MEGAWATT — Business Case
MEGAWATT v0.1 CONCEPT
Confidential · Internal · Connected
Business Case · Feasibility Assessment

MEGAWATTMW

A data-centre development marketplace for Australia
The realestate.com of powered land — matching build-ready sites and capacity to the operators and developers chasing them, with every construction lead flowing quietly to Connected.
A$48B
DC projects underway in Australia
2nd
Globally for AI-infra investment, behind US only
78
DC projects in development right now
ZERO
Self-serve marketplaces for dev sites
Scroll to explore the case
01 — The Verdict
CONDITIONAL GOWorth building — as an Australia-first development-site marketplace, not a colocation clone

There is a real, defensible gap.
And it sits exactly where Connected’s leads live.

The obvious version of this idea — a “find and compare colocation space” portal — is already crowded and serves the wrong buyer. But the layer Connected actually cares about, matching build-ready land and power to the people who develop it, has no self-serve marketplace anywhere. It’s still run deal-by-deal through commercial property brokers. That is the wedge.

A.

The market is large, fast, and capital-flooded

≈A$48B of data-centre projects are underway in Australia and the country now ranks second globally for AI-infrastructure investment. Capacity is set to more than double by 2030. This is not a niche bet on a quiet sector.

B.

The binding constraint is power and powered land — not money

The whole industry is now gated on finding sites with secured power and planning headroom. That is precisely the thing no platform organises and exactly what Connected already tracks internally.

C.

Incumbents cluster around colocation and US/EU intelligence

datacenters.com, UPSTACK and Cloudscene own colo procurement; datacenterHawk owns subscription intelligence (but is light in APAC). The Australian development-site layer is broker-gated white space.

D.

Property marketplaces are extraordinary businesses

REA Group runs ~58% EBITDA margins and raises listing prices ~15% a year on pure network effect. Even a fraction of that economics, on assets worth tens of millions each, is a serious prize.

02 — Market Opportunity

A sector running hot — and short on land.

Australia has become one of the most attractive data-centre markets on earth, driven by AI workloads, data-sovereignty rules and undersea-cable landings. The capital is here; the developable, powered sites are the scarce input.

A$0B
New DC projects underway nationally (Feb 2026)

Planned data-centre spend doubled in just six months as the AI build-out accelerated. [Mandala / GRC]

A$0B
15 projects fast-tracked by the IDA (Mar 2026)

Government endorsed a priority pipeline for accelerated approvals — demand the system is racing to enable. [USSC]

00
MW capacity 2024 → 2030 (+A$26B invest.)

National capacity is projected to more than double this decade. [Mandala]

0
Facilities operating · 78 more in development

A concentrated, high-value supply base — small enough to map, large enough to monetise. [Rok Solid / pv mag]

0%
Of capacity concentrated in NSW & VIC

Sydney and Melbourne dominate — and are exactly where land and grid headroom are running out. [USSC]

0%
CAGR — AU hyperscale segment to 2031

Hyperscale alone is forecast to grow from ≈US$6.3B (2026) to US$16.2B (2031). [Mordor]

Signals of intent: Blackstone bought AirTrunk for ≈A$24B · Goodman’s pipeline is ≈A$11.3B (≈75% data centres) · Amazon committed ≈A$20B and Microsoft ≈A$5B to Australian capacity.

03 — The Problem We Solve

The whole market is gated on one question:
where is the power?

Capital is no longer the bottleneck. Securing a site with contracted power, fibre, water and planning approval is — and the discovery of those sites happens through private broker relationships, not a searchable market.

“Power availability — not capital — is now the principal driver of investment decisions. A ‘will-serve’ letter does not equal powered land.”
— Ropes & Gray, Data Center Investment 2026

A slow system, a fast market

Transmission and interconnection lead times run 3–8 years, while AI capacity is priced in quarters. Grid congestion is pushing developers into regional and brownfield sites — and value migrates to whoever can find the way around the slowest piece of the stack.

Discovery is broker-gated & opaque

Powered land, dev-approved parcels and powered shells are matched bespoke by CBRE, Colliers and JLL — relationship by relationship. There is no realestate.com where a developer can search “20ha, 100MW secured, E-zoned, NSW.” That is the gap Megawatt fills.

Binding inputs: power · land · planning · fibre · water New national approval framework ties sites to national priorities Grid limits → regional & brownfield shift Interconnection lag 4–8 yrs
04 — Competitive Landscape

Everyone built the colocation portal.
Nobody built the dirt.

