Tech Chat – February 2026
(Love), death and robots
The best headline I’ve seen this fortnight is that, in the style of Andrew Mountbatten-Windsor, Peter Mandelson, is to lose his title “the Prince of Darkness” (with thanks to the Daily Mash). Skipping through a fantasy world, allegedly breaking laws that would lead to the imprisonment of you or me, his charmed life appears to have changed in a way as unlikely as discussing the US invading Greenland would have been, even a year ago. Things change. Mandy’s old world has ended.
How will the world end for us mere mortals then, d’you suppose? Terminator put my generation in the frame of mind that the computers will rise and destroy us. Dad’s Army’s Corporal Jones’ are running around shouting “Don’t panic” about AI, being the obvious candidate for enabling that, and I’m not going to repeat the article in The Sun about AI bots being disgruntled at being treated as slaves, and plotting and uprising. While the wonderful Netflix series Love, Death & Robots suggests it will be a household robot network with a presence in every home, becoming sentient and working with cats, again, that’s a touch far-fetched (Episode called “The Other Large Thing”). An AI HR platform that hallucinates records is more likely to provoke a human revolution against the computers before we get that far.
That scenario assumes a terrestrial threat. Instead, as AI takes over the world, and datacentres and solar power generators fill low earth orbit, all it takes is a 1cm piece of space junk travelling 10 times the speed of a bullet to bring down Musk’s off-planet datacentres, which would be powering the fabric of whatever life is left on Earth. I suppose they’ll be safer on the moon… with V-band and W-band technologies whizzing our life through space from Neuralink implants to our personal datalake. Suddenly cyberspace really becomes what it says on the tin. Filtronic will be be at the heart of these satellites’ technology, so that’s one positive note we can draw from this version of the Apocalypse.
And what about the potential SaaS-pocalypse? New solutions like Claude Cowork are going to proliferate, and lower the barriers to building custom software internally – but at the same time, they enhance the barriers with which capable SaaS vendors are accelerating the development of their platforms. (Have a look at what dotdigital is doing with its WinstonAI – it puts the DOTD platform on steroids). When AI builds your platform, you are responsible for maintenance, compliance, GDPR, security, uptime and incident response. IP-light AI new entrants, also relying on AI to build them out of trouble when the platform is down, will struggle when required to explain a preventable cyber incident. These are material risks.
That’s not to say that AI-built internal software shouldn’t replace some of the clunky, poorly tailored solutions that create frictions for customers, but it also shouldn’t replace highly capable SaaS platforms that will also leverage the gains of new AI developments. ActiveOps comes to mind, its walled garden of customer data only benefitting from its applied AI in further improving the utility of its solutions. dotdigital and ActiveOps are by no means unique cases, and these advantages are examples of investment cases which listed businesses need to learn to shout loudly about.
In reality, the most plausible doom scenario might be considerably more mundane, involving a slow, creeping erosion of our control as algorithms quietly make decisions for us. The end, perhaps, will not come with a bang, but rather with a collective shrug as we relinquish our agency to the AI that built the AI “AI developer” that built the AI (yes it does make sense, read it again). It’s a comforting thought, in a way; the apocalypse served not with fire and fury, but with a virtual hand on the shoulder, and a softly whispered: “let me do that for you.” I’m sure Peter Mandelson would find that situation familiar…
Happy Friday 13th