Edge-AI SaaS

Grid intelligence belongs at the edge, not the cloud.

VPP Dynamics delivers the first IEEE 1547-2018 Category III validated edge-AI platform for virtual power plant aggregation, faster than the regulatory threshold, on commodity hardware, with zero vendor lock-in.

● IEEE 1547-2018 Cat III — 9/9 validated IEEE 2030.14-2024 — 7/7 VPP functions 7.5M telemetry rows · 41-day HIL

Request Pilot Demo View Technology

4.5 ms

End-to-end edge pipeline

2,222×

Faster than IEEE Cat B limit

7.5M

HIL telemetry rows validated

Why VPP Dynamics

What the incumbents can't offer.

While traditional incumbents dominate gigawatt-scale portfolios with enterprise-heavy, proprietary frameworks, they lock out the middle market. None of them offer what VPP Dynamics has built: hardware-validated, edge-first, open-standard VPP control accessible below utility balance-sheet pricing.

⚡

Sub-5ms — Not Cloud-Dependent

Competitors route dispatch through the cloud (65–2,000 ms). Our Edge AI node responds in 4.5 ms, enabling frequency response and Volt-VAR control that cloud platforms physically cannot match.

🧠

AI at the Device Level

Legacy VPPs process ML in the cloud. We run LSTM inference on the edge gateway in 0.644 ms, ensuring the AI remains operational even during internet outages.

🔒

Open Stack, No Lock-In

Every incumbent uses a proprietary protocol stack. We run fully open, industry-standard interoperability profiles, allowing any certified smart inverter to connects natively without custom integration code.

📐

Compliance First, Not Claimed

Evaluation of our platform demonstrated full execution of all nine IEEE 1547-2018 Category III functions. The 41-day continuous dataset comprises of over 7.5 million rows of hardware telemetry and will be presented at CIGRE Canada 2026.

💰

Accessible Pricing

Tier 1 platforms require heavy enterprise contracts. VPP Dynamics offers a low-cost, scalable SaaS tier for small fleets, serving the co-operative, municipal utility, and community energy segment that are underserved by others.

🌐

Canadian + North American Native

Built for BCUC, AESO, IESO, and CAISO from day one. DNP3 for ENMAX/Alberta SCADA, IEEE 2030.5 for California Rule 21. Regulatory pathways are built in, not added later.

Five Core Pillars

Five pillars that reinforce each other.

Speed is only trustworthy because of Reliability. Cost is only sustainable because of Openness. Resilience is only credible because all three work without the cloud.

Speed

Real-time grid response

4.5 ms

End-to-end pipeline
On a commodity hardware

We respond to grid events 2,222× faster than the IEEE 1547-2018 Category B regulatory threshold.

Most VPP platforms route every control decision through the cloud — 200–2,000 ms of latency between voltage event and IBR response. VPP Dynamics deploys AI inference directly on each edge gateway. Our LSTM neural network executes in 0.644 ms, Volt-VAR in 0.50 ms, and the complete pipeline completes under 4.5 ms end-to-end — validated across 7.5 million samples over 41 days.

4.5 ms E2E0.644 ms LSTM2,222× Cat B marginIEEE 1547-2018 §6.5.1TFLite · OpenDSS

Cost

Accessible at every scale

Minimal cost
on commodity
hardware

Edge AI controller per site

Enterprise-grade grid intelligence at a fraction of the cost of traditional SCADA and centralised DERMS deployments.

Traditional SCADA upgrades cost $15,000–$25,000 per substation. Centralised DERMS adds six-figure licensing before a single DER is connected. VPP Dynamics deploys a low-cost Edge-AI Node at each site — running Cat III Volt-VAR, LSTM forecasting, and OpenDSS cosimulation simultaneously. Our SaaS starts at $500/month for 50 DERs.

Low-cost Edge-AI Node$500/mo Starter SaaSSub-$1K testbed

Resilience

Never cloud-dependent

∞

Autonomous offline
operation (no limit)

Our edge controllers operate indefinitely without cloud connectivity. IEEE 1547-2018 grid protection never depends on an internet connection.

The 2015 Ukraine grid attack and 2021 Colonial Pipeline incident proved what happens with a single cloud failure point. Each Edge-AI Node runs the complete IEEE 1547-2018 Cat III stack — Volt-VAR, FRT, VRT, enter-service, momentary cessation, islanding — entirely locally. Unlimited autonomous duration validated, exceeding the >4h target.

