What would a good data centre look like in Nova Scotia?

Nova Scotia has a chance to attract a huge amount of investment in data centres. I think we should be working hard to get it.

On September 23, Tim Houston said he was open to proposals, with five conditions covering water, power costs and public benefits. The response to his Facebook post included concerns about all three, along with doubts that the conditions would ever be enforced.

A good deal could help us build more power, improve the grid and train people for skilled jobs. It could give a host community a larger tax base and help local businesses grow. We should be asking what it would take to secure those benefits.

In May, I wrote about how to evaluate AI data centres. The location, cooling system, electricity supply and financial terms can make two facilities very different propositions. Since then, other provinces have been pursuing projects at a scale we should pay attention to.

Saskatchewan is already competing for the investment

On September 14, Bell and Saskatchewan announced plans to expand an AI data-centre hub to 1.2 gigawatts. Ottawa puts the potential capital investment at up to C$52.5 billion.

That figure includes the buildings, customers' computing equipment and associated power generation. Bell's expansion agreement is non-binding, with development proceeding in phases as customers commit and approvals are secured. The province reported the initial 300 MW project under construction in August. The additional 900 MW is the proposed expansion.

Saskatchewan is also setting terms. Its framework requires new proponents to supply their own power. Bell says the expansion would use partner-developed natural-gas generation, with closed-loop cooling that requires no municipal water.

I like the ambition. Saskatchewan is trying to attract a major new industry while addressing the demands it would place on existing infrastructure. Nova Scotia should be doing the same work.

We would need our own supply plan. New gas generation carries emissions, and any grid connection still needs clear terms for backup power and costs. Saskatchewan's approach gives us a Canadian example to examine, including the parts we might choose to do differently.

What people objected to in Nova Scotia

The analysis of Houston's Facebook post covered 860 top-level comments from September 23 to 27. About 58% were critical or opposed, 5% were supportive, and 37% were mixed, conditional, off-topic or questions. Replies were excluded. It is a snapshot of the people who chose to comment, rather than a poll of the province.

Only 5% were openly supportive of considering a major investment with conditions attached. To me, that response is an indictment of how we greet economic opportunity. I want us to show more interest in making a good project happen, and to use our concerns to negotiate something better.

Power and bills appeared in 134 comments; water in 122. People questioned whether the grid could handle a large new customer and whether households would end up paying for upgrades. Water concerns drew on drought, wells and restrictions around Dartmouth.

Other comments questioned the jobs: plenty of work during construction, then relatively few people once the facility opens. Some doubted that the province would enforce its conditions when a large investment was at stake. Promises of local benefits will be more convincing if residents can see the payments, operating limits and consequences for breaking the agreement.

The analysis also identified 28 comments about surveillance, digital ID or social credit. Those concerns have little bearing on the facility itself, which could support any number of digital services. Building a data centre here does not determine whether people participate in a digital ID system or use any particular service.

There were practical suggestions too: seawater cooling, dedicated generation, tidal power, and reusing industrial land such as the Northern Pulp site. One theme was to put the conditions into enforceable rules, with published baselines, independent monitoring and penalties.

Those suggestions deserve to be developed into proposals. We have investors looking for places to build and residents suggesting ways to make it work here.

Do the claims in the comments hold up?

The comments included claims about Ireland's electricity use, huge power bills and water consumption. Checking them gives us a better idea of what to ask for in a Nova Scotia project.

A large power customer could help us build a better grid

The comment that data centres use 22% of Ireland's electricity was accurate for 2024. The latest official figure is 23% in 2025.

Yes, they use a lot of electricity. That also makes them a potential source of money for new generation, transmission lines and grid upgrades. A company willing to buy power for years can help make those investments viable. With the right agreement, other customers could benefit from a stronger system and share less of its cost.

Ireland's percentage tells us the scale of the demand. Whether a project benefits everyone else depends on what gets built to serve it and who pays. Nova Scotia should negotiate for new supply and improvements that help the wider grid.

The same applies to the comment comparing a one-gigawatt facility with hundreds of thousands of homes. It is a large load: running at full power all year would use 8.76 terawatt-hours. A proposal at that scale should come with a substantial plan to fund its power supply.

A data centre can use far less drinking water than people fear

The water concerns have a local basis. Lake Major customers spent two months under mandatory restrictions before returning to voluntary restrictions in November 2025. The question is how much fresh water a new facility would actually need.

Water use depends heavily on how a facility is cooled. We have several ways to reduce it, including ones that avoid using drinking water for cooling altogether.

Google uses treated wastewater at its Douglas County, Georgia data centre. Water that has already gone through the municipal system can do another useful job before it is discharged or evaporates.

Microsoft's new cooling design keeps water circulating in a sealed system, avoiding the need to keep replacing water lost to evaporation. Microsoft expects new sites using it to start operating in late 2027. The tradeoff is somewhat higher electricity use.

For a sense of scale, consider a centre running 100 MW of computing equipment year-round at Amazon's reported 2025 water efficiency. It would draw about 105 million litres of water on site annually, roughly the irrigation used by three golf courses. That calculation uses Amazon's fleet average of 0.12 litres per kilowatt-hour and the northeastern U.S. golf-facility median of about 39 million litres a year.

We have other options here. Our cool climate can reduce the need for mechanical cooling. Along the coast, seawater could carry heat away without drawing on the drinking-water supply.

