# REPORT: Meta’s $9 Billion Alberta Campus Is a 1 GW Plan, Not a Built Data Centre


- Author: Maxime Marquette
- Publisher: MadMax (https://mad-max.co)
- Section: Reportage
- Published: 2026-08-10
- Canonical URL: https://mad-max.co/en/article/report-metas-9-billion-alberta-campus-is-a-1-gw-plan-not-a-built-data-centre

## The opening record is specific, and its limits are specific too

Data Center Knowledge reported on 5 August 2026 that Meta was developing a proposed 1 GW AI campus in Sturgeon County, Alberta, with more than $9 billion in investment and projected construction and permanent jobs. This article keeps the distinction between **reported fact**, **official attribution**, and **unconfirmed outcome** in view from the first line. The available material is strongest when it identifies a date, an actor, or a stated condition.

*A date can establish a claim without proving an outcome.*

The central issue is **the difference between an announced campus and an operating facility**. It cannot be reduced to a headline because the assigned sources describe both an event and the limits of what that event demonstrates. That is why the account below treats announced intentions, reported negotiations, and completed facts as different categories.

A careful reading begins with the chronology. It then asks who made the claim, what was actually announced, and which part of the account remains outside public confirmation. The discipline is simple. Do not turn a source boundary into certainty.

## The claim arrives with a date, not a finished outcome

### The Alberta announcement names its location

Data Center Knowledge reported that **Meta** was developing a campus in **Sturgeon County, Alberta**, described as the first Canadian data centre in its account. The report places the project in a construction and development context rather than presenting it as an operating site. In this part of the record, **Meta**, **Sturgeon County**, and **Canada** are the concrete reference points. The distinction matters.

*The public record is strongest where it is most specific.*

The source does not provide a primary project release in the assigned material, so the description is attributed to the published report. That is why the material supports a bounded reading of **Meta**, rather than a conclusion about every possible consequence. The relevant **timeframe** for Meta and **source attribution** for Sturgeon County remain attached to the claim. Nothing more is established.

### The projected investment is more than $9 billion

The Alberta project was reported with an investment figure of **more than $9 billion**. That number signals the intended scale of the proposal, but it does not by itself establish spending already completed or assets already in service. In this part of the record, **more than $9 billion**, **investment**, and **Alberta** are the concrete reference points. That is the live question.

*A large number does not erase an unfinished stage.*

The supplied account treats the figure as part of the development announcement, not as an audited total. That is why the material supports a bounded reading of **more than $9 billion**, rather than a conclusion about every possible consequence. The relevant **timeframe** for more than $9 billion and **source attribution** for investment remain attached to the claim. The sequence carries weight.

## A named actor changes the scale of the record

### The campus is described at one gigawatt

The proposed campus was described as a **1 GW** AI campus. Capacity language matters because a planned power scale and an active, verified load are not the same factual claim. In this part of the record, **1 GW**, **AI campus**, and **planned capacity** are the concrete reference points. The dates set the frame.

*A condition announced later can change the first agreement.*

The available item does not establish that the entire stated capacity is operating. That is why the material supports a bounded reading of **1 GW**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 1 GW and **source attribution** for AI campus remain attached to the claim. The stated condition is clear.

### Construction jobs are part of the projection

The report associated the Alberta development with roughly **3,000 construction jobs**. It is a forward-looking employment estimate tied to a project stage, not a count of workers already on site. In this part of the record, **3,000 construction jobs**, **employment estimate**, and **project stage** are the concrete reference points. The record is narrower.

*The evidence matters more than the language around it.*

No independent workforce roster is included in the assigned block. That is why the material supports a bounded reading of **3,000 construction jobs**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 3,000 construction jobs and **source attribution** for employment estimate remain attached to the claim. The available material stops there.

## The number is concrete, but its stage matters

### Permanent employment is a separate estimate

The same account projected **more than 300 permanent jobs** for the site. That is a different category from construction work and should not be merged with the temporary estimate. In this part of the record, **more than 300 permanent jobs**, **permanent roles**, and **construction work** are the concrete reference points. A plan is not a result.

