NVLink, every mention
35 scenes across 10 shows · ← back to NVLink
tap a year for its mentions
Latent Space 15
No Priors 9
BG2 Pod 8
the Y Combinator Startup Podcast 7
Invest Like the Best 5
Acquired 4
the MAD Podcast 3
the a16z Podcast 32 more shows
every year every show Latent Space 15 No Priors 9 BG2 Pod 8 the Y Combinator Startup Podcast 7 Invest Like the Best 5 Acquired 4 the MAD Podcast 3 the a16z Podcast 3 
All-In 2 
TBPN 2
Verbatim, from the transcripts: passages where NVLink comes up on Latent Space, No Priors, BG2 Pod, the Y Combinator Startup Podcast, Invest Like the Best
Ex-NVIDIA Engineer: Why AI Is About to Get 1000x Cheaper
- ▶ 20:55 Neil Movva With NVIDIA GPUs, one of the things that they really innovated on and did a great job with is the NVLink interconnect between GPUs. 5 times in the scene
Next 100x in AI: Inference, Networking, & Self-Optimizing Models — Philip Kiely & Ali Taha, Baseten
- ▶ 51:29 Philip Kiely Uh, tensor parallelism is not a good fit for local AI because it assumes a very high bandwidth interconnects like NVLink, uh, was, you know,
Multi-GPU Kernels, Intelligence per Watt, Heterogeneous Inference, and More | YC Paper Club · Y Combinator
- ▶ 9:43 Stuart (Stu) You also have asynchronous bulk device-initiated networking, which was not possible a few generations ago, and you also have a scale of architectures like NBL-L-Seventy-Two, which packs 72 GPUs inside a single NVLink domain, and soon this…
- ▶ 12:45 Stuart (Stu) It's relatively slow, so we have NVLink, which allows direct communication of data between the GPUs, and that's basically essentially, um, what NVIDIA DGX machines are all about. 3 times in the scene
- ▶ 14:28 Stuart (Stu) So it turns out that there are three ways to send data over NVLink. 3 times in the scene
Inside Nemotron & NVIDIA’s AI Lab | Bryan Catanzaro
- ▶ 46:05 Bryan Catanzaro And, uh, what it does is actually, um, reduces the amount of communication that has to be sent through NVLink during MOE computations by basically down projecting it.
The GPU Myth: State of AI Compute 2026 | Stephen Balaban
- ▶ 19:50 Stephen Balaban That's 72 GPUs all networked together via, um, NVLink.
Agent Inference at the "Speed of Light" — How NVIDIA moves like a $4.3 Trillion Startup
- ▶ 30:26 Kyle Kranen Let's say you're on an H 100, uh, the maximum NVLink domain, domain for H 100 for most DGX H 100 3 times in the scene
The Remarkable AGI Predictions of Daniel Gross
- ▶ 37:48 unnamed speaker And, and I was very confused because interconnect typically refers to sort of like the, the ethernet cables between servers, or like NVLink, or, ah, you know, the, what, what, what TPU is doing, ah, Ironwood between like the three Taurus…
NVIDIA’s Jensen Huang on Reasoning Models, Robotics, and Refuting the “AI Bubble” Narrative
- ▶ 42:55 Jensen Huang And, and I'm super proud that, that, um, uh, our latest innovation, MVLink, 72, we're the lowest cost
"Is there an AI bubble?” Gavin Baker and David George
- ▶ 24:40 Gavin Baker Fabric, which is a mixture of NVLink and either InfiniBand or Ethernet.
NVIDIA: OpenAI, Future of Compute, and the American Dream | BG2 w/ Bill Gurley and Brad Gerstner · Bg2 Pod
- ▶ 33:50 Jensen Huang Between Hopper and Blackwell, we increased, because of MVLink's 72, right, 30 x in one year. 2 times in the scene
Dylan Patel on the AI Chip Race - NVIDIA, Intel & the US Government vs. China
- ▶ 1:23:25 Dylan Patel Yeah, if you can run it for a bit, or just inference and take advantage of the huge NVLink, uh, NVL-CNII, you know, um, there, there's, there's ways you can, you could squint and say,… 2 times in the scene
🕰️ The Oral History of Windsurf (ft. Varun Mohan, Scott Wu, Jeff Wang, Kevin Hou, Anshul R)
- ▶ 45:55 Varun Mohan On top of that, within the machine, they've created this, this fabric called NVLink that allows you to communicate at ultra low latency.
Weekly Recap - Elon Vs Trump, Ukraine's Drone Attack, Cluely Update & OpenAI CRO
- ▶ 1:54:32 Sholto Douglas Um, so it used to be like eight GPUs in a pod, and then, uh, like you connect them over a worse, uh, interconnect.
