Loading video...

Video Failed to Load

Go Home

BG2. AI Compute Landscape ‘25-26, scaling intelligence, chip competition, memory tech, inference time reasoning & more. 👊💥Brad Gerstner Bill Gurley feat Dylan Patel

470,786 views • 1 year ago •via X (Twitter)

11 Comments

Bg2 Pod's profile picture
Bg2 Pod1 year ago

Spotify: Apple: YouTube:

Jeanne Roland's profile picture
Jeanne Roland1 year ago

PRINCESS BRIDE x ROBIN HOOD & MULAN in 14thc Belgium!🔥🏹 Light-hearted Coming-of-age Escapism 🏹🌺💋 Immersive 💞💔💘 Romantic 🔥 Slow-burning pining 💯 Unique & Original Packed with plot, intrigue, twists & turns, gorgeous boy shenanigans, & LOTS of longbow archery trials!🏹 Escape with Marieke into the intrigue, adventure, & romance of the all-male archers' guild of Saint Sebastian🏹 #free on #kindleunlimited 💞🏹🌺💋🔥 #romance #yaromance #cleanromance #historical #fantasy #booktwt #books #BookWormSat #bookworms #epic #medieval #archery #adventure #lovestory #foundfamily #friendship #yalit

Elon Musk's profile picture
Elon Musk1 year ago

@altcap @bgurley @dylan522p Interesting

Josh Wolfe's profile picture
Josh Wolfe1 year ago

@altcap @bgurley @dylan522p Strongly agree on memory. SK Hynix was my long pick at recent hedge fund friends gathering

Russell's profile picture
Russell1 year ago

@altcap @bgurley @dylan522p This is the salient point of why Data Centers will be looking to upgrade. It is not just about TCO anymore, it's about power and recovering that power for other uses.

Igor Kurganov's profile picture
Igor Kurganov1 year ago

@altcap @bgurley @dylan522p Please add chapters in the future. Hard to watch on X without them. Fan of your podcast!

Dishwasher's profile picture
Dishwasher1 year ago

@altcap @bgurley @dylan522p 90 more minutes of DYLAN PATEL?! sign me up!

CJDGiesen's profile picture
CJDGiesen1 year ago

The reason they said Pre-Train scaling is dead is because they haven't actually been able to scale these training clusters until @xai and @elonmusk figured it out. We will see with Grok-3 and beyond if that's actually the case or maybe with high-quality synthetic Data and larger clusters to make sense of this data it's actually the opposite. It might even be that we can only really see both the quality AND the benefit of synthetic datasets with clusters beyond a certain size, i.e. large enough training compute to make sense of chaotic data.

Jeff Amazon's profile picture
Jeff Amazon1 year ago

@dylan522p @altcap @bgurley I went into the neocloud learning from this great podcast actually. @dylan522p big respect sir 🫶

Bernie Gartenlaub's profile picture
Bernie Gartenlaub1 year ago

@altcap @bgurley @dylan522p Dylan is brilliant, but he needs to cut down on caffeine. How do Brad & Bill stay so calm?

needleye's profile picture
needleye1 year ago

@altcap @bgurley @dylan522p The interviewer should refrain from interrupting Dylan when he has a train of thought that is so perceptive

Related Videos