Our Guest Chris Lattner Discusses
The AI Coding Revolution Has a Dirty Secret
AI can write code faster than ever. So why do we still need software engineers?
Chris Lattner, creator of Swift and LLVM and co-founder of Modular, joins this week’s episode to talk about coding, vibe coding, and why more code doesn’t always mean better software. He explains how AI agents are changing developers’ jobs and why someone still needs to decide what’s worth building.
Chris also introduces Mojo, his programming language for high-performance computing, and shares how mistakes made launching Swift shaped his approach. We also discuss the pressure on open source maintainers and how to keep up with AI without chasing every new tool.
00;00;01;16 - 00;00;16;15
Chris Lattner
You say, like, let's build a new programing language to solve all of compute. Okay. That's that's an epic industry scale problem, right? That's not a little thing. That's not that's not a problem that normal people take on. And you go into that with a huge amount of ambiguity of what does success look like? Can we even do it?
00;00;16;19 - 00;00;24;15
Chris Lattner
How are we going to compete against the titans of the industry that have been working on this for decades? Right. Apple, Tesla, Google?
00;00;24;20 - 00;00;46;27
Geoff Nielson
Chris Leitner has built coding infrastructure for all of them that you've probably never heard of. But if you're a full stack developer, then LVM, Swift, and Clang are probably all familiar to you. Now, Chris is taking on what he calls an epic industry scale problem, creating a new programing language to break down siloed environments and cover all of compute.
00;00;47;03 - 00;01;02;24
Geoff Nielson
In a moment when AI is upending all of software development. Chris is the rare guy who doesn't just have a vision for the future. He's actually building it. I'm Jeff Nielsen, and this is digital disruption.
00;01;02;27 - 00;01;03;07
Geoff Nielson
Well, Chris.
00;01;03;07 - 00;01;03;17
Chris Lattner
Really.
00;01;03;17 - 00;01;22;26
Geoff Nielson
Excited to have you here today. And you know, so many things we can talk about. But maybe just to kick us off, I'm curious to just get your take on, I guess, like the state of coding in AI in 2026, like where we are now. You've been you've been at this for a long time. You've been building tools for programing programing languages for a long time.
00;01;22;28 - 00;01;27;05
Geoff Nielson
Like where are we now relative to where we've been? Yeah.
00;01;27;05 - 00;01;49;24
Chris Lattner
Well thank you Jeff. It's such a fun time right now. So I've been building developer tools for the last 20 plus years. I've built the Swift programing language. I've been building compilers in those low level technologies coders use to build tools. But I've been in AI for the last ten. And everything has changed. It feels like in the last year for coding and AI has really been driving so much innovation, so much change, but also a lot of uncertainty.
00;01;50;01 - 00;02;09;06
Chris Lattner
And everybody feels like they're either not going fast enough or they're falling behind or they're almost out of a job. But we also have this world where people keep hiring software engineers, right. And so I think it's really interesting when you look at how to square these things. And so my integrated take so I still write code and I love the tools.
00;02;09;09 - 00;02;27;25
Chris Lattner
Is that if you're not using them, if you're not like staying on the forefront and trying these tools out, then you're missing out. Right. I mean, there's a lot of great things for productivity. I find it empowering because I can do experiments I otherwise wouldn't do, and it's fantastic. But also, I think the the hyperbole gets ahead of things.
00;02;27;28 - 00;02;39;27
Chris Lattner
And I think people are quick to say, oh, all coding is solved and things like this. And yes, it is much easier to write code. The building software products is much more than writing the code.
00;02;39;29 - 00;03;01;28
Geoff Nielson
The hyperbole feels like a good word for it. And you know, I'm sure you're subjected to some of the hype around, you know, like coding being obsolete and like vibe coding being the future, which, by the way, I don't I don't buy you. I don't subscribe to that at all. I think it's like a fundamental misunderstanding of what engineers do to suggest that, yeah.
00;03;01;29 - 00;03;23;11
Geoff Nielson
I'm you know, you mentioned that there's more coders being hired than ever. One of the things I'm trying to square, though, that's interesting, is like, maybe I'm just spending too much time on Reddit, but it seems like there's a lot of moaning in the developer community right now that that AI like, you know, things are sad and the joy is being taken away or something like, what do you make of that?
00;03;23;13 - 00;03;51;09
Chris Lattner
Well, it's super bimodal, right? I mean, there's people that are, you know, things are over in the joy of creating things is dead. And then it's very opposite of everything is easy. Like, nobody should worry about this stuff and there's very little of this middle ground. And so, I mean, I see a lot of different things. I see people that don't like change in are holding on to old ways, and they're just putting their head in the sand and hoping that nothing will ever change around them.