*A deadline gives rhetoric a measurable edge.*

The source does not state when those permanent positions would be filled. That is why the material supports a bounded reading of **more than 300 permanent jobs**, rather than a conclusion about every possible consequence. The relevant **timeframe** for more than 300 permanent jobs and **source attribution** for permanent roles remain attached to the claim. The wording does not settle it.

### The source date fixes the snapshot

The assigned Data Center Knowledge item is dated **5 August 2026**. Its value is as a dated development snapshot, especially because data-centre projects can change status between announcement and operation. In this part of the record, **5 August 2026**, **Data Center Knowledge**, and **development snapshot** are the concrete reference points. That limit should stay visible.

*Silence from a party is not consent.*

The article does not supply a later project update. That is why the material supports a bounded reading of **5 August 2026**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 5 August 2026 and **source attribution** for Data Center Knowledge remain attached to the claim. This is not proof of completion.

## A deadline turns a statement into a test

### OpenAI is linked to a Georgia project

The wider development roundup identified **OpenAI** with **Project Camellia** in **Effingham County, Georgia**. The listing provides comparative context, not proof that this separate project shares the Alberta schedule. In this part of the record, **OpenAI**, **Project Camellia**, and **Effingham County** are the concrete reference points. The claim has a boundary.

*Reported talks are not a signed settlement.*

The projects are listed together, but their construction stages are not presented as identical. That is why the material supports a bounded reading of **OpenAI**, rather than a conclusion about every possible consequence. The relevant **timeframe** for OpenAI and **source attribution** for Project Camellia remain attached to the claim. The source fixes the timing.

### Paducah is framed as a conversion project

The report said **Brookfield** and **NextEra** planned to convert a former Department of Energy gaseous-diffusion site in **Paducah, Kentucky**. A conversion proposal carries different practical questions from a greenfield campus. In this part of the record, **Brookfield**, **NextEra**, and **Paducah** are the concrete reference points. Nothing more is established.

*The source defines the claim and its limits.*

The assigned summary does not establish the conversion as completed. That is why the material supports a bounded reading of **Brookfield**, rather than a conclusion about every possible consequence. The relevant **timeframe** for Brookfield and **source attribution** for NextEra remain attached to the claim. The status remains qualified.

## The official account has to be read in sequence

### The Paducah scale is stated as 1.2 GW

The Paducah proposal was described at **1.2 GW**. That figure is a stated development capacity and should not be read as an independent measure of current delivered power. In this part of the record, **1.2 GW**, **Paducah**, and **stated capacity** are the concrete reference points. The sequence carries weight.

*A stated reason is still a stated reason.*

No operating date is provided in the allotted source block. That is why the material supports a bounded reading of **1.2 GW**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 1.2 GW and **source attribution** for Paducah remain attached to the claim. That is a useful restraint.

### Galaxy has a Texas land footprint

The roundup described **Galaxy** as acquiring **500 acres** in **McGregor, Texas** for Project Merlin. Acreage identifies a land position, not an operating data-centre footprint. In this part of the record, **Galaxy**, **500 acres**, and **McGregor** are the concrete reference points. The stated condition is clear.

*The distinction between capacity and operation is decisive.*

The record does not equate land acquisition with completed build-out. That is why the material supports a bounded reading of **Galaxy**, rather than a conclusion about every possible consequence. The relevant **timeframe** for Galaxy and **source attribution** for 500 acres remain attached to the claim. The evidence has an edge.

## What was said before the condition is part of the story

### Project Merlin has a first-phase number

Project Merlin was reported with a **74 MW first phase**. A first-phase figure is necessarily narrower than an assertion about a fully realized campus. In this part of the record, **Project Merlin**, **74 MW**, and **first phase** are the concrete reference points. The available material stops there.

*No public confirmation means no public confirmation.*

The assigned material does not state later phases as operational facts. That is why the material supports a bounded reading of **Project Merlin**, rather than a conclusion about every possible consequence. The relevant **timeframe** for Project Merlin and **source attribution** for 74 MW remain attached to the claim. The chronology cannot be blurred.

