Tech Chat – September 2025
L-AI-ering up the use cases
Having watched events unfold from close quarters, my flatmate is all too keen to remind me of Downing Street’s BAFTA winning stagecraft: adding a series of already planned multi-million-pound commercial decisions, announce them at an unprecedented diplomatic event of soft power, and link pageantry and praise to a tangible economic benefit.
Timing is everything and the £31bn Tech Prosperity Deal MOU, including £22bn for AI infrastructure from Microsoft and £500m from NVIDIA, on top of £44bn previously committed, is certainly nice PR ahead of November, and of course, a much-needed injection to capitalise on our goldilocks circumstance (Huang’s words, not ours).
After many evenings arguing with my flatmate about whether AI is another bubble, what is clear is four-fold: 1) this is a step change in technology 2) there is a growing trend in ‘datafication’ of information 3) data is being used in more and increasingly novel applications 4) it is enabling (and is enabled by) investments in compute power and connectivity, of which a large proportion is more recently specifically focused on supporting AI uptake.
But then again perhaps similar arguments were made in 2000, long before I set foot in the City. I was three when my research colleague wrote a note on VoIP and its application, a breakthrough which was widely touted in the media as signalling the death of large telcos. The note instead suggested VoIP (voice over IP, carrying voice more efficiently over data protocols) would mostly benefit the telcos themselves, delivering massive efficiencies to their backhaul, including the opportunity for new products and solutions.
It is the same with AI: it will create new solutions through generative AI, but will most clearly deliver efficiencies to existing companies in doing what they do, but better, through more efficient processes. In the end the main outcome, from his point of view, was the end of the “telco” analyst, who mostly became tech analysts.
Stepping back, we see the core offering of AI is automation, optimisation, and augmentation. In context of the looming November Budget, this tech should theoretically help tackle the rising employment costs faced by enterprises of all sizes, which has then been compounded by recent NIC increases, reflected in the chart below from Next’s interim results. It seems the constrained supply side potentially has a way out with AI, and here, ActiveOps offers pure exposure to AI-driven workforce capacity optimisation for service operations hit by increasing employment costs.
The problems of rising wage costs are compounded with rising compliance costs. This is particularly important in regulated sectors like finance and insurance where service operations face intensified regulatory enforcement from the FCA and other bodies. Dear readers of Tech Chat know we love regulation.
The incentive for enterprises of all sizes is therefore to manage cost per employee via increasing productivity, and/or reducing the working time required to comply with non-discretionary tasks, like catching up on CPD hours, or maintaining conflict of interest or hospitality compliance registers. Conveniently, Skillcast does exactly this, and its stellar high margin performance and consistency surely justifies a premium.
More broadly there appears to be positive structural signals in complex data management and AI/ML applications across the whole ecosystem, including national security & defence (SRT Marine Systems), the enterprise back office (ActiveOps, Aptitude Software, GetBusy, Skillcast), infrastructure (Cordel, Eleco, Journeo, 1Spatial), commerce (EagleEye, PCI Pal) and more efficient marketing spend and monitoring (Celebrus, dotdigital, FADEL, Pulsar). The use case certainly spans more than prompting ChatGPT to write a strongly worded letter to your landlord or insurer, and the cost benefits to the enterprise and government are starting to show.
B2B/B2G AI looks like the best bet, tbh.
The ongoing debate rages as to whether AI mania remains a hype, or if there is something more substantial to result from the trillions invested in AI models and infrastructure by MANGO (the tech companies, not retailer). There is no question AI demand is pushing hardware to its limits. It has shifted the Overton window, enabling trillions of pounds of investment in new nano and quantum technologies and AI-specific infrastructure (almost) without question… and helps us stomach an audacious state visit.
Happy Friday
