20.7 Gigawatts of Ambition, 1,200 Megawatts of Reality
Alberta’s data center ambitions are making headlines. Billions in announced investment. Gigawatt-class proposals. A provincial government reshaping grid policy and tax frameworks to court AI-scale compute. From the outside, it looks like an infrastructure gold rush.
But underneath the announcements sits a constraint that should command the attention of every operator, investor, and executive evaluating Western Canada as a deployment market: Alberta’s grid operator reports approximately 20.7 GW of data center load requests active in its interconnection process — against an interim large-load connection limit of just 1,200 MW through the end of 2028. That cap is already fully allocated to two projects.
This is not a story about demand. Alberta has demand. This is a story about the physics of getting power to compute — and what happens when the queue dwarfs the capacity by a factor of seventeen to one.
For infrastructure leaders navigating site selection, capital deployment, or portfolio strategy in Western Canada, the question is no longer whether Alberta is a viable market. It’s which Alberta — grid-connected or self-supplied — and whether your power pathway closes before your capital timeline expires.
The Cap Is Full. Now What?
The constraint tightening in Alberta is not abstract. It is structural, quantified, and time-bound.
The Alberta Electric System Operator (AESO) established an interim 1,200 MW large-load connection limit, effective through end-2028. Both allocations are spoken for: 970 MW to the GLDC Load project and 230 MW to Keephills Data Centre Phase I. For any operator seeking grid-connected capacity above the threshold, the near-term answer is no — not maybe, not pending, but allocated.
This creates a bifurcation that will define Alberta’s data center market for the next three to five years. Operators must choose between two fundamentally different infrastructure models, each carrying distinct capital structures, timeline risks, and governance requirements. The grid pathway is gated. The self-supply pathway is open but complex. There is no middle ground.
Three Patterns the Pipeline Numbers Don’t Show
The experienced infrastructure teams evaluating Alberta are not looking at announced megawatts. They are looking at power delivery timelines, interconnection queue velocity, and the gap between proposal and energization.
Three patterns stand out.
First, the queue-to-capacity ratio is historically unusual. A 20.7 GW request pipeline against a 1,200 MW cap is not a normal demand-supply imbalance — it signals that the interconnection process itself has become the binding constraint, not generation capacity or site availability. Well-sited capacity with a credible power pathway will command a premium precisely because the queue makes alternatives scarce.
Second, the market is segmenting along power architecture lines. Projects like eStruxture’s CAL-3 — a $750M, 90 MW-class high-density facility targeting rack densities up to 125 kW — represent the interconnection-led colocation model anchored to Calgary’s carrier ecosystem. Meanwhile, proposals like the Beacon Langdon AI Hub (400 MW on-site generation) and Crusoe’s multi-site gas-powered AI campuses represent the “bring your own generation” (BYOG) model, where compute and power are co-developed as a single infrastructure program. These are not variations on a theme. They are different businesses.
Third, enabling power infrastructure is becoming the long pole. The Greenlight Electricity Centre — an up to 1,800 MW CCGT project with Phase 1 targeting approximately 900 MW and a 907 MW AESO allocation — has targeted FID in the first half of 2026 with earliest customer grid access around 2027 and plant start-up as early as 2030. That timeline tells you everything about the distance between announcement and availability for power-dependent campuses.
Gas, Grid, and Gravity: Alberta’s Infrastructure Stack
74.7% Gas: Dispatchable by Design, Carbon-Intensive by Default
Alberta’s generation mix is gas-dominant: approximately 74.7% of 2024 generation came from natural gas, with wind at roughly 17.2% and solar at 2.3%. For data center operators, this creates a duality.
On the upside, dispatchable gas generation means the system can ramp capacity faster than hydro- or nuclear-dependent provinces. Reliability and power quality — critical for AI/HPC workloads demanding high uptime — are structurally supported by a fleet that responds to load signals.
On the downside, carbon intensity is structurally higher than hydro-heavy provinces like Quebec and British Columbia. For tenants with strict Scope 2 emissions targets, this is not a deal-breaker, but it demands an intentional procurement strategy: additionality-based power purchase agreements, 24/7 carbon matching programs, or credible pathways to low-carbon firm supply. Operators who treat the carbon question as an afterthought will find it becomes a tenant acquisition problem.
