High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / belief

Published · transcript-backed

Alessio Fanelli: belief

6 Apr 2024 Latent Space Latent Space Chats: NLW (Four Wars, GPT5), Josh Albrecht/Ali Rohde (TNAI), Dylan Patel/Semianalysis (Groq), Milind Naphade (Nvidia GTC), Personal AI (ft. Harrison Chase — LangFriend/LangMem)

“I think when you do a new Silicon, you're like, Oh, we're going to be so much better at this thing or like much faster, much cheaper.”

— Alessio Fanelli

Source trail

Everything needed to verify it.

Speaker
Alessio Fanelli
Attribution
Verified speaker
Claim type
belief
Recorded
6 Apr 2024
Publisher
Latent Space

Transcript context

…three Answer the question! What do you think of them? I think, kind of, there's a couple things. It's like How do the other companies innovate against them? I think when you do a new Silicon, you're like, Oh, we're going to be so much better at this thing or like much faster, much cheaper. But there's all the other curves going down on the macro environment at the same time. So if it takes you like five years before you were like a lot better, five years later, once you take the chip out, you're only comparing yourself to the five year advancement that the major companies had to. So then it's like, okay, the, we're going to have like the C300, whatever, from, from NVIDIA. By the time some of these chips come up. What's after Z? What do you think is after Z in the road map? Because it's X, Y, Z, Anyways Yeah, yeah, it's like the age old problem, right? Like you build a chip, it has some cool thing, cool feature, and then like, a year later, NVIDIA has it in hardware, right? Has implemented some flavor of that in hardware. Or two generations out, right? Like, what idea are you going to have that NVIDIA can't implement, is like, really the question. It's like, you have to be fundamentally different in some way that holds through for, you know, four or five years, right? That's kind of the big issue. But, you know, like, those people have some ideas that are interesting, and yeah, maybe it'll work out, right? But it's going to be hard to fight NVIDIA, who one, doesn't consider them competition, right? They're worried about, like, Google and Amazon's chip. Right, they're not, and I guess to some extent AMD's chip, but like they're not really worried about you know, MADX or Etched or Grok or, you know, Positron or any of these folks.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.

Search evidence