### Google is associated with Project Tembo

The source also listed **Google** and **Project Tembo** in **Cheyenne, Wyoming**. Its inclusion shows the report tracks multiple developments rather than a single completed category. In this part of the record, **Google**, **Project Tembo**, and **Cheyenne** are the concrete reference points. The wording does not settle it.

*A sequence is not a substitute for proof.*

The source summary does not make a final completion claim for this project. That is why the material supports a bounded reading of **Google**, rather than a conclusion about every possible consequence. The relevant **timeframe** for Google and **source attribution** for Project Tembo remain attached to the claim. No further conclusion follows.

## The missing confirmation is not empty space

### Tembo carries a 2.7 GW figure

Project Tembo was described at **2.7 GW**. In a development roundup, that is a scale claim whose stage must remain attached to the source language. In this part of the record, **2.7 GW**, **Project Tembo**, and **Wyoming** are the concrete reference points. This is not proof of completion.

*The hardest fact is often the one that remains qualified.*

The record supplied for this article does not establish active use of that amount. That is why the material supports a bounded reading of **2.7 GW**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 2.7 GW and **source attribution** for Project Tembo remain attached to the claim. The category changes the reading.

### Core Scientific has a lease arrangement

The roundup reported that **Core Scientific** had **1.1 GW** of leased AI capacity under a **15-year agreement** with AMD. A lease agreement is a defined commercial arrangement, not a general description of every facility. In this part of the record, **Core Scientific**, **1.1 GW**, and **15-year agreement** are the concrete reference points. The source fixes the timing.

The supplied facts do not expand the agreement beyond the stated capacity and term. That is why the material supports a bounded reading of **Core Scientific**, rather than a conclusion about every possible consequence. The relevant **timeframe** for Core Scientific and **source attribution** for 1.1 GW remain attached to the claim. The public record is incomplete.

## A reported concession is not a public agreement

### AMD is named in the agreement

The Core Scientific entry identifies **AMD** as the counterparty in the 15-year agreement. Naming the counterpart matters because it confines the claim to that reported arrangement. In this part of the record, **AMD**, **Core Scientific**, and **AI capacity** are the concrete reference points. The status remains qualified.

No further contract terms are included in the assigned material. That is why the material supports a bounded reading of **AMD**, rather than a conclusion about every possible consequence. The relevant **timeframe** for AMD and **source attribution** for Core Scientific remain attached to the claim. The known facts are specific.

### DeepInfra is placed in Toronto

The list included **DeepInfra** in **Toronto**, with an announced **1.7 MW** site. This is a different scale category from the gigawatt proposals in the same roundup. In this part of the record, **DeepInfra**, **Toronto**, and **1.7 MW** are the concrete reference points. That is a useful restraint.

The comparison is about reported figures, not an assertion that all projects have the same maturity. That is why the material supports a bounded reading of **DeepInfra**, rather than a conclusion about every possible consequence. The relevant **timeframe** for DeepInfra and **source attribution** for Toronto remain attached to the claim. The answer is not supplied.

## The practical constraint sits behind the headline

### The Toronto site is linked to Blackwell systems

DeepInfra was reported to have more than **1,000 Nvidia Blackwell B300 GPUs**. Hardware quantity is a specific published claim and should not be stretched into a wider claim about Canadian AI capacity. In this part of the record, **1,000 Nvidia Blackwell B300 GPUs**, **DeepInfra**, and **Toronto** are the concrete reference points. The evidence has an edge.

The supplied summary does not independently verify the installation count. That is why the material supports a bounded reading of **1,000 Nvidia Blackwell B300 GPUs**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 1,000 Nvidia Blackwell B300 GPUs and **source attribution** for DeepInfra remain attached to the claim. That uncertainty is substantive.

### The roundup mixes development stages

The assigned source aggregates **construction**, **announcements**, and **plans** across several companies. That mix is the reason no single maturity label can safely be applied to every item. In this part of the record, **construction**, **announcements**, and **plans** are the concrete reference points. The chronology cannot be blurred.