$40M/Year in Energy Alone — and That’s the Average
Alberta’s wholesale pool prices can be attractive on average — Q2 2025 averaged $40.48/MWh, Q3 2025 came in at $51.29/MWh — but the averages obscure meaningful volatility. The same reporting periods documented tight-system events, including an Energy Emergency Alert Level 3 event and periods clearing at offer-cap levels.
For a 100 MW IT load operating at 90% load factor, annual energy consumption runs approximately 788,400 MWh, implying roughly $40M per year in energy-only costs at the Q3 average — before transmission, distribution, and ancillary adders. Spot exposure at that scale is not a cost management decision; it is a solvency-relevant risk.
Disciplined operators are building layered supply stacks: long-tenor financial hedges, physical PPAs, optional self-generation, demand response participation, and storage integration. The “Alberta power advantage” is real, but it is a risk-management story, not a low-price story.
125 kW Racks in Drought Country
Canada’s cool climate is frequently cited as a data center advantage, and the federal energy regulator explicitly references reduced cooling costs as a siting factor. For Alberta, this holds — with caveats.
AI/HPC workloads push heat densities well beyond traditional enterprise profiles. New Alberta builds are targeting rack densities up to 125 kW, which drives adoption of direct-to-chip liquid cooling and hybrid thermal rejection systems. These designs can increase water-management complexity, and several high-profile Alberta proposals have already drawn scrutiny around water use in drought-prone areas.
Investors should treat water strategy as a first-order siting variable: non-potable water options, closed-loop cooling architectures, and measurable water-use commitments are becoming permitting requirements, not optional sustainability narratives.
YYCIX, AWS Calgary, and the Peering Ecosystem That Makes Colo Work
Alberta’s connectivity story is strongest where data centers anchor into local peering ecosystems. YYCIX (the Calgary Internet Exchange) maintains switch presence across multiple Calgary sub-areas, with participation from major networks, CDNs, and local operators. YEGIX serves a similar role in Edmonton.
For colocation operators, this ecosystem gravity is the difference between a facility and a platform. Cross-connect density, carrier diversity, and peering adjacency drive enterprise hybrid-cloud demand — particularly now that the AWS Canada West (Calgary) region has materially changed workload placement calculus for Western Canada.
Latency fundamentals reinforce this positioning. Fiber propagation at approximately 200,000 km/s sets a hard floor of roughly 10 ms round-trip time per 1,000 km before routing and equipment overhead. Calgary’s geographic position relative to Vancouver, Seattle, Toronto, and Montreal makes it a structurally relevant node for workloads that need Western Canada proximity without trans-continental latency penalties.
Two Lanes, One Market — and No Room in the Middle
Colocation vs. Campus: Pick a Lane
The research frames Alberta’s opportunity as two distinct investment theses — and warns explicitly against “middle ambiguity.”
Lane A: Interconnection-led colocation in Calgary and Edmonton metros. Target sites with strong peering proximity and carrier density. Monetize cross-connect ecosystems and enterprise hybrid demand. The revenue model is contracted critical power (kW/MW), term length, and ecosystem gravity. Calgary’s exchange infrastructure provides a defensible anchor.
Lane B: Power-paired AI/HPC campuses near energy nodes. For ambitions above 100 MW, assume self-supply becomes central. Align with the province’s incentive direction toward BYOG approaches and structure tax and levy implications accordingly. The revenue model may be compute contracts (GPU clusters) with dedicated power solutions and high-density thermal engineering as enabling infrastructure.
Construction cost benchmarks from disclosed Alberta projects illustrate the capital intensity: CAL-3 implies approximately $8.3M per MW, the TNE/Data District rollout implies roughly $5.3M per MW across multiple sites, and Wonder Valley’s Phase 1 — which includes a 1.4 GW off-grid power system — implies approximately $8.6M per MW inclusive of massive enabling infrastructure.
The 2% Levy: Cost Line or Tax Strategy?
Alberta’s new data centre levy framework — up to 2% on qualifying computer equipment value for sites at or above 75 MW — introduces a fiscal layer that investors must model, not just note. The mechanism includes a lower rate for BYOG sites and deductibility against Alberta corporate income tax.
Structurally, this means ownership design, profitability timing, and intercompany arrangements should be engineered so the levy credit is usable during the ramp period when it matters most. Combined with Alberta’s 8% provincial corporate tax rate (the lowest among Canadian provinces) and GST-only sales tax structure, the net fiscal position can be competitive — but only if the levy is treated as a tax-structure variable, not just a cost line item.
