The company embeds compute adjacent to community facilities — leisure centres, swimming pools — and treats thermal output as a product rather than a liability

Decision Lens

The tension is scale versus signal. Deep Green’s Urmston facility is a 400kW modular site, and Zendo Energy is a pre-seed-stage software firm — neither qualifies as a peer benchmark for operators managing multi-GW portfolios. What demands attention is the underlying mechanism: disaggregating energy procurement to the workload level rather than treating the entire data center as a single load, while simultaneously exporting waste heat to adjacent community infrastructure as a recoverable asset. If either model proves economically defensible at larger loads, it challenges two assumptions embedded in most current procurement frameworks — that demand is monolithic, and that thermal output is a cost center rather than a revenue line.

90-Second Brief

As the week closes, deep Green has partnered with Zendo Energy to procure clean energy and deploy energy management software at its 400kW flagship data center in Urmston, Greater Manchester, with Engie as energy supplier. The facility routes excess server heat to warm Trafford Leisure Centre’s swimming pool, turning a waste stream into a community asset. Zendo, which emerged from stealth following a $2.2 million pre-seed round, also holds a 40GWh annual supply contract with Galaxy Data Centers in Redhill, Greater London. Both parties describe the partnership as a blueprint for the next generation of flexible, sustainable data center deployments.

What’s Actually Happening

Deep Green’s operating model departs from the standard data center template in two structural ways. The company embeds compute adjacent to community facilities — leisure centres, swimming pools — and treats thermal output as a product rather than a liability. The Urmston site routes waste heat to Trafford Leisure Centre’s pool. Earlier deployments in Exmouth and a planned 500kW installation adjacent to the York Stadium Leisure Complex in Huntingdon follow the same logic, suggesting deliberate site-selection strategy rather than opportunistic arrangement.

The energy software layer works differently from conventional approaches. Zendo’s Energy OS disaggregates demand to the individual workload level rather than managing the facility as a single commercial load, covering energy monitoring, forecasting, and capacity optimization. Zendo’s COO frames this as addressing what energy brokers currently miss: different workloads carry different demand profiles, and treating them uniformly leaves procurement efficiency uncaptured. The September 2023 contract with Galaxy Data Centers for 40GWh annually at its Redhill facility indicates the platform has moved beyond a single client relationship.

Why It Matters for Global Heads of Data Center Energy?

At 400kW, this project does not register as a portfolio event. The relevance is architectural. Two questions are worth stress-testing internally.

The first is workload-level procurement. If disaggregating demand profiles can genuinely reshape the load curve — reducing contracted capacity requirements, improving curtailment management, or enabling more precise PPA sizing — the efficiency argument holds at any scale. Most enterprise and hyperscale operators currently model energy demand at the facility or campus level. Whether granular workload visibility produces material procurement gains at 100MW+ loads is unconfirmed, but the structural question is legitimate: is efficiency being left on the table by treating the data center as a single load in contract negotiations?

The second is heat as a recoverable asset. Community heat reuse changes the planning and social license calculus for urban or peri-urban sites. In markets where grid capacity constraints and planning opposition are compressing site selection options, a heat export agreement with adjacent infrastructure could accelerate approvals and partially offset thermal management costs. Urban power scarcity is not a niche problem.

The Forward View

Zendo’s early commercial trajectory — two client contracts within its first operating year on a $2.2 million pre-seed base — suggests the energy software layer for smaller operators is gaining traction faster than the funding profile might imply. Whether workload-level procurement translates to enterprise or hyperscale environments will likely become clearer in the next 18–24 months if the firm secures a larger deployment and discloses measurable outcomes.

For Deep Green, the heat reuse model carries an inherent geographic constraint: it requires a co-located thermal offtaker, which narrows site selection. Whether the unit economics hold without a captive heat customer — exporting to a district heating network, for instance — remains untested as the company scales toward 500kW deployments.

The more consequential forward signal for senior operators may be regulatory. UK planning and energy policy is increasingly attentive to data center heat recovery. If policy shifts toward mandating or incentivizing heat reuse in urban data center approvals — a direction several European jurisdictions have already moved — operators without heat export infrastructure face a compliance dimension currently absent from most long-range capital planning.

What We’re Uncertain About?

  • Zendo’s claimed 25% monthly energy cost savings: This figure is self-reported and has not been validated by independent analysis or disclosed operator billing data. Resolution requires third-party audit results or multi-year performance data from a scaled deployment.

  • Workload-level procurement at hyperscale loads: The model is demonstrated at 400kW with a constrained workload set. At 100MW+ with heterogeneous tenant workloads, whether disaggregation produces a meaningful demand-curve shift — or simply adds operational complexity — is structurally untested.

  • Heat reuse economics without a captive offtaker: Deep Green’s model depends on adjacent thermal demand being available and reliable. The unit economics if heat must be rejected or routed to district networks at variable tariffs are not disclosed, which matters for any operator evaluating urban micro-site strategies.

  • Zendo’s enterprise integration readiness: With $2.2 million in pre-seed funding, Zendo is early-stage. Its compatibility with large-scale SCADA, BMS, or ERP environments, and its data security posture for multi-tenant hyperscale operations, are unconfirmed.

One Question to Bring to Your Team

If workload-level energy disaggregation can demonstrably reshape our contracted demand profile and reduce peak capacity requirements, which of our constrained urban facilities would be the right pilot environment — and does that decision currently sit with energy procurement, operations, or neither?

Sources

  • Datacenterdynamics — Deep Green taps Zendo Energy for Manchester data center renewable energy deal (Link)

Tags