inspectable layers before authority can expand.
Product modules
Each module exists to make extra MW usable, explainable, and auditable before autonomy expands.
Quantifies proof-adjusted capacity in Shadow Mode before any bounded autonomy is considered.
Summarizes usable capacity, binding constraints, data confidence, policy posture, and proof artifacts in one operator-legible output.
Maps generators, batteries, UPS reserve, cooling, and workload posture into a dispatch envelope that protects resilience margin.
Publishes flexible MW, ramp limits, reconnection posture, and no-proof gaps without implying utility approval or replacing studies.
Treats water and thermal posture as capacity constraints, reported as scenario evidence rather than broad sustainability claims.
Product boundary
Platform architecture
The platform is structured so the same path that ranks and gates action bundles also generates audit-ready evidence.
Read-only integration across mixed power, cooling, DER, and workload telemetry.
Physics-backed feasibility with learned residuals for site-specific behavior.
Constructs candidate bundles across workloads, cooling, bridge assets, and reserves.
Allow / repair / reject / no-proof with visible margins and fallback logic.
Operators and declared site policy remain final authority before any bounded control.
Structured logs, replay, Load Passports, accepted-headroom ledgers, and audit trails.
Proof chain
proof_root: 8f4c...91a
Can do
Rank candidate workload, cooling, and bridge-power bundles.
Cannot do
Cannot approve site actions or mutate authority.
Authority boundary
GridNinja can propose and verify candidate actions, but live authority stays inside declared site policy, deterministic checks, runtime assurance, and operator review.
Runtime Assurance
Deterministic verification checks the active dispatch envelope before a receipt and replayable proof record are produced.
Dispatch Envelope
See how GridNinja turns requested virtual capacity into an allow / repair / reject / no-proof dispatch envelope with binding constraints, evidence artifacts, and read-only Shadow Mode proof.
AI Data Center Load Passport
The Load Passport binds accepted capacity to ramp limits, reserve floors, freshness, no-proof gaps, and accepted-headroom evidence.
Constraint braid
The braid links power, cooling, storage, workloads, and telemetry trust to the runtime assurance outcome so virtual capacity is not treated as a generic optimizer score.
Constraint braid
Binding constraint
Transformer T2 ramp envelope
Proof object
Capacity Waterfall row 002
Accepted posture
Repair to declared ramp limit before acceptance
source
EPMS-A + PCC telemetry
constraint
T2_RAMP_LIMIT
proof_row
waterfall.power.row_002
Safety boundary
proposal layer
Advisory only
advisory-to-authority boundary
Solver + RTA gate
proof output
Receipt or no-proof row
ML / UI / fleet proposal
Solver / RTA / receipts / proof
Clipped to +1.1 MW for 30 minutes before it can be accepted.
T+00:18
solid verified evidence after repair: reserve floor remains intact
Runtime assurance is the approval boundary, not the proposal layer.
Fleet evidence
FleetOS can compare proof posture across sites, but site policy, operators, and runtime assurance remain the approval boundary.
Fleet evidence
FleetOS aggregates signed site envelopes, proof roots, accepted headroom, and no-proof gaps while local policy remains the authority boundary.
Artifact surfaces
Each artifact exists so operators and stakeholders can inspect capacity claims, refusal logic, and evidence quality without treating a dashboard as authority.
Operators, hyperscalers, investors
A site-specific capacity identity that summarizes proof-adjusted virtual capacity, binding constraints, freshness posture, and declared operating policy.
Illustrative sample · evidence chain completeExecutives, operators, reviewers
A transparent reduction from nominal headroom to safe, usable, auditable capacity after electrical, reserve, thermal, workload, and telemetry checks.
Illustrative sample · evidence chain completeOperators, auditors, fleet teams
A time-indexed record of accepted capacity states, rejected actions, no-proof intervals, and the proof root behind each decision.
Illustrative sample · evidence chain completeFacilities, BESS/UPS partners
A resilience-first view of UPS, BESS, and bridge-power margins so flexibility does not spend emergency posture silently.
Operators, CISOs, program owners
A remediation list for missing telemetry, incomplete topology, stale policy, or weak evidence that prevents safe approval.
Utilities, EMCs, partners
A planning-facing summary of flexible MW, ramp limits, reconnection envelope, confidence posture, and exceptions.
Related operator resources
Capacity Audit
GridNinja is not a generic optimizer. The platform is designed so the same telemetry, topology, policy, and runtime assurance path creates the evidence needed to inspect, refuse, audit, or operationalize capacity.
