Can today’s internet keep up with AI?

AI is data hungry

Each AI prompt sends data from your device to cloud servers for processing in a data centre before returning the result via the internet. Compared with typical browsing, these interactions generate far higher data volumes and require much more bandwidth when scaled across an organisation.

This surge is reflected in Omdia’s data, which shows AI‑driven internet traffic doubled in 2025 and is set to become the largest source of global internet activity by 2031.

As AI competes for bandwidth, everyday tools can start to falter: video calls stutter, shared platforms slow and small delays compound. And it’s not just bandwidth – AI is highly sensitive to latency, so inconsistent connections trigger lag, freezing and dropouts, all of which erode productivity.

Is UK real estate ready for AI‑level connectivity?

Most buildings connect to the internet in two ways:

• FTTC: Fibre reaches the street cabinet but the final stretch uses copper. Copper slows uploads and increases latency – a poor fit for cloud‑driven AI.
• FTTP: Full fibre into the building – far higher capacity and much lower latency, giving the stability AI workloads rely on.

The UK has made progress on full‑fibre availability, but adoption still lags. As of 2025, Ofcom reported that 78% of premises (residential and SMEs) can access FTTP, yet only 42% actually use it – around 13.1 million properties.

No equivalent adoption figure exists for commercial real estate, but adoption is widely understood to be lower. A 2025 government study highlighted how organisational inertia and legacy systems slow technology upgrades in UK businesses, including essential digital infrastructure. As a result, many offices remain on outdated connectivity even when faster fibre services are available.

Why does full fibre adoption stall?

Even where full fibre is available, switching is often delayed because of:

• older buildings with limited riser capacity
• wayleave delays
• upgrades deferred until the next refurbishment cycle
• upgrades confined to common areas, leaving tenants to arrange FTTP themselves
• occupiers assuming current speeds are "good enough"

The result is a widening digital gap: companies adopting AI, but buildings remain equipped for basic internet use.

What makes an AI-ready building?

In WiredScore’s latest “Resilience gets real” report, AI‑ready buildings need resilient digital, physical and cyber foundations – not just fast broadband.

Robust digital connectivity

High bandwidth, low‑latency networks across the whole building plus dependable indoor mobile coverage, especially as 2G/3G networks retire.

Physical resilience

AI increases reliance on uninterrupted power. Many buildings still lack adequate backup solutions. AI‑ready assets need resilient electrical systems, diverse feeds and protection against outages and flooding risks.

Cyber resilience

AI expands the attack surface. With HVAC, access control and lifts now network‑based, buildings need segmented networks, controlled vendor access and continuous monitoring.

To explore this in more detail, read my interview with Sanjaya Ranasinghe, VP of R&D at WiredScore.

What this means for occupiers

As AI becomes embedded into day-to-day business operations, occupiers may increasingly find that some buildings are not equipped to support evolving digital requirements.

In multi-let offices particularly, there can be a mismatch between what occupiers need and what landlords are able to provide. Even where occupiers want higher-performance connectivity, upgrades may be constrained by legacy infrastructure, limited riser capacity or landlord-controlled systems.

Over time, digital resilience could become a more important factor in relocation decisions alongside location, amenities and sustainability, particularly as occupiers look for buildings capable of supporting future technology adoption.

What this means for landlords and developers

For landlords and developers, digital infrastructure is becoming part of the core competitiveness of a building rather than a secondary technical consideration.

As occupiers become more reliant on AI-enabled workflows, expectations around full fibre connectivity, resilience and network performance are likely to increase.

Over time, this could widen the gap between newer and older stock, with digitally resilient buildings potentially better placed to attract and retain occupiers, while less adaptable assets risk drifting towards functional obsolescence.

For now, today’s office internet may still feel adequate, but as AI adoption grows, the limitations of existing building infrastructure could become increasingly difficult to ignore.