00;03;51;09 - 00;04;12;19
Chris Lattner
And maybe we'll go away again. I don't think that's realistic, but I see that and and I see a lot of people that are kind of, you know, super enthusiastically adopting new technologies and, overstating. Right. And so I see that I think that a lot of the middle of the road people object to that. They object to the over claims, they object to the noise.
00;04;12;21 - 00;04;30;12
Chris Lattner
They object to the people that don't acknowledge the problems. And so when I look at this coding revolution, because I think it is a huge moment for us, it comes with huge benefits. It also comes with some downsides. And so in a world where it's much easier to write code, you get more code. I don't want more code.
00;04;30;12 - 00;04;52;23
Chris Lattner
I want less code that does more right. I don't want more bugs, which comes with more code. I want higher quality products. Right. Open source communities in particular are getting overrun because there's always been for years. Now, this this balance between contributors want to add new things, but the long time maintainers have to keep the project sustainable over time.
00;04;52;26 - 00;05;12;12
Chris Lattner
And it's always been balanced because the people contributing to code, it was actually pretty difficult to do that. And so the maintainers could steer the overall architecture of a project. They could see what was in the best interest of the project over a long time, and then it balanced out. But now that writing code and creating patches and adding adding new cool things is so much easier, right?
00;05;12;13 - 00;05;30;28
Chris Lattner
This balance is shifted. And now a lot of the maintainers are grumpy. And so there's a lot of reasons to complain. And it's not just about the joy of writing code or getting getting getting more code to exist. It's about software engineering. It's about building products. It's about the higher level things that coding is a component of. I think that's that's where the struggle comes from.
00;05;31;00 - 00;05;49;02
Geoff Nielson
So, so where do you see the shape of, I guess, the development life cycle going? Because I can, you know, there's a world where we just invest a lot more in maintenance and cleaning up this code, and I've somebody probably come up with a better term for it than this, but I refer to it as code slop. Like, it's just this like like slop with everything else I related.
00;05;49;02 - 00;06;09;02
Geoff Nielson
Like, there's just this proliferation of shitty code that somebody has to deal with. Like, is the solution to just have more people dealing with it is the solution to just get better at requirements and doing smarter things up front. Some combination of both? Or do you reject the the dichotomy there?
00;06;09;05 - 00;06;23;27
Chris Lattner
I think it's a little bit more nuanced than that. So I think that the models are getting better. And so the, the outright slop is getting reduced as the models get better. I think that's one aspect of it. I think the people are also learning how to use the tools. And so I, I look at AI as a tool.
00;06;23;28 - 00;06;43;16
Chris Lattner
It's an amplifier. It's it's a super power fuze in the right way. It's a great curse. Fee is in the wrong way. Right. And so we as an industry need to figure out how to best use the tool. And that includes not just the production of code, but also includes code review and all the different things that go into the I think that really comes back to what is it that we want to build.
00;06;43;18 - 00;07;11;26
Chris Lattner
And I see a lot of people that are using a lot of tokens to do fairly meaningless work. It's like rewrite this thing into that thing and, and change it around in arbitrary ways, or re-engineer existing software, but in a different way. It's like, okay, well, that's not progress. That that is motion without without progress. And so this is where I think that we as an industry are seeing a lot of these things, and we're mis attributing progress to motion, but motions only good if it lines up in a direction that gets you someplace new.
00;07;11;28 - 00;07;29;03
Chris Lattner
Right. And I think that's that's also part of the confusion. I don't think it's a bad thing. I think it's a natural thing. Right. We're in an inflection point. There's a lot of cool things happening, and it will sort itself out. But I think that we're in that messy middle phase of trying to figure out how to best use the tools.
00;07;29;05 - 00;07;43;27
Geoff Nielson
It makes sense. And you use the word revolution to kind of describe the the stage that we're in, what's, you know, aside from all the fear and all the, you know, hyperbole, like what's got you genuinely excited about where this is going?
00;07;43;29 - 00;08;09;15
Chris Lattner
Well, so one of the things that I'll take a controversial take, which is most of what coders, including myself do, is not actually novel. So one of the things I'm working on is Mojo, which is a programing language. Right? And Mojo has a lot of really novel and cool things. I'm sure we'll talk about it later. But programing languages are well known and there's components that have been implemented the same the same thing in multiple different languages.