No Priors Ep. 95 | Best of 2024
- ▶ 2:06 Jensen Huang Ethernet, InfiniBand, NVLink, no NVLink, you know what I'm saying?
AI Semiconductor Landscape feat. Dylan Patel | BG2 w/ Bill Gurley & Brad Gerstner · Bg2 Pod
- ▶ 9:14 Dylan Patel Um, and so what's interesting is that Nvidia has, has seen that and built an architecture that has many chips networked together really well called NVLink.
New SEC Chair, Bitcoin, xAI Supercomputer, UnitedHealth CEO murder, with Gavin Baker & Joe Lonsdale
- ▶ 50:35 Gavin Baker and I think even more importantly, NVLink, although a lot of Ethernet is, um, you know, never bet against the internet, never, never bet against Ethernet. 2 times in the scene
State of AI 2024: Frontier Models, AI Geopolitics, Robotics | Nathan Benaich, Air Street Capital
- ▶ 44:49 Nathan Benaich So this company is like shipping new products like twice a year on the chip side, not even including like the interconnects and then, and V-Link and other things.
No Priors Ep. 89 | With NVIDIA CEO Jensen Huang
- ▶ 4:04 Jensen Huang And so that, that, that's the reason why we bought Mellanox and started fusing InfiniBand and MVLink, um, in such an aggressive way. 3 times in the scene
- ▶ 10:46 Sarah Guo I think about it going from single chip to, you know, server to rack, and VL-CNII. 3 times in the scene
Ep18. Jensen Recap - Competitive Moat, X.AI, Smart Assistant | BG2 w/ Bill Gurley & Brad Gerstner · Bg2 Pod
- ▶ 15:48 Bill Gurley That's where NVLink thrives.
Ep17. Welcome Jensen Huang | BG2 w/ Bill Gurley & Brad Gerstner · Bg2 Pod
- ▶ 3:31 Jensen Huang Uh, inventing new numerical precisions, we did it by new architectures, inventing the tensor core, uh, the way systems are for, are formulated, NVLink, um, added, uh, insanely, insane, insanely fast memories, HBM, and, uh, uh, scaling… 2 times in the scene
- ▶ 19:36 Jensen Huang And what came out of it was MVLink. 2 times in the scene
No Priors Ep. 81 | With Sarah Guo & Elad Gil
[LLM Paper Club] Llama 3.1 Paper: The Llama Family of Models
- ▶ 20:11 Eugene Cheah Yeah, so I'll, I'll draw the distinction between pipeline parallelism and techniques like deep, deep speed tree, which is essentially where the GPU has, uh, uh, NV-link connectivity to other, other GPUs to actually read the, the model…
State of the Art: Training 70B LLMs on 10,000 H100 clusters
- ▶ 42:20 Jonathan Frankle You know, for, for both the GH 200 and the GB 200 is this hybrid networking where you'll have blocks of NV link networked chips. 2 times in the scene
Supervise the Process of AI Research — with Jungwon Byun and Andreas Stuhlmüller of Elicit
- ▶ 52:39 unnamed speaker I don't know if that's like the best name to, to think about it, but I was using one of these chat which are documents features, and I put the AMD and my 300 specs on the, um, new, you know, Blackwell chips from NVIDIA, and I was asking… 2 times in the scene
Open Source AI is AI we can Trust — with Soumith Chintala of Meta AI
- ▶ 7:13 Soumith Chintala Um, similarly, GPU has several levels of memory, and then you have, like, different levels of network hierarchies, NVLink plus, like, uh, InfiniBand or Rocky or something like that, right?
- ▶ 19:16 Soumith Chintala So, I'm, like, I'm, like, I'm sure the other hardware providers are working on it, but.
Nvidia Part III: The Dawn of the AI Era (2022-2023) (Audio) · Acquired
- ▶ 1:05:35 Ben Gilbert So now we make a box, and it's a fully integrated NVIDIA solution with our GPUs, our CPUs, our NV link between them, our InfiniBand to network it to other boxes, and, you know, welcome to the show.
- ▶ 1:49:49 Ben Gilbert Which I think they basically just crank down the NVLink's data transfer speeds, so it's like buying a top of the line A-Eight-Hundred, but not with as fast of data connections as you need, which basically makes it so you can't train large…
- ▶ 2:45:29 Ben Gilbert You'd, of course, then need to build up the chip-to-chip networking capabilities, like NVLink, that very few have.
Ep 18: Petaflops to the People — with George Hotz of tinycorp
- ▶ 41:53 George Hotz Um, so, the bandwidth is the, is roughly 10 X less than what you can get with NV-linked A-Hundreds. 3 times in the scene
Nvidia: The Machine Learning Company (2006-2022) · Acquired
- ▶ 1:11:25 Ben Gilbert So at this point they've got NVLink, which is their, it's like the, what does Apple call it?