probably an important question rn: is the late 2020s gonna be constrained on chip production (fabrication + packaging) or chip deployment (datacenter + power)?
i hear a different take on this every day and i just want a clear answer. any good pieces i can read?
Users express optimism that AI chips will become extremely cheap soon, shifting the next constraint to bandwidth rather than late-2020s production or deployment shortages.
you guys are just proving my point. if i started a poll it would be stuck at 50%.
probably an important question rn: is the late 2020s gonna be constrained on chip production (fabrication + packaging) or chip deployment (datacenter + power)?
i hear a different take on this every day and i just want a clear answer. any good pieces i can read?
@khoomeik neither one will be the sole limiting factor and the world is not (yet) efficient enough to exploit one to the extent that all efforts go to solving the other
@khoomeik The constraint shifts every 18 months and whoever predicted the last one wrong is currently writing the definitive piece on the next one.
@khoomeik Easily the former. Power is so overrated as a bottleneck its not even funny
@big_algocracy yeah i guess price incentives also sorta force them to scale in lockstep or risk a supply glut
@khoomeik In the us it’s deployment, everywhere else on earth it’s supply
@khoomeik I would ask @gilbert_jc for his opinion on this, his operating on the field so he probably has interesting takes
@khoomeik sounds like almost certainly the former. look at it from first principles, which is easier to accomplish through brute force
@TattedWorks lol
@khoomeik The bottleneck keeps shifting from fabs to datacenters lately.
@khoomeik @insane_analyst @vikramskr
@khoomeik all of it
@khoomeik also curious for posts on this. my gut tells me that deployment constraints require less cutting edge tech and regulatory hurdles will clear up as everyone becomes more AGI pilled, making production the bottleneck
@khoomeik There won’t be a chip constraint in 3 years. That’s timeline for building new chips and standing up new fab capacity.
Also, in shortages, there is double ordering for chips and that self-corrects.
Can tell more in DMs
@khoomeik Deployment. Capital solves fab capacity. Nothing shortcuts power plants and permitting. Years of lead time in most markets. Gulf builds accelerating for exactly this reason.
@khoomeik Chips are going be extremely cheap and the next constraint may be bandwidth
@khoomeik !!!
probably an important question rn: is the late 2020s gonna be constrained on chip production (fabrication + packaging) or chip deployment (datacenter + power)?
i hear a different take on this every day and i just want a clear answer. any good pieces i can read?