00;08;09;16 - 00;08;27;22
Chris Lattner
Like you need arithmetic, you need basic types, you need these structures. Right. And so one of the things that, you know, I think programmers forget is that, you know, you may not be the expert at all things. You may be extremely knowledgeable, but there are people that are working on Rust or Swift or other languages that have tried different things.
00;08;27;22 - 00;08;47;25
Chris Lattner
And because it's outside of your field of view, you don't know what is happening. A has that has that field of view. It can see that. And what that means is it has that broader perspective. But you need to have that judgment to lean into what's what's actually the right thing to do. And so because a lot of what we do is it feels novel to us as first time we've done it, it isn't actually novel.
00;08;47;27 - 00;09;05;27
Chris Lattner
A lot of what that becomes is a problem. Like, how do you how do you draw the best from this new savant that has read all the world's code, but really doesn't know what your objectives are? How do you work with that as a pure programmer, and how do you develop something together? I think that's a that's a really big and interesting question and challenge.
00;09;06;03 - 00;09;28;11
Chris Lattner
Now, again, I don't think that reduces the value of the software engineer, the developer or the development of team to zero. I think what it does is it pushes more on a core product management like product design, customer centric workflows and things like this. I think it shifts that balance a little bit, but I still think that the act of creating code is still very important.
00;09;28;18 - 00;09;50;10
Chris Lattner
And honestly, I made it way easier to write code. But most code is read, not written. And so I think that's particularly true. And it's being read by eyes and many other things too. But but it does. You know, code hasn't gone away. And I think that a lot of people want to jump to a finish line that is still up for debate, which is where do we get to and when things settle out, like is code obsolete?
00;09;50;11 - 00;09;55;27
Chris Lattner
Well, I think some, some evidence is still necessary in that for that claim.
00;09;55;29 - 00;10;13;09
Geoff Nielson
I'm, I'm curious to dig into that a little bit because there is talk of like, you know, the end state of all of this being like, you know, natural language, you know, programing or whatever you want to call it that, that code in the traditional sense goes away and you just, you know, tell your computer what you want and, and and it does it it.
00;10;13;12 - 00;10;23;29
Geoff Nielson
Do you see that? I mean, let me maybe ask those two ways. Is that the direction we're heading in? And is that like a true objective or is it just something we're heading toward that we can never fully reach?
00;10;24;07 - 00;10;46;19
Chris Lattner
I think that can work in very limited cases. So if you say, I want a clock widget, you're you're what you're doing is you're referring to a specification of a product by a well known preexisting concept that's been done dozens of times before. Right. And so what you're doing is you're delegating the specification to a standard. What is a clock widget?
00;10;46;19 - 00;11;03;17
Chris Lattner
What is a calendar? What is like. These things are known. And so if you just want a standard thing off the shelf well yeah I can definitely give that to you. But softwares always have reusable components right. That's not actually a new thing right? The problem comes when you try to do something novel. We're trying to build something novel.
00;11;03;17 - 00;11;20;21
Chris Lattner
You have to specify what it is, what are the constraints, what are the trade offs, how does it work? Right. And when you do that, suddenly you do care about things like how much space does it use? How much compute does it require? Is it cloud local or not. Right. And you know, different things like this really do matter for architecture.
00;11;20;27 - 00;11;37;28
Chris Lattner
And so you need to ultimately specify how some of this stuff work. Now there's different theoretical ways to do it. I can't predict the perfect future. I don't I don't know I'm helping build it. But the only way to to to get there is to actually do the work. But you do need a way to specify things.
00;11;37;29 - 00;11;57;05
Chris Lattner
And so for higher level applications that are designed by reference to something else, I think that there could be that kind of situation. But for deeper systems components, I'm very skeptical. And often the deeper systems components, you care much more about the abstractions, the composability, how they work in different scenarios. It isn't just one outcome. You're you're building.
00;11;57;05 - 00;12;14;26
Chris Lattner
You're building something that can be used by many different constituents. And so suddenly you do want to reason about things like interfaces and abstractions and stuff like this. And code is a pretty good way to do that. Maybe there's going to be a better way, but so far I haven't seen that. English is not a precise specification.
00;12;14;29 - 00;12;39;26
Geoff Nielson
I talked to a lot of CIOs, and everybody knows that running a successful IT shop is extremely hard, and AI has only made it worse. You know the drill. Immature processes, a fragmented data ecosystem and an unlimited demand for it. It all makes for an impossible job. And yet nobody has invented a unified operating system for it. We're literally the last function without one until now.
