Summary

  • EverWind estimates the proposed campus will average 500,000 litres of water a day at 420 MW, or about 1.47% of the 34 million-litre daily sustainable-yield estimate cited for Landrie Lake.
  • A January 2026 utility assessment had already modelled 33.7 million litres a day of average 2028 demand from existing customers and proposed hydrogen projects. It predates the data-centre announcement, so neither the 34.2 million-litre arithmetic overlay nor the lake-yield comparison is a current supply forecast.

The most useful way to read Point Tupper’s water number is not as a verdict on whether a data centre can be built. It is as a test of which denominator the project is asking the public to use. EverWind says its proposed 420-megawatt campus would average 500,000 litres a day at full build, roughly 1% of Landrie Lake’s identified safe yield. Dividing the company estimate by the 34-million-litre figure in the water assessment gives 1.47%, not 1%. The difference is small in absolute terms, but the lake-wide denominator omits the utility’s other expected users and the supply infrastructure needed to serve them.

The 5 January 2026 environmental-assessment registration for the Landrie Lake Water Utility gives a more demanding planning context. Its 2028 Phase 1 scenario estimates average demand of 33.7 million litres a day and a maximum of 55.7 million. Existing customers account for 6.4 million litres of average daily demand; proposed EverWind green hydrogen accounts for 8.4 million, and Bear Head Energy for 18.9 million. The table does not include the data centre: it was filed months before the 16 October announcement. It is a scenario, not a measurement or a revised post-announcement forecast.

That timing matters. Adding the campus’s full-build estimate to the older average scenario produces 34.2 million litres a day. This is arithmetic only. It assumes the 2028 scenario and 420 MW campus operate at the same time, does not adjust for the planned Little River transfer, and does not establish a water shortage. The assessment lists a 34.0-million-litre daily sustainable-yield estimate based on a 2024 bathymetric study and a 36-million-litre variant using newer climate data. The overlay sits 0.2 million litres above the first estimate and 1.8 million below the second.

Neither comparison substitutes for a new utility model with approved supply volumes, timing, seasonal peaks and conservation assumptions.

Historical withdrawals offer a different reference point, but not a forecast. The assessment says Landrie Lake withdrawals averaged 6.2 million litres a day from 2016 through 2024, ranging from 3.5 million to 18.1 million. The maximum was one short-term event in October 2018. EverWind’s projected 0.5 million litres a day would equal about 8% of that historical average if compared mechanically. Yet a single industrial customer’s design estimate cannot be added to past withdrawals as though both figures describe the same year, operating conditions or approved allocation.

EverWind describes a closed-loop water-glycol circuit, with air as the primary cooling method and water-assisted cooling during hot conditions. Its release estimates 100,000 litres a day for the initial 95 MW and an average of 500,000 litres at the proposed 420 MW build-out. These are company design figures, not metered consumption. The public announcement does not provide a data-centre-specific water agreement, an allocation schedule, a peak-day estimate or a revised utility demand plan.

The difference between a recirculating cooling design and a utility’s net withdrawal is also material: the public figure does not spell out how much water is withdrawn, consumed, returned or required during extreme heat.

The planned supply system is not yet the same as water at the campus boundary. Nova Scotia approved the Little River pumping and transmission project’s environmental assessment on 4 March, subject to 41 conditions. The province says other permits and approvals are typically required before a project proceeds. The registration states that the actual Little River withdrawal volume would be subject to a separate Surface Water Withdrawal Approval. Environmental-assessment approval therefore does not prove that the transfer is operating or that a particular customer has a firm allocation.

Provincial guidance also requires an approval for direct surface- or groundwater withdrawals above 23,000 litres a day; the available release does not say who would hold any approval for utility-delivered campus water.

Water is one part of a larger delivery package. EverWind announced more than $10 billion of planned private investment—over $8 billion for the campus and $2 billion for four wind projects—without identifying whether the dollar figures are Canadian or U.S. dollars. The company says the campus would fund attributable electricity-connection and network-upgrade costs, use several hundred megawatt-hours of batteries, and reduce or stop grid draw when needed. It has not published the battery’s power rating, dispatch terms, firm curtailment obligations or a public cost-allocation formula.

Required electricity studies and approvals remain outstanding. Those commitments cannot be treated as already contracted capacity or financing.

The project is staged too. EverWind targets 95 MW for late 2027 and another 325 MW in 2028, subject to commercial, technical and regulatory conditions. Global News identified Anthropic as the tenant for the initial phase, while Canadian Press reported a 20-year lease; the public reports do not disclose price, minimum payments, termination rights or remedies. A lease for the first phase does not establish that the later 325 MW will be built, nor does it settle water use for either stage.

The local sequence makes the disclosure gap more concrete. Port Hawkesbury residents raised questions about Landrie Lake in local reporting on 7 October. A 14 October report said the county had not received a detailed proposal. EverWind’s announcement two days later supplied a high-level water estimate, but not an allocation agreement or a new integrated demand table. That is progress in specificity, not a substitute for a public operating plan.

A credible next filing should reconcile the campus with the utility’s existing scenario: the forecast demand by phase, the amount and timing of Little River supply, the volume available under each approval, and the conditions under which large users would reduce demand. It should distinguish average use from hot-weather and peak-day use, and gross withdrawal from net consumption or return. It should also say how water-system capacity and any expansion costs are assigned among existing customers, hydrogen proposals and the data centre.

The point is not that 500,000 litres is inherently too much, or that Landrie Lake is already short of water. The utility’s own assessment describes a planned transfer that could increase yield, but its volume and permits matter. The narrower conclusion is more useful: a campus estimate that looks small against a lake-wide yield is not yet a service commitment. Point Tupper becomes investable infrastructure only when the new load, the transfer and the allocation rules appear in the same dated, permit-backed plan.

Sources