AI Data Centres Are Becoming a Local Permission Problem: What SMEs Should Learn

Reddit’s technology community is debating a Guardian report on Erin Brockovich’s campaign against AI data centres, and the discussion is worth watching if you run an SME in the UK, EU, or US.

This is Hajikreena’s view on the signal: the AI conversation is moving beyond model launches and chatbot features. It is becoming a practical question of land, electricity, water, planning consent, community trust, and operating resilience.

The Guardian reported on 29 June 2026 that Erin Brockovich is campaigning around the impact of AI data centres, with concerns focused on local resources and community power. Google News also surfaced related recent coverage from DW, CNET, USA Today, and others on public concern around the rapid expansion of data centres.

The Reddit thread is not proof that every local claim is correct. It is a useful signal that the public debate around AI infrastructure is becoming sharper, more local, and more political.

For businesses, that matters.

AI tools do not run in a vacuum. Every automation project depends on a physical and regulatory stack: data centres, cloud contracts, energy supply, network reliability, security controls, and rules about how data is processed.

Until recently, many SMEs treated AI as a software buying decision. Pick a vendor, connect a few apps, and automate the workflow.

That is no longer enough.

If AI infrastructure becomes contested, three risks become more important for operators:

1. Cost volatility: Energy, compute, and cooling pressure can feed into cloud and AI pricing.

2. Regional constraints: Data residency, environmental permitting, and local planning disputes can affect where workloads are hosted.

3. Reputation risk: Customers and staff may ask whether an AI system is efficient, responsible, and necessary.

The wrong conclusion is to pause useful automation because the infrastructure debate is uncomfortable.

The better conclusion is to become more selective.

AI should be used where it removes real operational friction: response handling, lead qualification, CRM hygiene, invoice triage, support workflows, knowledge retrieval, compliance checks, and repetitive admin. It should not be added everywhere simply because the tool is available.

A lean automation strategy is easier to defend than a bloated one.

List every AI-enabled tool in the business, including browser extensions, CRM add-ons, customer support bots, meeting note tools, and workflow automations. Many companies have more AI in production than leadership realises.

Keep the workflows that save measurable time, reduce errors, or improve customer experience. Retire experiments that create little value but still send data to external systems.

When buying AI or automation software, ask:

Where is data processed and stored?

Can we choose region or retention settings?

What happens if model prices change?

Which tasks are handled by AI, and which remain human-reviewed?

Can logs and outputs be audited?

Every important AI workflow needs a fallback. If the model provider, API, or cloud region has a problem, the business should still be able to serve customers and process core work.

If AI touches customer data or customer decisions, explain the purpose in plain English. Trust is easier to maintain when automation is visible, bounded, and useful.

The data centre debate shows that AI is becoming part of public infrastructure, not just private software. That means business leaders need to think like operators, not just buyers.

The winning SMEs will not be the ones that adopt the most AI. They will be the ones that automate carefully, govern sensibly, and keep humans in control of the moments that matter.

GOFTUS helps SMEs design AI workflows that are practical, auditable, and aligned with real business outcomes. If you want to review where automation can safely reduce admin, support load, or operational drag, speak to GOFTUS about an AI workflow assessment.