Map the players on two axes — what they trade (capacity vs development land) and how (self-serve platform vs bespoke advisory) — and a clear quadrant of white space opens up exactly where Megawatt would sit.

Self-serve platform Bespoke / relationship-gated Colocation / capacity Development / powered land
OPEN
WHITE
SPACE
datacenters.com
UPSTACK · OCOLO
Cloudscene
datacenterHawk
CBRE · Colliers · JLL
MEGAWATT
Crowded

Colocation procurement

datacenters.com, UPSTACK, OCOLO, Datalok, Cloudscene — mature self-serve markets for renting rack space and capacity. Serves tenants, not builders.

Partial

Market intelligence

datacenterHawk — proven subscription model (≈US$4M revenue, acquired Mar 2026) selling supply/demand data. Strong in North America & Europe, light in APAC. Sells insight, not transactions.

Bespoke

Development advisory

CBRE, Colliers, JLL run land acquisition, powered-shell leasing and JV matching — but deal-by-deal, behind relationships. No searchable, self-serve layer.

Open

Self-serve development marketplace

The intersection of “development land” + “open platform” — searchable, Australia-first, organised around power. No incumbent. This is Megawatt.

05 — The Wedge

What Megawatt actually is.

A two-sided marketplace for data-centre development opportunity in Australia, organised around the one variable the industry transacts on — power — with an intelligence layer that is the real prize for Connected.

The listing schema is the IP

Not bedrooms and bathrooms — but available MW & time-to-energisation, grid/substation proximity, zoning & planning status, FIRB exposure, water & cooling, fibre, and asset stage (raw land → dev-approved → powered shell → operational with spare capacity). Nail that taxonomy and you own a structure the brokers don’t have in self-serve form.

Two leads, not one

Transaction leads (“I want this site”) route honestly to the lister — that’s the product that makes operators trust and pay for it. Build-intent leads — the aggregate pattern of who’s searching powered land — are visible only to the platform operator. That second stream is Connected’s pipeline, and harvesting it cannibalises nothing.

Australia first, development-led

Start where the gap is widest and Connected’s intelligence is deepest: Australian development sites and powered land. Layer in colocation capacity as a fast-follow once there’s traffic — rather than fighting datacenters.com and UPSTACK on colo from day one. The US (your existing remit) is Phase 3.

06 — Why Connected, Specifically

This is one of the few firms that can win it.

A builder has no asset-level conflict

Connected doesn’t want the land or the tenants — it wants to build on the site after it trades. So landowners and operators listing supply aren’t threatened by the owner, which neutralises the usual “secret marketplace owner” problem that kills neutrality.

The intelligence already exists

The Sentinel Dossier, the greenfield-vs-brownfield whitepaper and the US market-intelligence work are a populated supply side waiting to be productised. Connected can seed the map from public signals before a single operator self-lists.

Owned niche & credibility

Critical industries — data centres, AI infrastructure, advanced manufacturing — is already Connected’s positioned territory. A Megawatt platform compounds that authority instead of starting cold.

AU + US reach, in-house

The marketing function spans both markets, so the eventual US expansion has a team and pipeline behind it — and the platform feeds the BD leads on both sides quietly.

07 — Business Model

Four revenue lines — plus the unbilled one.

Megawatt earns directly from the marketplace, and indirectly — and most valuably — through the construction pipeline it surfaces for Connected.

Line
What it is
Comparable
Listings & placement
Operators, landowners and brokers pay to list sites/capacity and to feature them. Per-asset value is enormous vs residential.
REA core model
Intelligence subscription
Paid access to capacity, pipeline and planning data — the APAC layer datacenterHawk under-serves.
datacenterHawk ≈US$4M ARR
Demand-side leads
Qualified buyer/tenant enquiries routed to listers on a per-lead or success basis.
Portal lead fees
Connected pipeline (hidden)
Aggregate build-intent signals convert to construction work — the strategic reason the platform exists.
Internal · uncapped
Why the model is worth building — the proof next door

REA Group / realestate.com.au — FY2025

A single-country property marketplace built on network effects. The economics are the argument: dominate the listings layer and you earn pricing power and software-grade margins. Megawatt targets a thinner but far higher-value B2B version of the same machine.

