What's coming down the pipeline for DevOps in 2025

Thanks to the power of localised models, generative AI will allow us to graduate from anecdotal and loosely organised thinking to using the entire history of an organisation to make good decisions about infrastructure and SRE.

ML has long been an established tool in the monitoring toolbox. However, we have yet to realise the potential of combining ML and generative AI to add understanding and context to monitoring data, let alone automatically fix or suggest fixes when problems are detected.

While Platform Engineering forms the backbone of Developer Experience, Internal Developer Portals act as the central nervous system–connecting developers to the so-called platform tools and best practices, fostering seamless collaboration, visibility, and control across the engineering ecosystem.

Regulation seems to be ever-tightening, and the organisations that end up doing this well are those that add suitable automation in order not to drown under the weight of increasingly thorough auditing. Writing 'regulation as code' is a discipline we expect to grow in 2025.

Part of the feedback loop of building new infrastructure involves trying to estimate capacity. But in environments where this is predictable, it can make more sense to repatriate back to running on your own servers. This doesn't necessarily mean running a data centre in the basement again, but it does mean that you need staff who understand technologies, such as virtualisation and Linux again.

Using building blocks to reduce the amount of code you need to write has been on the rise for a few years now – but the rise of AI also brings its own unique vector. AI is already extensively used to write boilerplate code outside of this context, and next year, I expect to see low/no-code platforms using this in the background.