Nathan Lambert

Amazon has always optimized for the small customer too, though, right? Obviously, they optimize a lot for the big customer, but when they started, they just would go to random Bay Area things and give out credits, right? Or just put in your credit card and use us, right? Back in the early days. So they've always... The business has grown with them, right? And Virgin.

So why does Amazon... Why is Snowflake all over Amazon? Because Snowflake, in the beginning, when Amazon didn't care about them... was still using Amazon, right? And then, of course, one day, Snowflake and Amazon has a super huge partnership. But, like, this is the case. Like, Amazon's user experience and quality is better.

So why does Amazon... Why is Snowflake all over Amazon? Because Snowflake, in the beginning, when Amazon didn't care about them... was still using Amazon, right? And then, of course, one day, Snowflake and Amazon has a super huge partnership. But, like, this is the case. Like, Amazon's user experience and quality is better.

So why does Amazon... Why is Snowflake all over Amazon? Because Snowflake, in the beginning, when Amazon didn't care about them... was still using Amazon, right? And then, of course, one day, Snowflake and Amazon has a super huge partnership. But, like, this is the case. Like, Amazon's user experience and quality is better.

Also, a lot of the silicon they've engineered makes them have a lower cost structure in traditional cloud storage, CPU, networking, that kind of stuff than... in databases, right? Like the, you know, I think like four of Amazon's top five revenue products, uh, margin products are like gross profit products or all database related products like redshift and like all these things, right?

Also, a lot of the silicon they've engineered makes them have a lower cost structure in traditional cloud storage, CPU, networking, that kind of stuff than... in databases, right? Like the, you know, I think like four of Amazon's top five revenue products, uh, margin products are like gross profit products or all database related products like redshift and like all these things, right?

Also, a lot of the silicon they've engineered makes them have a lower cost structure in traditional cloud storage, CPU, networking, that kind of stuff than... in databases, right? Like the, you know, I think like four of Amazon's top five revenue products, uh, margin products are like gross profit products or all database related products like redshift and like all these things, right?

Like, um, so, so Amazon has a very like good Silicon to a user experience, like entire pipeline with AWS. I think Google they're in for their Silicon teams. Yeah. They have awesome Silicon internally, TPU, the YouTube chip, um, you know, some of these other chips that they've made and, uh, The problem is they're not serving external customers, they're serving internal customers, right?

Like, um, so, so Amazon has a very like good Silicon to a user experience, like entire pipeline with AWS. I think Google they're in for their Silicon teams. Yeah. They have awesome Silicon internally, TPU, the YouTube chip, um, you know, some of these other chips that they've made and, uh, The problem is they're not serving external customers, they're serving internal customers, right?

Like, um, so, so Amazon has a very like good Silicon to a user experience, like entire pipeline with AWS. I think Google they're in for their Silicon teams. Yeah. They have awesome Silicon internally, TPU, the YouTube chip, um, you know, some of these other chips that they've made and, uh, The problem is they're not serving external customers, they're serving internal customers, right?

I really don't think so. We went through a very long process of working with AMD on training on their GPUs, inference and stuff. And they're decent. Their hardware is better in many ways than NVIDIA's. The problem is their software is really bad. And I think they're getting better, right? They're getting better faster, but the gulf is so large.

I really don't think so. We went through a very long process of working with AMD on training on their GPUs, inference and stuff. And they're decent. Their hardware is better in many ways than NVIDIA's. The problem is their software is really bad. And I think they're getting better, right? They're getting better faster, but the gulf is so large.

I really don't think so. We went through a very long process of working with AMD on training on their GPUs, inference and stuff. And they're decent. Their hardware is better in many ways than NVIDIA's. The problem is their software is really bad. And I think they're getting better, right? They're getting better faster, but the gulf is so large.

And, like, they don't spend enough resources on it, or haven't historically, right? Maybe they're changing their tune now, but, you know, for multiple months, we were submitting the most bugs, right? Like, us, semi-analysis, right? Like, what the fuck? Like, why are we submitting the most bugs, right?

And, like, they don't spend enough resources on it, or haven't historically, right? Maybe they're changing their tune now, but, you know, for multiple months, we were submitting the most bugs, right? Like, us, semi-analysis, right? Like, what the fuck? Like, why are we submitting the most bugs, right?

And, like, they don't spend enough resources on it, or haven't historically, right? Maybe they're changing their tune now, but, you know, for multiple months, we were submitting the most bugs, right? Like, us, semi-analysis, right? Like, what the fuck? Like, why are we submitting the most bugs, right?

Because they only cared about their, like, biggest customers, and so they'd ship them a private image, blah, blah, blah, and it's like, okay, but, like... I am just using PyTorch and I want to use the publicly available libraries and you don't care about that. Right. So they're getting better. But like, I think AMD is not possible.

Because they only cared about their, like, biggest customers, and so they'd ship them a private image, blah, blah, blah, and it's like, okay, but, like... I am just using PyTorch and I want to use the publicly available libraries and you don't care about that. Right. So they're getting better. But like, I think AMD is not possible.

Because they only cared about their, like, biggest customers, and so they'd ship them a private image, blah, blah, blah, and it's like, okay, but, like... I am just using PyTorch and I want to use the publicly available libraries and you don't care about that. Right. So they're getting better. But like, I think AMD is not possible.

Intel is obviously in dire straits right now and needs to be saved somehow. Very important for national security, for American technology.