AI, Connected

How AI is reshaping infrastructure, energy and the networks that connect it

Artificial intelligence is reshaping our daily lives faster than almost any technology before it.

Whereas the internet before it helped us to shop, consume content and research, AI is helping us create, build and digest information at an unprecedented rate.

Businesses are fundamentally rethinking how they operate around it, roles are being redesigned, and workers are under pressure to adapt or fall behind.

In September 2026, RETN surveyed 2,000 consumers across the United States, Europe, and Asia, 54% said they would be significantly affected within a few days if AI services disappeared.

With worldwide AI spending forecast to reach $2.59 trillion in 2026, this dependence is fuelling unprecedented investment in data centres – and consumers have noticed: 53% named AI tools as the biggest driver of new data-centre construction. Big Tech companies have signed agreements that extend five to ten years into the future, committing around $2 trillion to new capacity.

However, the technology doesn’t rely on data centres alone. It depends on four connected parts: the data centres where models are trained and run, the power that feeds them, the carriers whose networks move the traffic, and the people and businesses using it.

As an independent provider with a network spanning Europe, Asia, and North America, RETN is seeing new opportunities emerge in locations where connectivity is still developing, alongside growing investment in long-haul routes to connect the next generation of digital infrastructure with users.

What Next?

6 Practical Considerations for AI Infrastructure & Network Planners

Plan connectivity early. Where AI infrastructure is being developed outside established data-centre hubs, network planning may need to begin years before the facility becomes operational.
Design for physical diversity. Multiple circuits or suppliers do not necessarily provide independent infrastructure. Critical routes should be assessed for shared ducts, corridors, landing stations and other physical dependencies where relevant.
Plan for point-to-point connectivity. AI workloads can create substantial data movement between data centres and compute environments. Dedicated, scalable interconnection needs to be considered alongside public IP transit.
Plan for unpredictable growth. AI has so far added only a few percent to IP traffic, but demand for high-capacity links is already rising and much more is still to come. With no clear forecast for how AI traffic will scale, building in headroom and upgrading early is safer than waiting for demand to arrive.
Price resilience properly. Genuine physical diversity can require new infrastructure and significant civil engineering. The cost of resilience therefore cannot be assessed on bandwidth price alone.
Treat networking as part of the AI supply chain. Long lead times for specialised optical and networking equipment mean that connectivity infrastructure needs to be planned alongside compute, power and data-centre capacity, rather than treated as a final-stage requirement.