00;12;39;28 - 00;13;01;19
Geoff Nielson
That's why Infotech created CIO analytics, a single platform to run the IT department. We call it API driven advisory. It's powered by your data, our analyst insight, and it works with your existing tools right out of the box. If you're interested, click on the link in the description to request access and check out CIO Analytics and Action for yourself.
00;13;01;21 - 00;13;27;14
Geoff Nielson
An Infotech team lead can get you started. Same day. Better context, better insight, better IT performance only with API driven advisory, only with Infotech. The idea that all of coding gets reduced to clock applications and things that aren't novel or innovative, right? Like if that was the point, why would you need code at all? So so you know, maybe, maybe, maybe.
00;13;27;14 - 00;13;40;15
Geoff Nielson
Let me ask you, you know the question this way. But like what? What makes a good software engineer? And do you see that as changing or is it just the same? Fundamental skills are more important than ever.
00;13;40;18 - 00;14;01;19
Chris Lattner
Well, so the challenge with the question that you pose is that you assume that all software engineers are the same, and I don't and I don't think that's true. I think the software engineering is a broad category and there's multiple different buckets, right. And it could be that you're building end user applications, and your job is to make them as quickly and cheaply as possible.
00;14;01;21 - 00;14;21;11
Chris Lattner
Quality isn't really the thing. You're in a contract shop and you're just trying to get the thing out the door, right. It could also be that you're building the Linux kernel, and you're building a core operating system component that billions of people and countless dollars depend on, right? And so I think that what we need to do is we need to calibrate what is what is that job?
00;14;21;11 - 00;14;53;14
Chris Lattner
How do we think about it? What are the rules? And so the smaller the impact if you get to cool I'm building the clock widget to make you know. Hello kitty. But for clocks or something. Right. Well, cool that then. The stakes are low and the job is really outcome driven, right? The more you shift towards building core infrastructure like compilers or operating system kernels or things like this, the more you realize you're building an artifact with tens of millions of lines of code, thousands of stakeholders, and it's a very social problem.
00;14;53;16 - 00;15;13;21
Chris Lattner
And so AI is a great way to drive the engineering. Be soft, to drive consensus. You still have to be able to reason about what success looks like. You still have to motivate a significant team. In the case of Linux, it's a hugely distributed team, and that actually is a very different social set of problems. That being the most brilliant singleton engineer isn't really enough.
00;15;13;24 - 00;15;29;23
Chris Lattner
Like, you can be the most brilliant person that doesn't get along with other people and it won't go very far in your career, right? And so I think the software engineering has always been, you know, you get taught coding in school, but really what should be taught is how to build projects, how to work with other people, how to scale out.
00;15;29;28 - 00;15;47;21
Chris Lattner
And a lot of that is not taught in school. At least it wasn't when I was in school. A lot of that's learned on the job. It's learned by joining a team, learning the culture, learning, learning how to plan and execute things. What are we doing over the next four quarters? Right. These kinds of things are much more a part of software engineering and AI codings.
00;15;47;21 - 00;15;51;02
Chris Lattner
Awesome, but hasn't really helped solve those problems yet.
00;15;51;04 - 00;16;21;13
Geoff Nielson
Well, it's funny because it's taking me back now to the conversation we were having about like the, you know, malaise among software engineers. And I've, I have more of like a product management and development background. I've worked with a number of software engineers. And if you are a software engineer and I've met many who, like, want to be just left in a dark corner alone to write code, yeah, that's probably a pretty depressing outlook when we say, hey, actually, there's going to be more socializing than ever, and they say, no, take me back to to the olden days, I guess.
00;16;21;13 - 00;16;23;16
Geoff Nielson
But I don't know, like.
00;16;23;19 - 00;16;50;08
Chris Lattner
The word if you if you look the word Renaissance, right. Renaissance is really kind of talking to there's a before and after there's a moment. It's not that everything ends. It's not that the world is stopped after the Renaissance. It's actually a new and different and vibrant place, but in new ways. Right. And some people pull forward and make it through and they, they, they're happy or on the outside, some people are grumpier.
00;16;50;11 - 00;17;05;06
Chris Lattner
Right. But but it is a big change. So I think it's very likely that some people will be grumpy. Right. Some people do need to evolve and and they'll be pushed on their boundaries. And I think that again, for any individual engineer, you have a choice. Do you want to lean into things? Do you want to evolve? Do you want to grow?