The source itself is a roundup rather than a uniform project-status register. That is why the material supports a bounded reading of **construction**, rather than a conclusion about every possible consequence. The relevant **timeframe** for construction and **source attribution** for announcements remain attached to the claim. The source does not go farther.

## The source separates a plan from an operating fact

### Individual dates are not supplied for each project

The block notes that individual project dates are **unclear** for some entries. That missing precision matters when comparing the announced campuses. In this part of the record, **individual dates**, **unclear timing**, and **project entries** are the concrete reference points. No further conclusion follows.

No date should be inferred simply because the roundup has a publication date. That is why the material supports a bounded reading of **individual dates**, rather than a conclusion about every possible consequence. The relevant **timeframe** for individual dates and **source attribution** for unclear timing remain attached to the claim. The final outcome remains open.

### No primary releases were consulted

The limitations state that **primary releases** were not consulted for the listed developments. The report can therefore support attributed reporting, not a claim of primary-source confirmation. In this part of the record, **primary releases**, **limitations**, and **attributed reporting** are the concrete reference points. The category changes the reading.

This is a declared source limitation, not a claim that no company has ever issued a release. That is why the material supports a bounded reading of **primary releases**, rather than a conclusion about every possible consequence. The relevant **timeframe** for primary releases and **source attribution** for limitations remain attached to the claim. The distinction matters.

## A narrow fact can still carry a wide consequence

### The Alberta project leads the Canadian angle

The **Sturgeon County** item is notable in the assigned material because it is presented as **Meta** entering the Canadian data-centre landscape. The article keeps that characterization tied to its reporting source. In this part of the record, **Sturgeon County**, **Meta**, and **Canadian data centre** are the concrete reference points. The public record is incomplete.

The designation comes from the Data Center Knowledge account supplied for this piece. That is why the material supports a bounded reading of **Sturgeon County**, rather than a conclusion about every possible consequence. The relevant **timeframe** for Sturgeon County and **source attribution** for Meta remain attached to the claim. That is the live question.

### The $9 billion figure needs its qualifier

The phrasing is **more than $9 billion**, not a precise final cost. Retaining the qualifier prevents a projection from being recast as a completed expenditure. In this part of the record, **more than $9 billion**, **projection**, and **final cost** are the concrete reference points. The known facts are specific.

The figure remains an announced investment scale in the assigned source. That is why the material supports a bounded reading of **more than $9 billion**, rather than a conclusion about every possible consequence. The relevant **timeframe** for more than $9 billion and **source attribution** for projection remain attached to the claim. The dates set the frame.

## The stated rationale is not independent proof

### A gigawatt is a planning signal

The reported **1 GW** designation makes the Alberta campus large on paper, while the word **planned** keeps its operational status qualified. Both parts belong in the same sentence. In this part of the record, **1 GW**, **planned**, and **operational status** are the concrete reference points. The answer is not supplied.

The facts supplied do not say the campus is already delivering its named capacity. That is why the material supports a bounded reading of **1 GW**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 1 GW and **source attribution** for planned remain attached to the claim. The record is narrower.

### Employment figures cannot be fused

The report distinguishes **3,000 construction jobs** from **more than 300 permanent jobs**. Treating them as a single employment result would erase the difference between temporary and ongoing work. In this part of the record, **3,000 construction jobs**, **300 permanent jobs**, and **employment categories** are the concrete reference points. That uncertainty is substantive.

The sources do not provide a realized hiring total. That is why the material supports a bounded reading of **3,000 construction jobs**, rather than a conclusion about every possible consequence. The relevant **timeframe** for 3,000 construction jobs and **source attribution** for 300 permanent jobs remain attached to the claim. A plan is not a result.

## The next date is where the file becomes answerable

### The comparison should not create equivalence

The entries for **OpenAI**, **Google**, **Galaxy**, and others are useful comparison points, but the roundup does not establish that each project is at the same phase. In this part of the record, **OpenAI**, **Google**, and **Galaxy** are the concrete reference points. The source does not go farther.

