Chapter 2
Network infrastructure and AI
Powering up. Staying resilient.
AI workloads require more storage, power, compute, and network throughput. On top of this, successful AIOps require integration across systems, clean and structured data, ongoing optimisation, and strong governance.
Dealing with legacy systems
44% of NHS digital leaders say incompatibility with current legacy systems and infrastructure is a big worry. (5)
AI capabilities are relatively easy to demonstrate. They’re much harder to operate. You may need to reassess:
- Integration across systems.
- Clean and structured data.
- Ongoing tuning and optimisation.
- Governance and oversight.
- On-prem versus cloud strategies.
- Bandwidth and latency constraints.
- Segmentation and security models.
(5) Block (2025)
The hidden challenge:
Do you have enough team capability to get your infrastructure AI-ready?
For stretched network teams, this can be an unwanted extra pressure.
That’s why many Trusts are starting to ask a different question:
“Do we want to build and manage this ourselves, or consume it as a capability?”
There isn’t a single right answer.
But it’s a decision that needs to be made early.