Eduard

Software engineer & amateur cyclist.

[Eduard @EduardHL]

Many investors view AI mainly as a way to make existing software faster and cheaper, but that misses the more important point: AI unlocks entirely new capabilities.

Take software for dentists as an example (VMS). Switching to a cheaper product with identical functionality would not be worth the hassle. But an AI-native system that can answer calls, book appointments, chase unpaid bills, and handle routine patient communication could eliminate the need for a secretary. This could be enough to justify switching.

Incumbents still have an advantage and could retain customers if they move quickly. But strong moats often breed complacency, making them slower to react. Beyond new features, AI can also reduce switching costs by making migrations easier and less error-prone. The best software engineers usually don't work on ERPs or VMS, so early versions are buggy and painful to adopt. With modern AI coding agents, initial software quality should be meaningfully higher.

There was a clear step change in AI coding agents in December 2025, which likely explains the recent selloff in many software stocks. The financial impact on incumbents could take a few quarters to show up, just as it took years for Kodak to fail after the digital camera appeared.

Some investors with strong judgment will identify incumbents whose moats are unlikely to be eroded and earn exceptional returns, but this seems far from straightforward. Owning data that can be used to unlock new AI functionality should be an advantage for incumbents.

By mid-2025, ASML was reasonably priced, a clear AI beneficiary, and the key risks (like China developing EUV) seemed unlikely in the short to medium term. This situation feels fundamentally different.

[Eduard @EduardHL]

@rogerfields120 Well done!

[Eduard @EduardHL]

@vkhosla @elonmusk A community notes scientific town square, where people like @friedberg or @cremieuxrecueil influence government policies on things like food safety or vaccines would be great. Public discussions would create trust and consensus/collaboration should improve outcomes.

[Eduard @EduardHL]

@BillAckman In "From Third World to First" Lee Kuan Yew explains how they made Singapore a cohesive multicultural society, no easy feat. If a country wants to be more multicultural (Singapore was from the start), it should plan for success. Only opening borders might be wishful thinking.

[Eduard @EduardHL]

@michaelxpettis Why not make income transfers to couples who have babies (to address the low fertility rate)?

[Eduard @EduardHL]

@paulg Singapore improved underperformance of Malay students in science and mathematics by creating community-based self help groups, which included extra tuition in the evenings (From Third World to First, pages 210-212). Inspiring story, especially in comparison to removing algebra.

[Eduard @EduardHL]

@cyclingalps It looks like you both already enjoyed beautiful sceneries (which are nice to explore on a bike). Happy anniversary!

[Eduard @EduardHL]

@Noahpinion The Price of Time by Edward Chancellor

[Eduard @EduardHL]

@InternetH0F I would use some of the money to build it a nice habitat enclosed in glass (like in zoos). I would keep it home as a reminder that death can come at any moment and that time is precious.

[Eduard @EduardHL]

@yishan Very insightful thread, thanks. Option 1 (written submissions) with more than 2 people with different views on the topic agreeing on a set of claims could be most conducive to truth. @CommunityNotes with an original tweet being debated, edited and agreed by the whole community.

[Eduard @EduardHL]

@mulagostarr Your tweets appear more on my « For you » feed post elon twitter. I’m grateful for that. Some activity on linked-in might be people showing « professional engagement ». Views might be a better metric on twitter as many (me included) rarely like/repost. Hopefully you don’t leave.

[Eduard @EduardHL]

@cyclingalps I cycled all these except Cleuson (I now have thicker tires for the last gravel stretch 😉). I wouldn't have discovered all these climbs so easily without your blog.

[Eduard @EduardHL]

@paulg Given that definition, elegance is exposed in the book « Deep Simplicity: Bringing Order to Chaos and Complexity », recommended by Charlie Munger.

[Eduard @EduardHL]

@stats_feed How did you calculate Switzerland? The canton of Zürich is not the same as the city of Zürich.

[Eduard @EduardHL]

@yishan The book Big Debt Crisis by Ray Dalio may help better understand these things https://t.co/ZRpJCOEdQU

[Eduard @EduardHL]

@Rainmaker1973 Looks fun!

[Eduard @EduardHL]

@BillAckman @SRuhle Thanks for speaking your mind. I believe many people (who seldom tweet) really appreciate it. Paul Graham recommends to ignore the haters https://t.co/aNn6WG18u5

[Eduard @EduardHL]

@MartinVetterli @gvdgoot I also know highly talented males who have abandoned the academic route because the risk/reward looked poor. If this is representative, scientific progress might suffer. Finding other ways to improve diversity, even if it takes longer than 2032, could be worth considering. 2/2

[Eduard @EduardHL]

@MartinVetterli @gvdgoot Diversity is a desirable outcome, no doubt. Trying to achieve it by setting targets on female hires might produce undesirable second order effects. I heard female postdocs wanting to pursue an academic career say « I don’t need to put as much effort as my male counterparts » 1/2

[Eduard @EduardHL]

@garrytan Zurich had an open drug scene that was eliminated in the 90s with the introduction of the four-pillar policy (prevention, therapy, harm reduction and enforcement)

https://t.co/nfKWgWnAEf

https://t.co/fn0KRyA2bl