Shared inclusion in a development list is not evidence of shared construction status. That is why the material supports a bounded reading of **OpenAI**, rather than a conclusion about every possible consequence. The relevant **timeframe** for OpenAI and **source attribution** for Google remain attached to the claim. That limit should stay visible.

### The record supports a limited conclusion

The data-centre list documents a wave of **reported projects** and capacity figures, including the **Alberta campus**. It does not turn every announcement into a completed facility. In this part of the record, **reported projects**, **Alberta campus**, and **capacity figures** are the concrete reference points. The final outcome remains open.

The distinction between prospective scale and operating reality is the core limitation of this source set. That is why the material supports a bounded reading of **reported projects**, rather than a conclusion about every possible consequence. The relevant **timeframe** for reported projects and **source attribution** for Alberta campus remain attached to the claim. The claim has a boundary.

## The conclusion is a boundary, not a flourish

Meta’s Alberta proposal is significant because of its stated scale, but the assigned evidence describes a development plan. The story is not less consequential for being qualified; it is more useful when its stage is stated plainly. The record supports a firm distinction between what is documented and what is merely expected. **the difference between an announced campus and an operating facility** is therefore best understood through the named dates, actors, and qualifications rather than through a claim of certainty the sources do not provide.

*The conclusion must stop where the evidence stops.*

The next disclosed decision, implementation step, or official statement may change the picture. Until then, the responsible conclusion is both plain and limited: **the available evidence** establishes the stated events, while **the final result** remains outside the supplied record.

**By Maxime Marquette, columnist**

## Columnist's transparency note

### Editorial stance

I am not a journalist but a **columnist and analyst**. My expertise lies in observing and analysing the geopolitical, economic and strategic dynamics that shape our world. My work is to take political strategies apart, to understand global economic movements, to put the decisions of international actors in context and to offer analytical perspectives on the changes that are redefining our societies.

I do not claim the cold objectivity of traditional journalism, which stops at reporting facts. **I claim analytical clarity**, rigorous interpretation and a deep understanding of the complex issues that concern us all. My role is to give meaning to the facts, to place them in their historical and strategic context, and to offer a critical reading of events.

### Method and sources

This text keeps the basic distinction between **verified facts and interpretive analysis**. The rule never changes: a factual claim is published only when a verifiable source supports it, and the sources this article actually used are listed under **Sources**, never here.

**Primary source families** used by the publication, when the story has them: official statements from governments and international institutions, public statements by political leaders, reports from intergovernmental organisations, dispatches from recognised international news agencies.

**Secondary source families**: specialised publications, internationally recognised news media, analyses from established research institutions, reports from industry organisations.

When an article cites statistical, economic or geopolitical data, it comes from the institutions that produce it (intergovernmental organisations, central banks, national statistics offices), and the exact institution is named under **Sources**.

### Nature of the analysis

The analyses, interpretations and perspectives in the analytical sections of this article are a **critical, contextual synthesis** based on the information available, the trends observed and the expert comments cited in the sources consulted.

My role is to interpret these facts, to set them in the context of today's geopolitical and economic dynamics, and to give them a coherent meaning within the larger story of the changes shaping our time. These analyses reflect expertise built through continuous observation of international affairs and an understanding of the strategic mechanisms that drive global actors.

This article describes a documented state of affairs on its publication date, not a prediction: later developments may shift its outlook. No update is promised in advance; when an article is corrected or completed, the change is dated in the text.

## Sources

### Primary sources

- [Data Center Knowledge development roundup](https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026)

### Secondary sources

- [Reference 1: Data Center Knowledge development roundup](https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026)

- [Reference 2: Data Center Knowledge development roundup](https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026)

- [Reference 3: Data Center Knowledge development roundup](https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026)

- [Reference 4: Data Center Knowledge development roundup](https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026)

- [Reference 5: Data Center Knowledge development roundup](https://www.datacenterknowledge.com/data-center-construction/new-data-center-developments-august-2026)

---

Source: Maxime Marquette, MadMax. When citing, name the author and link https://mad-max.co/en/article/report-metas-9-billion-alberta-campus-is-a-1-gw-plan-not-a-built-data-centre