Alberta does not have a megawatt shortage of announcements. It has a scarcity of power that is actually contracted, allocated, or permitted as dedicated generation. That is why Power Certainty is 25% of the DC Intel score - more than sponsor, site, commercial commitment or execution.
The request book makes the weighting obvious. About 19,565 MW of large-load requests as of 30 July 2026 sit against about 12,800 MW of 2025 maximum demand. AESO Phase I allocated 1,200 MW and stopped there: 970 MW to GLDC, 230 MW to Keephills. A request is not an allocation.
Three different power objects
The tracker keeps mixing these if a reader is not careful:
- A connection request. Goldfinch carries 950 MW as an AESO connection request (P3102). That is real process, not a grid right.
- A signed or allocated load contract. Phase I at GLDC and Keephills. This is the scarce grid object in the current public record.
- Dedicated generation. Meta’s approved 932 MW Greenlight plant with FID; Bitdeer’s approved 101.133 MW gas plant; Beacon’s proposed Heartland 920 MW and Indus 1.494 GW plants; Teton behind-the-meter designs.
Adding those three as if they were one Alberta total produces a nonsense figure. Methodology says not to add MW across rows. Facility power, IT load, on-site generation and AESO requests are different measures.
What the leading names actually hold
Meta / Sturgeon is the full stack in public: 250 MW contracted Capital Power supply for initial service, a 970 MW Phase I load contract, and a dedicated plant with FID for later service. The remaining power risk is sequence - interim supply before Greenlight is in service - not the absence of a plan.
Bitdeer Fox Creek is a behind-the-meter story with a grid adjunct: an approved on-site plant plus a 99 MW interconnection, and the campus is already under construction.
Keephills holds the other Phase I contract and existing site infrastructure. It does not yet hold a data centre. Power got it to Probable. Construction evidence is what it still lacks.
Beacon Indus shows how far a generation-led campus can get without winning: municipal redesignation, a completed federal review trigger, and an AUC hearing on 1.494 GW of on-site gas - with no final AUC decision, FID, anchor customer or data-hall permits. That is why it is Probable, not Highly Credible.
Dedicated generation is not a skip code
Bring-your-own-generation / Phase 2 is described in provincial grid materials as a prioritized path, still developing. It can matter. It does not erase AUC decisions, gas supply, capital or customers.
Synapse Olds is the worked example in the other direction. A 1 GW campus concept ended when the AUC denied the associated power plant in August 2026. The proponent confirmed it would not proceed as proposed. Power decided that outcome.
Winners, in the only sense this model uses
“Winners” here means names that keep clearing execution evidence, not names that win a ranking contest. On present public facts, the power filter already matches the Highly Credible set: allocated or contracted load, or approved dedicated generation, plus construction or approval. Proposed GW roadmaps have not passed that filter.
This article does not forecast which AUC applications will be approved, how much behind-the-meter capacity will reach commercial operation, or when Phase 2 rules will settle. It does say that announcements which remain requests should not be underwritten as if they were power.
The ranked pipeline is in Alberta pipeline 2026, ranked by credibility. The shortlist is in projects most likely to proceed.