What to notice is that this wasn't really a startup idea at first but it was someone noticing that commercial wireless keyboards had solved a problem that the DIY ecosystem had not; a number of successful products seem to be from bringing an existing capability into a community that need it.

AI infras is starting to run into the same land using and energy also as heavy industries : when the datacenters compete for grid capacity water and industrial land at scale the local governments just stop treating them like just advanced softwares

tokenmaxxing is becoming harder to justify could be a change in the labor market => when capital was free the companies optimized aggressively around retention and internal status spending but high rates + slow growth oblige firms to back toward productivity and operating leverage.

A big amount of user loyalty comes from products that feel finished more than always evolving. Early Kindles felt closer to bookshelves than software platforms

My Kindle Touch, judging by the plastic, is old, on its second battery. I love that the reading experience has not changed at all in the time I’ve had the device. I never thought of it as a bookshelf, but that’s a great way to put it.

In contrast, my iPhone changes with each update, but often I find not for the better – I hated the new control center at first, and while I made it mostly match the old one, the tap targets are smaller than they used to be, on my 4.7” display.

Nice to see that human intuition depends on periodicity when structures become unperiodic , classification becomes harder even if the rules under are deterministic

The market still prices the AI companies like software businesses with some huge margins while the economics are starting to look like cloud or infrastructure with big capex heavy competition and rapidly falling prices.

A lot of AI aspects probably don’t need to be permanent cloud services as local hardware improves part of the industry may change from renting intelligence to on-device computing.

One of the most dangerous words in engineering is “statisticaly impossible” At enough scale edge cases stop to be theoretical and start become production events.

The cost savings didn’t come from a single finding but from institutional alignment and standardization and political continuity. So seems that infra costs are often organizational problems as much as engineering ones

If the growth is increasingly led by investment (like AI, infrastructure) more than by the consumption then the cycle dynamic could be very different from the previous expansions