Texas makes the case for clear cost protections

The comments blamed data centres for rising Texas electricity prices. They contribute to demand, but household bills also reflect fuel prices and the cost of maintaining and expanding the grid. A Dallas Fed explanation points to natural-gas prices and spending on ageing transmission and distribution infrastructure, alongside demand from population growth and other industries.

There is research that tries to isolate the data-centre effect. A May 2026 Dallas Fed presentation estimates that data centres made wholesale prices in Texas's ERCOT region about 3% higher during 2021–2025. That measures prices before the retail charges on a household bill. It supports a real effect, but does not explain the whole rise in residential prices.

Texas does offer a concrete policy example. Senate Bill 6 requires financial-commitment standards for large loads, including possible security or contributions toward transmission construction. Such protections can address the risk of a utility building for demand that arrives late, falls short or disappears.

Nova Scotia should settle who carries those risks before the utility spends the money. A developer can fund its connection, commit to minimum payments and provide security for obligations. The agreement should also cover any grid services used by a facility with its own generation.

The benefits can last beyond construction

The comments about jobs raise a fair question: what remains after the building is finished?

For its expanded Richland Parish campus, Meta, Facebook's parent company, forecasts over 7,500 jobs at peak construction and 1,000 operational roles. It plans more than US$50 billion in investment for a 5 GW campus. The permanent workforce is much smaller than the construction workforce, but it is still substantial.

Meta is also paying to train people. Its America's Workforce Academy launched with a US$115 million first-year commitment. The company promises free skilled-trades training, support while participants learn, and a guaranteed job for graduates. The credentials are designed to transfer across employers and industries. Louisiana is one of four states in the initial program.

Separately, Meta committed US$5 million to scholarships at Louisiana Delta Community College. Richland Parish high-school graduates, starting with the class of 2026, are eligible for full scholarships for data-centre-related trade courses. The academy's job guarantee is a separate commitment from these scholarships.

I would want a similar offer here: paid training, recognized qualifications and a job at the end. It could give more Nova Scotians a route into better-paid skilled work. Electricians, mechanics and other tradespeople could later use those skills on homes, businesses and energy projects across the province.

A major build will compete for the workers we already have. Funding apprenticeships and training before construction would help us increase the number of skilled workers as the demand grows. That would be a lasting benefit of the investment.

The electricity agreement is worth studying too. Entergy says its agreements would cover Meta's full cost of service and provide a combined US$2.65 billion in benefits to customers over 20 years. Those savings are forecasts, but they show the kind of return we should be negotiating for.

The supply plan includes seven new natural-gas plants, as well as batteries, nuclear upgrades, transmission and support for renewables. For Nova Scotia, I would want to use that buying power to help fund more clean electricity, with a reliable supply when wind output is low.

Google's Michigan agreement with DTE offers another example. Google announced 2.7 GW of new resources, including solar, storage and demand flexibility, and committed to cover its electricity and infrastructure needs. That includes the ability to reduce demand when the grid needs relief.

We should explore whether this investment could also help advance tidal power. Nova Scotia is accepting proposals for tidal demonstration projects in the Bay of Fundy. A large customer could help pay for testing and, if the technology performs, commit to buying its power. That could give a tidal developer the money and future revenue needed to move toward a larger project. Tidal power still needs to prove its costs, reliability and environmental performance, but a well-funded customer could help us do that work.

Another local benefit could be cheaper heat. Servers warm up as they run. Instead of releasing all that heat outdoors, pipes and heat pumps can carry it to nearby buildings for heating and hot water. Homes, a school or a community facility could use it, reducing the fuel or electricity they would otherwise buy for heat.

Fortum is preparing to do this with Microsoft's data centres in Finland, with recovery of server heat expected to begin in 2027. Here, we could look for sites near buildings that need heat and negotiate for the developer to help pay for the connection. Lower heating bills would be a benefit people could see each winter.

Turn Houston's conditions into a deal we want

Houston's five conditions fit this approach. They call for meeting or exceeding federal principles, protecting water and the environment, and protecting ratepayers. They also require lasting local benefits and a financial return beyond ordinary taxes.

The federal principles already address more than the short list suggests. They call for developers to cover the electricity-system costs attributable to their projects, including generation, transmission and substations. They also address water use, heat recovery, disclosure and independent verification.

Nova Scotia can build on that by specifying the obligations for each project. The power agreement should identify the required upgrades, who pays, and what happens if the development changes. Water approvals should specify the source, consumption limits and monitoring. Community agreements should name the payments, training commitments and delivery dates.

Ontario is proposing a separate rate class for new data centres, with details still under discussion. New York recommends US$1 million in community investment per megawatt of utility demand. Its benchmark is voluntary and was issued during a permitting pause, but it puts a number on the benefits communities might negotiate.

We should put the full deal in public view: what the company pays, what government contributes, and what residents receive. Any tax concessions or public spending should be counted when judging the return.

The province should now work with municipalities and Mi'kmaq rights holders to identify promising sites. It should establish what power and infrastructure each site needs, and give developers clear requirements and decision dates. Then it should actively seek companies willing to build on those terms.

Nova Scotia is already procuring more renewable electricity. Data-centre investment could help us finance more of it, alongside the grid improvements needed to serve new industry.

This is the "yes, if" approach to development I want to see here. We have a chance to attract major investment and use it to build things we need: more power, skilled workers, better infrastructure and stronger local businesses. Saskatchewan is pursuing that opportunity. Nova Scotia should be out competing for it too.