Skip to main content
Global data centre energy

Energy Intelligence Platform for Data Centres

GridForce AI is Aerum Labs’ energy intelligence model — built for Australian data centres navigating the NEM’s new active grid participation framework.

Operator-led · Built for the grid edge
Context / mandate

The Australian Government’s March 2026 Data Centre Expectations require operators to demonstrate demand flexibility, peak-load management, and grid stability contribution as part of regulatory approval. GridForce AI is the evidence and optimisation layer that makes that demonstrable — starting with a read-only Phase 1 deployment that carries zero operational risk.

Read the alignment
Readout · operating envelope Cited figures
<500 ms
Agent response to grid frequency events
5 min
Market dispatch interval agents reason within
30–40 %
Of electricity bills from peak demand charges
<5.26 min
Unplanned downtime/year for five-nines uptime
Inside the product

What your operators see.

A live, volatile site turned into a short list of clear, costed actions — with your operator in control.

GridForce Console · DC-01 Live · 14:02:11 UTC
Asset stack
Grid frequency
50.00 Hz
Live activity
Operator-led
Recommended action
Peak window in 14 min
Discharge storage to cap demand at 38 MW
Holds your site under the demand-charge threshold for this peak interval. Estimated saving ~$4,200.
Operator-led · every action explained
0 auto-overrides · enterprise security standards
Illustrative simulation · not connected to a live site
How the loop works
01

Sense

Reads every energy asset on your site in real time — solar, storage, grid, and compute load — in one view.

02

Anticipate

Sees what’s coming — demand climbs, price spikes, grid disturbances — before they reach your bill or your uptime.

03

Recommend

Surfaces the precise next move, with the dollar impact and the reasoning attached — not a wall of alerts.

04

You approve

Nothing acts without your operator’s sign-off. The intelligence does the reasoning; your team stays in control.

Human in the loop
The problem

The energy-management gap that AI workloads create.

The point where existing systems stop being adequate, and operators start losing money and reliability.

01 —

Grid access is constrained and slow

Operators in major metros face multi-year connection timelines and capacity bottlenecks on aging grid infrastructure. The platform must optimise around limited grid access, not assume it will always be available.

02 —

AI workloads are unpredictable by nature

Training runs and inference bursts generate instantaneous load spikes with little warning. Agents need to anticipate demand shifts, not just react — which requires continuous forecasting, not periodic monitoring.

03 —

Renewable assets require active orchestration

Co-located solar and BESS introduce bidirectional power flows that change minute to minute. Effective orchestration means reasoning across generation, storage, and grid state simultaneously — not managing each in isolation.

04 —

Poor dispatch decisions compound into real cost

Peak demand charges, suboptimal storage dispatch, and reactive decisions accumulate into material cost. Agentic management exists to close the gap between what operators know and what they can act on in time.

The approach

A new intelligence layer for energy at the edge.

Stack tiers
04 → 01
01 — Visibility

Always watching

Continuous visibility across every energy asset — solar, storage, grid connection, and compute load — unified in a single real-time view.

02 — Intelligence

Always learning

Models that improve over time, building a precise understanding of your site’s unique energy behaviour and demand patterns.

03 — Action

Operator-led action

Explainable recommendations with the reasoning behind each one — so your team acts with confidence, not guesswork.

04 — Resilience

Always protected

Hard safety limits ensure the platform never acts outside your parameters. Grid events, frequency excursions, and load anomalies are absorbed before they reach your infrastructure.

What to expect

Measurable outcomes. No overpromising.

Based on published research and comparable deployments, this is what operators can reasonably target.

# Target Outcome Mechanism Qualifier
01 Up to 25% Quantified · cited Reduction in energy waste

AI-optimised dispatch reduces over-provisioning and curtailment losses, lowering cost per rack without compromising availability.

Supported by IEEE research on AI microgrid optimisation

02 Reduce Directional target Demand-charge spikes

Demand charges can be 30–40% of electricity bills. Better forecasting and smarter dispatch reduce unmanaged peaks over time.

