and you don’t want to use automation
megascale makes that impossible, bc organizational complexity doesn’t increase linearly 1:1 with each new staffer, it increases geometrically
if you have 500 million users and 700,000 are hit by a terrible edge case, that’s still only 0.14% of your user base and it doesn’t “make sense” to put resources into it
staffing for 500m users is also basically impossible
and you don’t want to use automation
megascale makes that impossible, bc organizational complexity doesn’t increase linearly 1:1 with each new staffer, it increases geometrically
adsei.org/2026/02/18/p...
adsei.org/2026/02/18/p...
One pushback: it feels like a overreading that consumer AI will vanish post crash - what will swap in are on device, open source/weight models. This is already happening.
What will die is any form of megascale training, and, as you say, the hype driving uncritical adoption.
One pushback: it feels like a overreading that consumer AI will vanish post crash - what will swap in are on device, open source/weight models. This is already happening.
What will die is any form of megascale training, and, as you say, the hype driving uncritical adoption.