Reid Hoffman

So I would hazard strongly that there's something like that in the background. Now, it could be that they had some access to chat GBT, certainly some of the data and evidence suggests that in terms of the way that it answers and does certain things. It could be that they actually had access to a compute cluster of of size because the so-called training run really makes sense.

And I cross check this across multiple groups, you know, like outside groups saying, hey, you know, what is it? What makes sense here? And they're like, yeah, for the final training run on a serious compute cluster, that could be the dollars that was that was spent on this in order to make it happen. Doesn't include talent, doesn't include all these other things.

And I cross check this across multiple groups, you know, like outside groups saying, hey, you know, what is it? What makes sense here? And they're like, yeah, for the final training run on a serious compute cluster, that could be the dollars that was that was spent on this in order to make it happen. Doesn't include talent, doesn't include all these other things.

And I cross check this across multiple groups, you know, like outside groups saying, hey, you know, what is it? What makes sense here? And they're like, yeah, for the final training run on a serious compute cluster, that could be the dollars that was that was spent on this in order to make it happen. Doesn't include talent, doesn't include all these other things.

Yeah, exactly. And I think that the... I think it's nearly certain that it's dependent upon the large-scale compute, the larger models in some way. The only question is we don't know in what ways and how. And I think that's one of the things that everyone's investigating. And so I think the kind of market frenzy on, oh my God, AI can be here without large-scale compute.

Yeah, exactly. And I think that the... I think it's nearly certain that it's dependent upon the large-scale compute, the larger models in some way. The only question is we don't know in what ways and how. And I think that's one of the things that everyone's investigating. And so I think the kind of market frenzy on, oh my God, AI can be here without large-scale compute.

Yeah, exactly. And I think that the... I think it's nearly certain that it's dependent upon the large-scale compute, the larger models in some way. The only question is we don't know in what ways and how. And I think that's one of the things that everyone's investigating. And so I think the kind of market frenzy on, oh my God, AI can be here without large-scale compute.

And of course, by the way, AI can be here. I invest in startups that train small models and so forth. But the large models still bring certain critical elements to the table in terms of ability to change small models, ability to get to performance. Like if you say, well, hey, moving from 10,000 GPUs to 100,000 GPUs, we only get a 20% better coder, medical assistant, legal assistant, tutor.

And of course, by the way, AI can be here. I invest in startups that train small models and so forth. But the large models still bring certain critical elements to the table in terms of ability to change small models, ability to get to performance. Like if you say, well, hey, moving from 10,000 GPUs to 100,000 GPUs, we only get a 20% better coder, medical assistant, legal assistant, tutor.

And of course, by the way, AI can be here. I invest in startups that train small models and so forth. But the large models still bring certain critical elements to the table in terms of ability to change small models, ability to get to performance. Like if you say, well, hey, moving from 10,000 GPUs to 100,000 GPUs, we only get a 20% better coder, medical assistant, legal assistant, tutor.

Like, well, in a wide variety of those areas, that really matters. That increase in cost when you amortize it across people accessing it across the entire unit, the billions of people that could use it, that's actually completely worth it and makes total economic sense.

Like, well, in a wide variety of those areas, that really matters. That increase in cost when you amortize it across people accessing it across the entire unit, the billions of people that could use it, that's actually completely worth it and makes total economic sense.

Like, well, in a wide variety of those areas, that really matters. That increase in cost when you amortize it across people accessing it across the entire unit, the billions of people that could use it, that's actually completely worth it and makes total economic sense.

Yeah. By the way, I think that all of these questions are in the classic short-termism versus long-termism. Because if you're saying, hey, I'm building this CapEx thing of $50 to $80 billion, A, I can train much better intelligence, but B, I'm also, these are kind of data centers in terms of serving intelligence through various apps, you know, to the world.

Yeah. By the way, I think that all of these questions are in the classic short-termism versus long-termism. Because if you're saying, hey, I'm building this CapEx thing of $50 to $80 billion, A, I can train much better intelligence, but B, I'm also, these are kind of data centers in terms of serving intelligence through various apps, you know, to the world.

Yeah. By the way, I think that all of these questions are in the classic short-termism versus long-termism. Because if you're saying, hey, I'm building this CapEx thing of $50 to $80 billion, A, I can train much better intelligence, but B, I'm also, these are kind of data centers in terms of serving intelligence through various apps, you know, to the world.

You know, you could say, well, the payoff is longer than I'd like as a public market. I'd like the payoff to be three years and maybe it'll be five years or seven years. That's the range you're talking about. I find that the general discussion on the X tens of billions of dollars to be short-sighted.

You know, you could say, well, the payoff is longer than I'd like as a public market. I'd like the payoff to be three years and maybe it'll be five years or seven years. That's the range you're talking about. I find that the general discussion on the X tens of billions of dollars to be short-sighted.

You know, you could say, well, the payoff is longer than I'd like as a public market. I'd like the payoff to be three years and maybe it'll be five years or seven years. That's the range you're talking about. I find that the general discussion on the X tens of billions of dollars to be short-sighted.

Generally, me as a private citizen, me as a venture investor, obviously making zero comment as a Microsoft board member. But the and so I think that it's an extremely important area to be investing in. And I'm actually glad that that we as a industry are doing this.