Steve Burke
speaker
449 appearances
1 recordings
1 series
first heard Jul 2026
last heard 15 Jul
Steve Burke’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 1 in all, peaking in Jul 2026 with 1.
Appearances
And so increasingly as they make more and more of this, presumably for NVIDIA's devices, you will continue to see consumers...
really get shafted if there's not new supply online soon and and likewise if they're focusing all their efforts and resources on hbm for whatever reason uh who's to say that the new fabs will even make consumer memory anyway right like there's they might just pick more hbm
Yeah, I mean, I guess it's good to have more capacity.
There's a lot of government money floating around, too, right?
Like, Micron got a $5 billion tax break in New York as part of the, I guess, Big Beautiful Bill.
Yeah, well, I mean, I do think personally, I think that they are colluding again.
I think they are price fixing.
It might be through like a parallelism rather than a literal behind closed doors, meaning like.
we're just going to announce our changes in earnings reports and then our friendly competitors will kind of mirror us and so yeah yeah like it's kind of a prisoner's dilemma like if they don't raise prices the other guy will if they and if they don't build capacity the other guy will they won't get the long-term deals right yeah and i mean also you know the
The wafers themselves, I have some interesting numbers I wrote down, but I was looking up, when you were asking about HBM, I went and pulled up some of the original coverage I did of HBM, which started to come online, or at least was being actively developed somewhere around, I don't know, 2013, plus or minus a bit, for AMD GPUs.
So yeah, when AMD was...
Working on Fiji and its Fury GPUs.
They debuted HBM actually for consumer devices.
They had HBM on Vega and on a couple other architectures in like the 2015 to maybe 2018 area, plus or minus a year or two.
And massively expensive, just like it is today.
The power efficiency was great, which is really why they used it, although they were bandwidth starved too.
But I pulled up, I had done some reporting at the time where I spoke to a bunch of the people in the know in the industry.
And at the time, 2017 or so, 2016 area, the HBM2 memory for one of the devices should have cost around $150 cost for them to buy.
And the interposer and the packaging should have been another $25, $175.
And this is old technology now.
Showing 201–220 of 449 · page 11 of 23
← Previous
Next →