∞ autonomous validated9/9 local functionsIEEE 2030.7 alignedDefense-in-depth

Reliability

Hardware-validated

9 / 9

IEEE 1547-2018 Cat III
HIL validated

Every compliance claim is backed by 7.5 million rows of hardware-in-the-loop telemetry — not simulation, not theory.

VPP Dynamics is the only student-led project to demonstrate full IEEE 1547-2018 Category III compliance across all nine mandatory functions on a physical HIL testbed over 41 continuous days. Bus voltages measured at 0.9638–0.9877 p.u. Cap-MAPE 8.82% beating the <10% target. Will be resented at CIGRE Canada 2026.

7.5M HIL rows · 41 daysCap-MAPE 8.82%7/7 IEEE 2030.14-2024

Openness

Vendor-neutral by design

100%

Open protocols
no lock-in

Every protocol, every standard, every interface is open. Any DER hardware. Any utility. Any ISO market. Any cloud.

The VPP market’s biggest hidden cost is vendor lock-in. VPP Dynamics was architected on open standards: SunSpec Modbus, MQTT v5, OpenADR 2.0b, IEEE 2030.5, DNP3 — all on open-source software. Any SunSpec-certified inverter connects without writing a single line of custom integration code.

SunSpec · MQTT · OpenADR100% open stackIEEE 2030.5 · DNP3OpenAPI documented

1 / 5

Ready to pilot edge-AI VPP at your site?

10-site pilot program open in BC and Alberta. ESA permits and utility interconnection included in onboarding support.

Apply for Pilot → Read the Specs

Edge-AI VPP Platform · Technical Reference

The technology behind 4.5 ms grid response.

A five-stage edge pipeline — from disturbance detection to reactive power publish — executing entirely on-device in under 4.5 ms. Every number below is measured, not modelled.

4.5 ms

Total E2E Pipeline

0.644 ms

Forecast Inference

2,222×

Under IEEE Cat B Limit

6.6×

Faster Than Cloud

7.5 M

Validated Telemetry Rows

Disturbance Event Detected

0.17 ms

MQTT/TLS message arrives at Edge-AI subscriber thread. QoS-1 at-least-once. 20 Hz inverter telemetry (V, I, f, P, T) from SunSpec Model 101.

MQTT · QoS-1 · Port 8883

OpenDSS Power Flow Solve

3.0 ms

Newton-Raphson single iteration across LV feeder. Produces Layer 2 bus voltage v_pu at pv_end for Volt-VAR dispatch.

OpenDSS · Layer 2

AI Forecast Inference (LSTM)

0.645 ms

2-layer LSTM model. Cap-MAPE = 8.82%. Feed-forward signal to Volt-VAR controller.

TFLite · LSTM

Volt-VAR Curve Interpolation

0.50 ms

IEEE 1547-2018 Cat III 4-point piecewise-linear. V1=0.95→+44%Q V4=1.09→−44%Q. τ=5s lag. Validated Q=6.60% WMax.

IEEE 1547-2018 · Cat III

Reactive Power Q Publish

0.17 ms

Q setpoint (%WMax) published via MQTT/TLS QoS-1 to IBR register via Modbus/TLS. Inverter injects reactive power at PCC.

SunSpec · Modbus · Q inject

Pipeline time budget — proportional decomposition (total 2.98 ms = 100%)

OpenDSS 85.2%

6.5%

0.17 ms MQTT receive

3.0 ms OpenDSS power flow

0.645 ms LSTM inference

0.50 ms Volt-VAR interpolation

0.17 ms MQTT publish

Total E2E Decision Time

4.5 ms

0.17 MQTT receive

3.01 OpenDSS solve

0.645 LSTM inference

0.50 Volt-VAR calc

0.17 MQTT publish

2,222× under IEEE limit

Category B: 10,000 ms

6.6× faster than Cloud

Cloud: 65.67 ms

IEEE Standards Compliance

Hardware-validated. Not claimed.

