2025: The year we rode the enterprise AI wave

With the start of the Christmas festivities and the year coming to a close, the team at Computer Weekly has been looking back on what the last 12 months has meant for corporate IT. What is absolutely clear is the unstoppable rise of artificial intelligence (AI).

In fact, just over 12 months ago, the world started hearing of China’s Deepseek, a large language model (LLM), that perfectly undermined the US’ AI ambitions. In what likely set the stage for the next phase in the evolution of AI, Deepseek’s R1 model demonstrated that it is certainly not necessary to throw vast amounts of computational resources and spend a huge amount of money on AI acceleration hardware, to achieve good results. And this is certainly very good news for budget- and resource-constrained enterprise IT departments tasked with “doing something with AI” because the boss has heard that every business is doing this.

But as the CTO of Thomson Reuters told Computer Weekly in August, businesses like Thomson Reuters are looking to transform their products with AI.

Such transformations require both changes in culture and technical infrastructure and over the last 12 months there have been plenty of developments bolstering the AI capabilities in datacentres. AI has fuelled an unprecedented surge in network demand, which means new long-haul networks to enable the rapid scaling of capacity needed to support datacentre AI workloads.

Datacentres that run AI workloads are power-hungry, which raises questions about how environmentally friendly they really are. As datacentre operators build out more and more AI datacentre capacity, research from datacentre resiliency think tank, the Uptime Institute, revealed this year that operators are seemingly becoming more negligent when it comes to collecting sustainability data about how their sites operate.

Against a backdrop of geopolitical unrest and concerns about the US government snooping, the UK’s reliance on overseas cloud providers came into sharper focus during 2025, prompting more thought and consideration from UK IT buyers about the topic of data sovereignty.

AI was the main focus at this Consumer Electronics Show as PC makers demonstrated their latest devices, which now offer a neural processing unit (NPU). AI is moving to the edge and this means end user computing is fast becoming AI-enabled.

Every enterprise software company is adding AI-capabilities to their product portfolio and, going forward, some of this functionality may well end up running at the edge, on an AI PC with an NPU. For now, the AI in enterprise software very much relies on machine learning in public clouds, but the shift to AI inference means these workloads can run on-premise.

The AI genie is soaring. 2026 is set to be a bumper year for AI agents. Happy Cjristmas.