lukas221

we will, but data centers will compete for that hardware with retail consumers

you allocate tokens from top down - first exclusivity deals - Citadel pays $10 bil to get exclusivity access to GPT-6 for 3 months before anyone else, then you price it $1000/mil, then whatever compute is not used you sell GPT-5.9 at $500/mil...

the moat always has been and will be compute

and we are fast approaching limits which will be hard to overcome - electricity, chips

it's simple, how much dollars you get out for every dollar put into tokens

as Jensen said, get ready for $1000 per mil token

those for which this price makes sense will push out those for which it doesn't - to lower models or to local models

but those who want to run local models need to compete for hardware with the data centers, which have strong scale effects thus will always be able to out price local hardware allocations - can already be seen now as hardware makers get out of retail business

people mostly use RDBMS because they want the advanced querying that SQL provides, not because of the isolation levels

but ignoring that, serializable isolation level means the database acts AS IF the transactions are serial. but most databases in fact will execute them concurently, with careful tracking to make sure they appear serial

the tragedy of allowing stupid people to vote.

pick one:

- stupid people vote without understanding what they vote for

- stupid people don't vote, but it's not a democracy anymore

before mass media we had the priests and the Church which decided what is truth and what is not.

I would argue that not using serialization isolation level by default is like not using a memory safe programming language by default.

Sure, sometimes it's too slow, but it should be the default.

Very few people can write correct database code at the other serialization levels. Most think they can, but it's harder than correct multi-threading, because databases do weird unintuitive things for performance.