Alex Wagner

And what deep seek proved was that you can do this with inferior chips and you can do this with a smaller team and you can do this much more efficiently. And I think, you know, that was going to happen one way or the other. So like the grand, like conspiracy theory thinking of like, you know, this is a Trojan horse that China has thrown into the United States to destroy our industry.

And what deep seek proved was that you can do this with inferior chips and you can do this with a smaller team and you can do this much more efficiently. And I think, you know, that was going to happen one way or the other. So like the grand, like conspiracy theory thinking of like, you know, this is a Trojan horse that China has thrown into the United States to destroy our industry.

Doesn't really hold water to me. It's what everybody in the U S was aiming toward.

Doesn't really hold water to me. It's what everybody in the U S was aiming toward.

Doesn't really hold water to me. It's what everybody in the U S was aiming toward.

Yeah, I mean, isn't it amazing that Stargate and DeepSeek R1 come out like basically at the same time? And it was, I mean, I wrote about this in my newsletter last week. It was just like, wow, like these two things seem like they're barreling towards each other and something has to give. So, okay, so why didn't the US firms develop this? Couple reasons.

Yeah, I mean, isn't it amazing that Stargate and DeepSeek R1 come out like basically at the same time? And it was, I mean, I wrote about this in my newsletter last week. It was just like, wow, like these two things seem like they're barreling towards each other and something has to give. So, okay, so why didn't the US firms develop this? Couple reasons.

Yeah, I mean, isn't it amazing that Stargate and DeepSeek R1 come out like basically at the same time? And it was, I mean, I wrote about this in my newsletter last week. It was just like, wow, like these two things seem like they're barreling towards each other and something has to give. So, okay, so why didn't the US firms develop this? Couple reasons.

First of all, it could just be a natural resource curse. Honestly, without the constraints, they didn't have to think about this way. Now, their models have become much more efficient over time. And that makes a big difference because they've been, again, heading this direction. It just hasn't been an imperative toward them.

First of all, it could just be a natural resource curse. Honestly, without the constraints, they didn't have to think about this way. Now, their models have become much more efficient over time. And that makes a big difference because they've been, again, heading this direction. It just hasn't been an imperative toward them.

First of all, it could just be a natural resource curse. Honestly, without the constraints, they didn't have to think about this way. Now, their models have become much more efficient over time. And that makes a big difference because they've been, again, heading this direction. It just hasn't been an imperative toward them.

So they've been making the models more efficient, but it hasn't been the only way they can do things. And so therefore, they haven't been forced to innovate this way. I think if they had similar constraints as the DeepSea team did, they probably would have come up with it. So that's one. But the other side of it, and I think this is what they would all tell you, is that they still believe that

So they've been making the models more efficient, but it hasn't been the only way they can do things. And so therefore, they haven't been forced to innovate this way. I think if they had similar constraints as the DeepSea team did, they probably would have come up with it. So that's one. But the other side of it, and I think this is what they would all tell you, is that they still believe that

So they've been making the models more efficient, but it hasn't been the only way they can do things. And so therefore, they haven't been forced to innovate this way. I think if they had similar constraints as the DeepSea team did, they probably would have come up with it. So that's one. But the other side of it, and I think this is what they would all tell you, is that they still believe that

scaling up like Elon Musk is putting together a million GPU data center. Mark Zuckerberg says he wants to build a data center that's like the size of half of Manhattan. We already know that Altman's been out there with Trump talking about this $500 billion initiative, which might be like a fifth of the size in the end, but it's still big.

scaling up like Elon Musk is putting together a million GPU data center. Mark Zuckerberg says he wants to build a data center that's like the size of half of Manhattan. We already know that Altman's been out there with Trump talking about this $500 billion initiative, which might be like a fifth of the size in the end, but it's still big.

scaling up like Elon Musk is putting together a million GPU data center. Mark Zuckerberg says he wants to build a data center that's like the size of half of Manhattan. We already know that Altman's been out there with Trump talking about this $500 billion initiative, which might be like a fifth of the size in the end, but it's still big.

And I think there's still a lot of belief in Silicon Valley that Bigger is still better, and you can build better models if you throw all these resources toward it. The most optimistic case is that they will take the innovations that they're seeing with DeepSeek and then they will use that efficiency to make even more use out of the architecture that they have.

And I think there's still a lot of belief in Silicon Valley that Bigger is still better, and you can build better models if you throw all these resources toward it. The most optimistic case is that they will take the innovations that they're seeing with DeepSeek and then they will use that efficiency to make even more use out of the architecture that they have.

And I think there's still a lot of belief in Silicon Valley that Bigger is still better, and you can build better models if you throw all these resources toward it. The most optimistic case is that they will take the innovations that they're seeing with DeepSeek and then they will use that efficiency to make even more use out of the architecture that they have.