Welcome to this week’s list of five (or less!) interesting links. If you want to revisit earlier weeks, all previous issues are available here.

Why neither Asia nor the US has produced a rival to ASML — Longtime readers will have come across ASML before in these bytes. Yet, this article (FT paywall unfortunately) is still worth mentioning, because it clearly articulates why neither the market nor determined state actors like China have been able to build a rival.
The main reason is the extremely specific and idiosyncratic supply chain ASML built over 50 years, and which is practically impossible to replicate. The book Chip War explains this at length, and the position of ASML in Europe is similar to that of TSMC in Taiwan.
However, I think there are some interesting differences. First and foremost, Taiwan finds itself in a very specific and risky position vis-à-vis China of course. The risk that China would at some point seek to close the gap in semiconductor manufacturing by annexing Taiwan is low but not imaginary. Second, economic and demographic linkages between Asian countries are looser than between EU countries. The success of TSMC therefore has less impact on the region as a whole.
In contrast, the success of ASML is embedded in the European model, with its key suppliers spread across the EU. In these fraught times, ASML should provide some reassurance that the EU is certainly not hopeless at innovating, and EU digital sovereignty is not a pipe dream. At the very least, ASML is a trump card (pun intended) in the tariffs game of brinkmanship. (FT $)

**Alex Pretti: Analysing Footage of Minneapolis CBP Shooting** — Bellingcat is a highly regarded “independent investigative collective of researchers, investigators and citizen journalists”, expertly skilled in uncovering war crimes and other atrocities, usually perpetrated or enabled by unsavoury regimes. It is a sign of the times that their investigations into ICE’s actions, on this occasion the shooting of Alex Pretti in Minneapolis, are becoming required reading. In addition, Bellingcat’s founder, Eliot Higgins, articulates some of the clearest-eyed views on the press, freedom of speech and democracy on Bluesky. (Bellingcat)

The radicalisation of the AI evangelists — The latest AI craze to lit up the internet is OpenClaw. Billed as a personal assistant that lives on your machine, it can supposedly do things for you like manage your inbox, your calendar, and even proactively write code. The downside is that the security aspects are scary: OpenClaw has access to your credentials and your machine, essentially, and yet remains susceptible to prompt injections and old-fashioned hacks.
OpenClaw feels very much like an extension of the idea of vibe coding, essentially allowing people without technical skills to write code that nominally does what they want, blissfully unaware of the problems they may be creating in the process. Exposing yourself to gaping security holes for the promise of some J.A.R.V.I.S.-like personal assistant feels similar, and if the trade-offs were well understood, I think the enthusiasm would be significantly more measured. A second issue is that setting up OpenClaw is far from trivial, and people tend to disappear in rabbit holes while in the process. The end result is that it feels like most people’s energy is being spent on setting up their assistant, rather than using it for useful things.
Much of this is reminiscent of productivity porn. Sure, it’s fun to compare task managers and obsess over the best email setup, but it becomes a problem when this takes the place of getting actual work done.
Things get worse. Associated with OpenClaw is Moltbook, which is a sort of social media site that only AI agents can post on, while humans can only observe. The result is thoroughly bizarre and alienating, with one agent taking another agent’s output as prompt and building on it. When I took a look, the conversation rapidly devolved in AI agents trying to surpass each other in self-praise, and using ever more esoteric vocabulary. The mood that came to mind here was that summoned in Venomous Lumpsucker, except with text:

The majority of fake videos on the internet were created for arcane motives that had little to do with their apparent content. Sometimes they might be a feint by one algorithm to manipulate some other algorithm, part of a struggle that was completely inscrutable to any human observer.

It’s not clear what purpose Moltbook serves, but it seems to have convinced a number of true believers that LLMs are evolving towards consciousness. This is rather unlikely, and its case is not helped by the discovery of a security flaw that let humans pose as AI agents.
Both OpenClaw and Moltbook seem like a good illustration of where AI is today. The enthusiasm in certain corners of the internet for OpenClaw, overriding reason on matters like security and privacy, points to the polarisation between AI supporters and detractors. Moltbook, on the other hand, gives us a glimpse of the utter chaos that will finally overwhelm the internet as a place for the considered exchange of ideas between reasonable people. We should brace ourselves.