About InferaStack

From power to intelligence. Energy-aware, sovereign AI infrastructure for Australia.


Who We Are

An Australian AI infrastructure company operating from Sydney.

Our focus. InferaStack is building the control plane for energy-aware, sovereign AI infrastructure in Australia. Its approach combines GPU compute, workload orchestration and energy integration to support AI inference across dedicated facilities and distributed edge sites. The proposed commercial model brings together managed compute operations and inference services, with site assets able to be funded and owned separately.

Our company. InferaStack Pty Ltd is an independent Australian company. The company is newly formed but built by experienced infrastructure operators, combining data-centre and AI-systems experience with distributed-energy and virtual power plant expertise.


Leadership

These profiles describe individual experience, distinct from projects delivered by InferaStack.

Dexter Zhang

Founder, Co-CEO

AI infrastructure specialist spanning data-centre engineering, cloud and DevOps, large-scale production operations and AI systems.

Professional background

Dexter Zhang is the founder of InferaStack and an AI infrastructure specialist with a background spanning data-centre engineering, cloud and DevOps, large-scale production operations and AI systems. He leads InferaStack's work on sovereign AI infrastructure, distributed GPU compute and the integration of data-centre and energy-aware compute systems. His career path runs from data-centre engineering (Data Center Engineer, Vocus Group) through DevOps and cloud engineering (DevOps Engineer and DevOps Team Lead, Trend Micro; DevOps Consultant, Contino) to large-scale production operations and incident management, AI systems engineering, and now AI infrastructure founder and operator. He holds AWS Certified Solutions Architect – Professional, AWS Certified DevOps Engineer – Professional and AWS Certified Developer – Associate, together with credentials in cloud security, data-centre engineering and project management. He is Founder and Chair of ACIA, working on technology-industry engagement, ecosystem development, professional communities and innovation initiatives.

Martin Ma

Co-founder

Energy, software and the coordination of compute load with distributed energy systems.

InferaStack on LinkedIn  (opens in a new tab)


What We Are Building

Three complementary product and infrastructure lines, each described at its actual stage.

AI grid is the industry category described in NVIDIA's public material; InferaStack AI Grid is InferaStack's product within that category. The name does not imply NVIDIA approval, validation, co-design or endorsement.

Proof of concept · technical validation

InferaStack AI Grid

Energy-aware, distributed AI infrastructure for Australia, designed around NVIDIA systems at qualified sites.

InferaStack AI Grid page →

Design · first deployment on the roadmap

Sovereign AI Infrastructure & Colocation

Dedicated GPU environments in Australian data centres, designed and commissioned by InferaStack and owned by the client.

Dedicated GPU Deployments →

Private preview for invited applications

InferaStack Gateway

An OpenAI-compatible layer for model requests, metering, budgets and metadata-only audit records. In development and not generally available to customers.

API preview →


How We Work

Four commitments that govern what we build and how we describe it.

Validate before scale

Validation and site acceptance come before deployment. Scale follows measured results and demand.

Work within site limits

AI compute is designed around each site's energy resources and operating requirements. Device protection and owner policy come first.

Place workloads by data

Placement follows data classification. Sensitive data is designed to remain in controlled environments; only eligible work is designed to run on InferaStack AI Grid nodes.

State relationships precisely

Memberships, partnerships and discussions are described by their actual status.


Programs & Relationships

Program memberships, the signed channel agreement and active infrastructure relationships, stated by actual scope.

Signed channel partner agreement · Registered tier

BytePlus Channel Partner

InferaStack is a Registered channel partner of BytePlus Pte. Ltd. and resells the Lumina image and video generation platform in Australia. InferaStack is independently owned and operated. Scope, availability and pricing are agreed per project.

Product line resold: Lumina by BytePlus (opens in a new tab)

NVIDIA Inception member

NVIDIA

InferaStack has been a member of the NVIDIA Inception program since August 2026. This is not a partnership, endorsement or supply commitment.

Partner Program participant

NEXTDC

Official program participant, supporting a route to dedicated GPU colocation. The first deployment remains on the roadmap.

AWS Partner Network member (Enrolled stage)

AWS

AWS Partner Network member at the Enrolled stage. The Gateway is hosted on AWS in Sydney as a private preview for invited applications. Model processing locations vary; no AWS endorsement is implied.

Energy integration partner

ConnectVPP

Energy integration partner for proposed energy-aware AI infrastructure at qualified sites. The design remains in technical validation.

Dedicated GPU Deployments · Program announcements


ACIA products

InferaNote and InferaCard are ACIA products.


Events

Speaking engagements. Company announcements are published on the news page.


Start a conversation

Discuss a scoped inference pilot, a customer-owned GPU deployment or hosting an InferaStack AI Grid site.