Standard	Scope	Validation Status	Key Result
IEEE 1547-2018 Cat III	DER Interconnection — 9 mandatory functions	● Full — HIL validated	4.5 ms pipeline; all buses ±10% over 41 days
IEEE 2030.14-2024	VPP Architecture — 7 core functions	● Full — hardware deployed	4-node testbed; 7/7 functions operational
Modbus/TCP Security	MBAP Security 2018 · IEEE 1547 Annex B	● Implemented	mTLS port 802; 2048-bit RSA; ECDHE-PFS
NIST SP 800-131A	Cryptographic key strength	● Compliant	2048-bit RSA; ECDHE+AESGCM; TLS 1.2 min
IEEE 2030.11-2021	DERMS Functional Spec	◐ Partial	Edge voltage control; full utility DERMS in pilot
NERC CIP-015-1	Internal Network Security Monitoring (Sep 2025)	◎ Pilot phase	Cilium/Hubble INSM; mandatory from pilot day 1
IEEE 1815 (DNP3)	Utility SCADA Protocol	◎ Pilot phase	Required for AESO/ENMAX Alberta integration

AI / ML Stack

Training frameworkPyTorch 2.x

Edge inferenceTFLite / ONNX Runtime

Federated learningFlower (flwr)

Dispatch optimizerCVXPY + Stable-SB3

Model registryMLflow

Language Stack

AI/ML · DERMS · APIPython 3.11+

Edge agent · MQTT gatewayGo 1.22+

Operator dashboardTypeScript / React

InfrastructureKubernetes / Terraform

SecurityVault · Cilium · Cosign

Research & Insights

Evidence-based grid intelligence.

VPP Dynamics publishes research findings, technical notes, and market insights grounded in our validated hardware-in-the-loop testbed. Every claim is measured, timestamped, and reproducible.

⚡

Research · IEEE · Capstone

IEEE 1547-2018 Category III Validation on a 4-Node Hardware-in-the-Loop Testbed

Full validation of all nine mandatory DER interconnection functions over 41 continuous days. 7.5 million telemetry rows. Pipeline latency of 4.5 ms, 2,222× below the Category B threshold. To be presented at CIGRE Canada 2026, Calgary.

Read paper →

🧠

Technical · Edge AI

Why Grid Intelligence Belongs at the Edge: A Latency Analysis

Empirical comparison of edge (4.5 ms) vs cloud (65.67 ms) control latency for IEEE 1547-2018 Volt-VAR and frequency response. Demonstrates that cloud-dependent VPP platforms cannot physically meet Category B frequency response requirements.

VPP Dynamics · May 2026

Read analysis →

🌐

Market · VPP · Canada

The 100–1,000 DER Gap: Why Tier 1 VPP Platforms Ignore Canada's Growth Market

Existing VPPs serve GW-scale portfolios. The 100–1,000 DER segment — municipal utilities, co-operatives, community energy — is unserved. Analysis of market size, regulatory pathway, and why edge-AI SaaS is the right model.

VPP Dynamics · April 2026

Read analysis →

🔒

Security · NERC CIP

NERC CIP-015-1 and the VPP Aggregator: What Sep 2025 Means for Edge Deployments

FERC Order 907 approved CIP-015-1 on June 26, 2025. A VPP aggregating 50+ DERs under FERC Order 2222 is now classified as a medium-impact BES Cyber System. What this means for edge deployments, and how VPP Dynamics addresses INSM compliance from day one.

VPP Dynamics · March 2026

Read brief →

📊

Market · Funding

VPP Dynamics Pilot Program: 10-Site BC/Alberta Deployment Plan

Detailed overview of the 24-month commercial pilot

VPP Dynamics · June 2026

Read brief

🔋

Technical · Federated ML

Privacy-Preserving Federated Learning for DER Forecasting: Why Raw Data Should Never Leave the Edge

GDPR Article 25 and PIPEDA both classify energy consumption data as personal data. Federated learning with differential privacy (ε=1.0) lets the VPP platform improve its LSTM models without raw telemetry ever leaving the customer premise.

VPP Dynamics · February 2026

Read paper →

CIGRE Canada 2026 · Accepted Paper

Edge AI Framework for Real-time Distrinution Energy Management Systems Virtual with Power Plant Integration

Edge AI Framework for DER Management with VPP Integration: Hardware-in-the-loop validation of IEEE 1547-2018 Category III compliance using a 4-node testbed. CIGRE Canada 2026 Conference Proceedings, Calgary, AB.

Grid intelligence belongs at the edge, not the cloud.

What the incumbents can't offer.

Five pillars that reinforce each other.

The technology behind 4.5 ms grid response.

Hardware-validated. Not claimed.

Evidence-based grid intelligence.

Let's talk about your DER portfolio.

Get in touch

Vancouver

Email

Phone

Request a demo