AI Data-Centre Power and Capacity Strategy Director

The mandate

The provider's accelerator procurement curve implies far more energised capacity than its signed utility positions support, yet commercial forecasts mix reserved, probable and speculative demand. The defined problem is to determine what power can credibly become usable, where and when, and which customer commitments the resulting capacity can safely underpin.

The required artefact is a Canadian AI-infrastructure capacity strategy containing a reconciled demand envelope, site and interconnection portfolio, power-quality and cooling constraints, staged build-versus-lease choices, energy-cost sensitivities, carbon implications and a decision calendar. It must distinguish nameplate megawatts from deliverable IT load at the required rack density.

Milestone one closes after month one with a source-traceable load curve and utility-position verification. At month three, milestone two supplies site archetypes, constraint maps and downside scenarios; milestone three at month five delivers the recommended portfolio, negotiation mandates, capital sequencing and committee paper.

Acceptance rests with the COO and finance sponsor: they must be able to rerun demand, energisation-delay, curtailment, density and price assumptions; each recommended site must have verified dependencies and a decision owner; and the plan must fit the approved capital envelope. Broker claims or unqualified queue dates cannot support acceptance.

The client will provide accelerator orders, sales probabilities, current leases, utility correspondence, engineering standards, capital parameters and access to shortlisted partners. Detailed electrical design, permit applications, utility representation, property acquisition and construction delivery sit outside the engagement.

Why this is external work

Infrastructure leaders are consumed by live deployments, while commercial teams are measured on reservations and finance sees inconsistent capacity categories. None has neutral ownership of the demand-to-power reconciliation. A specialist external view can challenge both optimism and excessive contingency before the committee locks scarce capital into long-lived sites.

What you will own

• Rebuild the accelerator demand envelope from contracted, reserved, weighted-pipeline and scenario demand with explicit confidence and cancellation logic.
• Verify utility, landlord and developer claims against interconnection stage, upstream works, transformer lead times, redundancy and curtailment conditions.
• Translate facility power into usable AI capacity after cooling, electrical loss, rack density, network and maintenance constraints.
• Compare owned, leased, powered-shell and capacity-reservation pathways across timing, control, cost, reversibility and stranded-asset exposure.
• Model site portfolios under energisation delay, hardware efficiency shifts, customer concentration, power-price volatility and deployment slippage.
• Frame negotiation mandates for the three preferred markets, including evidence conditions, option structures and walk-away triggers.
• Submit a capital-allocation paper naming the selected portfolio, deferred choices, downside protections and dates when assumptions must be refreshed.

Candidate qualifications

• Directed data-centre power, site or infrastructure-capacity strategy for hyperscale cloud, colocation or compute-intensive platforms in North America.
• Converted accelerator or high-density computing demand into facility, utility and cooling requirements at a portfolio rather than single-site level.
• Evaluated interconnection queues and energisation dependencies with direct utility, developer and engineering evidence.
• Built capital scenarios that separated nameplate capacity, commissioned power and sellable IT load under schedule and efficiency uncertainty.
• Negotiated leases, options, capacity reservations or development terms involving material power availability and delivery risk.
• Presented a multi-site capacity recommendation to an investment committee and can identify which assumption most changed the approved allocation.

Non-negotiables

• Available for on-site working sessions in Toronto and travel to three shortlisted Canadian markets within five months.
• Holds no undisclosed brokerage mandate, developer economics or utility advisory conflict connected to the candidate sites.
• Will distinguish verified energisation evidence from commercial representation in every model and recommendation.
• Can supply client-rerunnable models and transfer all calculation logic, interviews and source records at acceptance.

1. 49 words maximum. Describe the largest AI or high-density capacity plan you led and the power assumption that most altered its site portfolio.
2. 49 words maximum. How would you convert a utility interconnection position into a confidence-weighted energisation date during month one?
3. 49 words maximum. Identify any current developer, utility, broker, colocation or hardware relationship relevant to Canadian site selection.