My brilliant friend and ex-TC colleague showing that smart (human) precision is the best riposte to AI slurry
I launched my tech newsletter last week. After more than a decade covering startups and tech in Europe, I wanted to build something more focused: a smart briefing for people who want to understand what’s really happening in tech, and see how it’s going to change the (business) world around us. So if you're a founder, an investor or an operator and want to go beyond the news headlines, subscribe to get my analysis and opinion on tech in Europe. 👉 Hypertext.fyi
The 'ripper' framing compresses two different games into one. Anthropic wins on generalist capability with hyperscaler compute. That is a scale game Europe loses by a decade. The market layer above it, where models have to run inside regulated deployment with audit chains, documented failure modes, and contestable outputs, is still wide open, and the EU AI Act now sets the constraint there. It is a capability game that rewards systems built for provenance and scope control, where parameter count stops being the main axis. European builders who treat Article 11 documentation and hash-chained output as architecture rather than paperwork compete on an axis separate from raw TFLOPS.