In August 2026, Cape Town's Housing Assembly and the UK nonprofit Foxglove filed a legal challenge against Equinix's proposed 174 MW Cape Town data centres. Rest of World reported on September 17 that they also asked the South African Human Rights Commission (SAHRC) to investigate and to impose a moratorium on hyperscale expansion. They further demand rules that make operators disclose water and electricity use. The facilities would use about 4.4 billion litres of water a year, according to Rest of World, and the groups' own submission equates that to the annual use of more than 18,000 South African households.
The disclosure demand is right. The blanket moratorium is the weaker half of the package.
The strongest case for a pause
The case for a freeze is serious, and it should be stated fairly. Cape Town has a recent memory of acute water scarcity. The Constitution guarantees everyone an environment that is not harmful to their health or wellbeing (section 24) and access to sufficient water (section 27), both in the Bill of Rights. The groups' joint submission to the SAHRC, filed on August 24, 2026, says South Africa has more than 60 data centre facilities with roughly 500 MW of disclosed capacity. It says the planned hyperscale pipeline would more than triple electricity demand, and that operators face no obligation to disclose consumption (submission). If regulators cannot see cumulative demand, approving projects one at a time is guesswork. In that situation a pause looks like ordinary prudence.
Why the pause overreaches
The problem the groups identify is missing information, and a moratorium does not create information. It stops projects without producing a single metered figure, and it does so across a very uneven field. Existing South African facilities are far smaller than the proposed campuses. Water use also varies widely with cooling design. A national freeze treats a closed-loop or air-cooled build the same as a water-hungry evaporative one, so it removes any reason for operators to choose the better design.
A freeze also has costs the submission does not weigh. Rest of World notes, citing McKinsey, that African compute demand could reach 2.2 gigawatts by 2030, about five times current levels. South Africa hosts the most data centres on the continent. Demand for cloud, payments, health and government services will not pause because approvals do. It will move to facilities in other countries, taking jobs, tax base and latency benefits with it. The submission itself argues the sector creates few jobs relative to investment. If that is true, an independent cost-benefit assessment, which the groups also request, should be able to show it without a national stop order.
What proportionate rules look like
The SAHRC has already framed the issue correctly. Its 22 May 2026 media advisory invited submissions on electricity and water use, transparency, accountability and regulatory adequacy, with a deadline of 30 July 2026. The advisory says the input could guide recommendations to Parliament and relevant government bodies. Reporting on the process quotes the Commission as saying a key issue is the availability, consistency and transparency of information on electricity and water demand, per the Sowetan.
That diagnosis points to a targeted response:
- Mandatory reporting before approval. Applicants should publish projected and actual annual water and electricity use, cooling method and energy source, so regulators and neighbours can test claims.
- Binding commitments. Disclosed figures should become permit conditions, with periodic audited reporting and public results.
- Cumulative-impact review. Municipal and national authorities should assess pipeline demand together, since the submission argues approvals are currently weighed project by project.
- Additionality on power. The groups object that Eskom is marketing a coal-fired surplus to operators. Requiring new clean generation to accompany very large loads addresses that without banning the load.
- Conditional, project-specific pauses. If a particular project's water source cannot be verified, delay that project until it can. Do not freeze the sector.
This approach keeps the legitimate part of the campaigners' case. It gives communities the numbers they say they have been denied. It also gives operators a predictable rulebook. Investors can price a disclosure and performance standard. They cannot price an open-ended moratorium whose end date depends on an inquiry that has not yet been designed.
Where the legal challenge fits
The litigation and the SAHRC process can do real work here. A court or the Commission can insist that permits be granted only with adequate information, and that is a precedent worth setting. Turning that into a country-wide ban would go beyond what the evidence in the record supports. The record shows an information gap and a fair worry about water and coal-fired power. It does not yet show that every planned facility is unworkable.
South Africa can be a serious host for AI infrastructure and still protect its water. The route is transparent, enforceable rules applied to each project, and Parliament and the SAHRC now have the raw material to write them.