Outcomes vary by site — baseline audit required

03 Reduce Directional target Exposure to unplanned downtime

Early visibility of potential fault conditions and grid instability gives operators a window to act before an incident becomes an outage.

Response time depends on operator protocols

04 Increase Directional target Renewable utilisation

Smarter scheduling of co-located solar and BESS increases self-consumption and cuts grid import during peak tariff periods.

Subject to asset configuration and site conditions

01 is the only quantified target on this page. 02–04 state a direction of travel only — magnitude is established against your own baseline during the initial audit, not claimed here.

Common questions

FAQ

Answers to the questions operators ask before requesting a conversation.

What does GridForce actually do, in plain terms?

GridForce connects to your energy assets — solar, battery storage, grid connection, and compute load — and watches them continuously. When a demand peak, price spike, or grid event is approaching, it calculates the best response and presents your operator with a clear, costed action to approve. Nothing happens without human sign-off.

How quickly does the platform respond to grid events?

Agent response to grid frequency events is under 500ms. Dispatch recommendations are generated within the 5-minute market interval, in time for your operator to act before the window closes.

Does GridForce take automated actions without operator approval?

No. Every recommended action requires explicit operator approval before execution. The platform is designed to eliminate the information gap, not to remove the operator from the loop. Hard safety limits are enforced at all times regardless of operator input.

What assets and infrastructure does GridForce integrate with?

GridForce is designed for data centres with co-located generation and storage — on-site solar, battery energy storage systems (BESS), and grid connections. It works alongside your existing SCADA, BMS, and metering infrastructure. Integration requirements are assessed during the initial conversation.

Is GridForce suitable for a facility still in planning or development?

Yes. Early engagement is often where the most value is created — GridForce can inform asset sizing, storage configuration, and grid connection strategy before capital is committed. We work with operators at build, expansion, and operational stages.

How is our data handled and is the platform secure?

GridForce is built to enterprise security standards. Operational data is encrypted in transit and at rest, access is role-based, and the platform is designed with the security requirements of critical energy infrastructure in mind. A detailed security fact sheet is available on request.

Regulatory alignment

Three required capabilities. One evidence layer.

The Australian Government’s March 2026 Expectations for data centres and AI infrastructure developers explicitly require operators to demonstrate demand flexibility, peak-load management, and contribution to grid stability — as a prerequisite for Commonwealth regulatory prioritisation. The AEMC’s concurrent access standard framework classifies large data centres as active grid participants, introducing new technical obligations at connection.

R1

Demand flexibility

AEMO-aligned telemetry captured continuously across compute load, generation, and grid import.

R2

Peak-load management

BESS dispatch records evidencing managed peaks against the demand-charge threshold.

R3

Contribution to grid stability

Demand response evidence and frequency-event response records at the connection point.

GridForce AI Phase 1 generates the AEMO-aligned telemetry, BESS dispatch records, and demand response evidence that form the documented basis of compliance demonstration — before any commercial commitments are made.

Source: Australian Government, Expectations for data centres and AI infrastructure developers, 23 March 2026 (industry.gov.au)

From the founder

“The BTM Energy Gap That Australia’s Data Center Boom Is Creating”

Data Center Dynamics · July 2026 · Opinion
Request a conversation

Your energy decisions can be better. Let’s make them intelligent.

Tell us about your site. We’ll analyse your load profile, grid exposure, and current tooling, and show you exactly where energy costs are slipping through the gaps.

30 minutes. Focused on your facility, not a walkthrough.
We arrive with data on your site. No blank-slate conversations.
No obligation. If it’s not a fit, we’ll tell you that too.
A written energy assessment follows if the numbers stack up.
Requests reviewed within two business days.
No spam · your data is never sold

Request demo access

~30 seconds · Tell us about your setup.

No spam · your data is never sold · unsubscribe anytime

Request received.

We’ll review your details and send your demo access shortly. Watch for a note from the GridForce team.

See GridForce run on your site Request demo access