Skip to content

Capacity decisions · Proof before autonomy

Understand your capacity. Know the limits.

Review your next AI workload or tenant commitment against declared facility limits. GridNinja offers a paid, bounded assessment of authorized historical inputs, with a decision brief, supporting model record and unresolved questions for review.

Scope an assessment See a sample decision brief

Historical inputs. No live connection or equipment control.

Illustrative facility animation · SyntheticInspect facility
Synthetic data center facility with graphite server racks, cooling fans, electrical equipment and amber service routes.
Synthetic example · 1-hour windowModel screen: REPAIR
Requested increment
7.0 MW
Modeled eligible increment
5.8 MW

Additional to the 20.0 MW reference. 2026-09-22 · 00:00–01:00 UTC.

Economics unestimated. No site action is authorized.
Accepted and delivered capacity: not applicable.

DEMO-01-Bv1.0.0 ↗

Our ambition

Data centers that become dependable assets for grids and accountable neighbors in their communities.

Grid constraintsFacility perimeterLocal community

A fenced data center, separate grid infrastructure, and a nearby local community. The scene does not define an assessment boundary or operating authority. Synthetic concept. Grid services, resource impacts and community benefits require separate evidence and review.

  1. Respect grid constraints

    Investigate the facility boundary, interconnection conditions and the evidence required for any flexibility claim.

  2. Examine local impacts

    Water, noise, heat reuse and local costs need their own scope, evidence and accountable review.

  3. Make responsibility explicit

    Keep modeled findings, commercial decisions and permission to operate distinct.

Today, GridNinja starts with a paid, bounded capacity assessment of authorized historical inputs. The civic ambition does not add services or establish community benefits.

The assessment mechanism

From physical constraints to inspectable decisions.

  1. 01Model the constraints

    Bring the workload, power, cooling and reserve conditions into one bounded assessment.

  2. 02Check the request

    Compare the proposed profile with the declared limits, evidence quality and time window.

  3. 03Inspect the evidence

    Review the result, its assumptions and the business question that still needs an answer.

A worked decision · synthetic

The limit is a finding. The commitment is a decision.

Fixture B records a proposed increment and a smaller modeled revision. The next question belongs to the service and commercial decision owner.

A useful assessment can also return a negative result or identify missing evidence.

Explore the four scenarios
DEMO-01-B / v1.0.0Model screen · REPAIR

Would the reduced 5.8 MW profile still meet the service and commercial requirement?

Requested7.0 MW
Recorded revision5.8 MW
All values are additional load at the same meter and for the same hour. The request exceeds the modeled increment by 1.2 MW.

Additional to the 20.0 MW reference. 2026-09-22 · 00:00–01:00 UTC.

The proposed 7.0 MW increment exceeds the fixture’s modeled 5.8 MW increment. A reduced workload profile requires review.

Synthetic. Economics unestimated. Accepted and delivered capacity: not applicable. No operational authority.

Read this decision brief Download PDF

Start with your decision

What are you ready to commit?

One clear question. The right evidence. An accountable decision owner.

Illustrative compute infrastructure

AI cloud infrastructure and operations

Admit the next AI workload.

Review a proposed workload increment before a service commitment. Keep the facility conditions and service need together.

Explore the decision
Illustrative tenant-capacity hall

Colocation operators and infrastructure teams

Commit colocation capacity.

Examine a proposed tenant increment before a commercial commitment. Make the reserve, evidence and SLA questions explicit.

Explore the decision

Considering cooling investment or bridge power and on-site generation? Feasibility and economics require a separate assessment scope.

Inspect the basis

Engineering you can inspect.

Start with the sample output. Review the method and the platform direction.

Explore public evidence
GRIDNINJADECISION BRIEF

DEMO-01-B · v1.0.0

A smaller commitment needs reviewSynthetic example · No operational authority

Sample output · Synthetic · v1.0.0

A versioned decision brief

A requested profile, modeled revision and unresolved question in a traceable record.

See a sample decision brief
Modeled result → review Gate operating permission

Method and authority boundaries

Proof before autonomy

Why a modeled result, a business decision and permission to operate remain distinct.

Understand the boundaries
Illustrative facility · declared boundary

Platform direction · In development

Physical constraints

The intended relationship between workloads, facility limits and inside-the-fence orchestration.

Explore the platform direction

GridNinja is developing an AI Data Center Virtual Capacity Control Plane and a runtime-assured virtual capacity engine for inside-the-fence orchestration. Proof before autonomy.

Delivery responsibilities, relevant experience and reviewer availability are established during scoping. Public team credentials and customer outcomes are not yet published.

About the work

Research informing the direction

Peer-reviewed research helps frame the questions. It does not validate GridNinja's models, demonstrate customer outcomes, or confer operating authority.

  1. Bounded flexibility

    AI data centres as grid-interactive assets

    Nature Energy · 2026

    A 256-GPU field trial demonstrated bounded power reduction while preserving the tested service requirements.

    Study limit: External research on one cluster; not a GridNinja result or a whole-site guarantee.

  2. Emissions tradeoffs

    Flexible data centers reduce power system costs but can increase emissions

    iScience · 2026

    Power-system modeling finds that flexibility can lower system costs while emissions depend on the regional generation mix.

    Study limit: Emissions increased in some modeled settings. Household bill savings and GridNinja economics are not established.

  3. Local water

    The environmental footprint of data centers in the United States

    Environmental Research Letters · 2021

    U.S. research links data-center water impacts to cooling, electricity generation, and local water stress.

    Study limit: Historical spatial modeling; not a current inventory or a site-specific water assessment.

A bounded, paid assessment

One question.
A reviewable answer.

Agree the decision, evidence and responsibilities before analysis begins.

You bring
A capacity question, a decision owner, a facility boundary and permission to use the relevant historical inputs.
You receive
A decision brief, assumptions and evidence gaps, a supporting model record, and a review of unresolved questions.

Deliverables, review rounds, exclusions, price and schedule are agreed during scoping. Do not send operational files or credentials through this website.

Review the deliverables Data handling and permissions

Your next capacity decision

Start with your operating constraints.

A scoping conversation establishes fit. Paid work starts only after agreement.

Scope an assessment