Concentrated Bets, Concentrated Risk
The concentration of Alberta’s near-term capacity in two allocated projects creates single-point-of-failure risk at the market level. Grid instability events — documented in quarterly market reports — reinforce the importance of redundancy in power supply architecture. Fiber diversity, cooling system resilience, and fault-domain isolation at the campus level must be engineered early, not retrofitted after occupancy.
From Announcement to Energized Rack: The Gap Nobody Models
Alberta’s data center pipeline contains proposals spanning from $750M to $70B. The distance between announcement and energized rack space is where most programs encounter friction.
The most common failure patterns mirror infrastructure coordination challenges globally, but Alberta adds province-specific dimensions. Underestimating the AESO interconnection timeline — particularly the distinction between queue position and executed load contract — can strand capital in development phases that extend well beyond initial projections. Misaligned rack density assumptions between shell design and actual tenant requirements create expensive retrofit cycles. Cooling systems designed for traditional enterprise loads cannot simply be upgraded for 125 kW rack densities without structural and mechanical rework.
The governance gap between facilities engineering, IT architecture, and financial planning is particularly acute for BYOG campuses, where the operator is simultaneously a power developer, a data center operator, and — increasingly — a participant in Alberta’s evolving emissions compliance regime. Federal Clean Electricity Regulations, in force with binding limits beginning in 2035 for covered generation units, add a long-term compliance trajectory that any gas-backed campus must underwrite explicitly.
Most infrastructure failures are coordination failures between physics, schedule, and governance. Alberta’s market structure, with its dual-pathway model and evolving regulatory framework, amplifies coordination complexity.
The Teams That Are Actually Advancing
The teams positioning successfully in Alberta share several operational postures.
They are underwriting power certainty as the primary variable — not site acquisition speed, not shell build timelines, not aspirational megawatt targets. Projects with explicit paths to contracted grid capacity, behind-the-fence generation, or credible hedging structures are advancing. Everything else is waiting.
They are designing for density class from day one: liquid-cooling readiness, heat rejection flexibility, and measurable water strategy are embedded in initial design standards, not treated as future phases. They are building network gravity plans early, anchoring to IXPs and carrier ecosystems before construction, not after. They are modeling carbon compliance as a cost-of-capital driver for gas-backed campuses, incorporating CCS readiness, credit structures, and operational constraints into baseline financial models.
And they are treating Alberta’s policy environment — the levy framework, the BYOG incentive direction, the grid pathway reform — as a dynamic variable that rewards engaged operators and penalizes those who assume stability.
Questions Your Alberta Strategy Should Answer
- What is our time-to-power versus time-to-procure gap in Alberta, and does our capital deployment timeline survive the interconnection reality?
- Are we designing for actual density class requirements or aspirational specifications that will require costly retrofit?
- Do we have a credible power pathway — grid allocation, self-supply, or hybrid — that closes within our investment horizon?
- How does the provincial levy interact with our ownership structure, profitability timing, and tax position during ramp?
- What is our campus-level fault-domain model, and where is single-point resilience exposure hiding?
- Have we underwritten federal emissions compliance costs for any gas-backed generation in our supply stack?
- Where is stranded capital exposure if interconnection timelines extend or policy frameworks shift?
Reynar Point of View
Alberta’s data center market is real, material, and structurally constrained. The 20.7 GW of demand requests against a 1,200 MW cap is not a temporary bottleneck — it is a market-shaping condition that will define winners and losers through the end of the decade.
The operators who will succeed here are the ones who treat energy as governance, not procurement. Who engineer power certainty before compute scale. Who understand that in a constrained market, the ability to close a power pathway is the competitive advantage — not the announcement of one.
Reynar IT approaches these environments with an infrastructure-first lens: energy-integrated strategy, vendor-neutral assessment, and mission-critical execution discipline. Alberta rewards operators who respect the physics. The constraint is the signal.
Coming in Part 2: Alberta’s Legislative Response
The constraint story doesn’t end at the grid. In Part 2, we examine Alberta’s legislative response to the massive surge in electricity demand caused by the global artificial intelligence boom. Through Bill 8 and Bill 12, the province has established a “Bring Your Own Power” framework, requiring large data centres to generate their own electricity to protect the grid’s stability and keep costs low for regular consumers. We’ll break down how these regulations grant the government expanded ministerial authority to manage grid connections and introduce a targeted levy that incentivizes self-sufficiency over total grid reliance — and what it means for operators building their Alberta strategy today.