A$0B
Revenue (+15% YoY)
0%
≈ EBITDA margin (A$970M)
0%
Annual listing price rises

realestate.com.au draws ≈4× the audience of its nearest rival on pure network effect. [REA FY25]

08 — Risks & How We Manage Them

The honest objections — and the design answers.

A clear-eyed business case names what could break it. Each of the four real risks has a structural mitigation that has to be built in from day one, not bolted on.

Highest

The “secret owner” / neutrality problem

+
A marketplace lives on perceived neutrality. If suppliers believe it’s secretly a competitor’s lead-funnel, they won’t list — and if covert ownership surfaces while transaction leads are being skewed to Connected, it damages both brands.
Mitigation — Run the transaction layer genuinely honestly (leads go to listers). Keep Connected’s harvest to the aggregate build-intent layer only. Operate Megawatt as a separate brand and entity with editorially credible, independent-looking content. Because Connected is a builder, not an asset competitor, the worst-case discovery story shrinks to “construction firm owns a useful industry platform” — unremarkable, and common practice.
High

Cold-start / two-sided liquidity

+
Marketplaces die waiting for both sides to show up. No supply → no demand → no supply.
Mitigation — The thinness of the market is the advantage here. With ~178 operating and ~78 in-development sites, the supply side is small enough to seed manually from public signals and Connected’s existing intelligence. A map that already looks complete pulls demand in; their activity then pulls suppliers into self-listing. “Do things that don’t scale” is viable when the market is hundreds, not millions.
Medium

Build cost & data maintenance

+
A credible platform needs accurate, maintained data (power status, planning) — that’s an ongoing analyst cost, not a one-off build.
Mitigation — Phase it. Validate with a clickable prototype and a seeded AU dataset before committing engineering spend. The intelligence work is partly being done internally already; the marginal cost is curation, not origination.
Medium

Incumbent response (brokers / datacenterHawk)

+
A CBRE or a well-funded datacenterHawk could move into AU development listings.
Mitigation — Move first in the AU development niche while incumbents are focused on colo (them) or NA/EU (datacenterHawk). Brokers are structurally conflicted about cannibalising their bespoke fee model with a transparent platform — that hesitation is the window.
09 — The Path Forward

How we’d actually build it.

A staged path that spends real money only after each gate is cleared — validate cheaply, seed manually, open carefully, then scale to the US.

Phase 0
Weeks 0–6

Validate

  • Clickable prototype
  • Seed 30–50 AU sites
  • Test with 5–8 developers / operators
  • Go / no-go gate
Phase 1
Q1–Q2

Seed MVP

  • AU development-site map
  • Power-led listing schema
  • Curated supply from intel
  • Two-lead routing built in
Phase 2
Q3–Q4

Open & monetise

  • Self-serve listings
  • Intelligence subscription
  • Colocation fast-follow
  • First revenue lines live
Phase 3
Year 2

Scale

  • US market (existing remit)
  • Premium data products
  • Connected pipeline at scale

Immediate next steps

1

Build the Phase 0 clickable prototypeAU sites · power-led search · enquiry flow demonstrating both lead paths

2

Pressure-test with a handful of developers & landownersWould they list? Would they pay? What schema do they need?

3

Decide the entity & brand separationStandalone Megawatt brand / entity, arm’s-length from Connected

4

Scope the seed dataset from existing intelligenceSentinel Dossier + market-intel work → curated supply side

Evidence & Sources

  1. Mordor Intelligence — Australia Data Center Market (2026)
  2. Mordor Intelligence — Australia Hyperscale Data Center Market (2026)
  3. Mandala report (via Global Risk Community) — AU capacity & investment forecast
  4. United States Studies Centre — Powering the Cloud (Apr 2026)
  5. pv magazine / Rok Solid — AU grid constraints & project counts (Feb 2026)
  6. Ropes & Gray — Data Center Investment 2026: power constraints
  7. Techerati — The Real Constraint is Power (Dec 2025)
  8. Data Center Knowledge — AU new approval framework (Mar 2026)
  9. datacenters.com · UPSTACK · OCOLO · Cloudscene — colocation marketplaces
  10. DCD / RocketReach — datacenterHawk model & acquisition (Mar 2026)
  11. CBRE Australia — data-centre advisory services
  12. Online Marketplaces / REA Group — FY2025 results
  13. The Urban Developer · Partners Group — AU site & brownfield transactions