00;17;05;06 - 00;17;21;12
Chris Lattner
Do you want to learn, or do you want to entrench? And do you want to complain on Reddit? Right. And so it's really a choice. And so there's different outcomes. I don't think that it's going to be complete doom and gloom and whatever. But there's definitely a change. And denying that there's a change. It's not going to get you anywhere.
00;17;21;12 - 00;17;37;27
Geoff Nielson
So I, I completely agree with you on that. I think that's a great summary of it. You've talked in the past about this notion of agenda coding and being in an era now of agenda coding. What does that mean?
00;17;37;29 - 00;18;02;03
Chris Lattner
So coding, I look at it as saying you have a new kind of artistic teammate who is brilliant in certain ways and just knows incredible amounts of everything, but is not a fully well-rounded engineer. And so what you're doing is you're working with a new teammate and they're brilliant. They're very capable at certain things. You need to keep an eye on them.
00;18;02;05 - 00;18;18;29
Chris Lattner
You have to understand what you're what you're looking from, what what you're looking for. You have to coach them. You have to you have to work with them. You can again, depends on your outcome. If you don't really care how anything works, then it can be fantastic and they can have very little oversight. And you can you can vibe code in the positive way.
00;18;19;01 - 00;18;41;08
Chris Lattner
It could be that you're building long term infrastructure that needs to be bulletproof. And, you know, you have you have stakeholders that don't want it to crash, right? In which case, yeah, you should be more human in the loop. Right. But agenda coding is really about the you know, I look at it is you have AI teammates that now are helping build the product with you and you're orchestrating and working together with them.
00;18;41;14 - 00;19;06;14
Chris Lattner
I mean, to extend this has always been about the tech lead manager, right. And we've had we've had engineers, we've had non-technical managers. And then we have people that are both and technical leads. These are these are key roles that we've had in software engineering. But I've always liked the tech lead manager, somebody who can manage, people who can oversee, coach, guide people, but is still very technical, often hands on and helping build the code.
00;19;06;20 - 00;19;22;05
Chris Lattner
And so what I think we're seeing is we're seeing a shift. Now, a lot of engineers have not wanted to be managers and not wanted to deal with other people, not want to write performance reviews, all that kind of stuff. But they're taking on some of those other rules. And one of the great things about agents is you don't have to write a performance review.
00;19;22;08 - 00;19;42;13
Geoff Nielson
So no, it it is, I'm sure in a lot of ways better than managing a person. And I'm, you know, this just may expose the last time I was involved in this stuff, but it almost I don't know if you remember, like there was this big era of like pair programing a number of years back. I feel like it's it's like pair programing, but you're paired up with like, you know, AI versus a person.
00;19;42;14 - 00;19;44;07
Chris Lattner
I think it's a great analogy. Absolutely.
00;19;44;08 - 00;19;44;23
Geoff Nielson
Yeah.
00;19;44;27 - 00;20;04;06
Chris Lattner
Well, and also with that, with that, I mean you look at if you put yourself in the, in the frame of pair programing, you're not competing. I think this, this is actually one of the biggest questions is, are we trying to compete with the AI and prove how valuable we are, or are we partners? And we're looking at how to get the most out of both of us together.
00;20;04;08 - 00;20;19;22
Chris Lattner
And I think that one of these is very adversarial and negative. And again, it's the head in the sand kind of approach. The let me proof how terrible the AI is and why I shouldn't have to use it. The other approach can be very empowering, saying, look, this is a tool. I want to go fast. I want to be able to achieve new things.
00;20;19;22 - 00;20;38;21
Chris Lattner
I want to, I want to learn and solve problems faster than I was able to solve. And from that framing, suddenly it's very productive. And you can look at these new tools that are coming and you say dispassionately, does this help me get more done or not? Like get past the hype? Like a lot of people have hype and there's a lot of PS in the world, but can I get more done faster?
00;20;38;22 - 00;20;54;22
Chris Lattner
Does it make me happier as a programmer? Does it get the mechanical the mechanical nonsense out of the way for this big refactor? Like, these are objective questions you can ask. And I think if you frame on that, even kind of the middle of the road programmers that are, you know, they're not they're not AI pilled like, oh my gosh, like we're just going hog wild.
00;20;54;22 - 00;21;02;15
Chris Lattner
But they're also not head in the sand type people. I think that's really the best framing. Like, how does that help you?
00;21;02;17 - 00;21;28;16
Geoff Nielson
It's I like that and I want to like I want to push on that a bit because there's a world where the future of coding is like it's the same as it's always been, but like throughput is just higher, right? And in theory, maybe quality gets a bit higher to, but there's another, I guess, slightly more interesting world where the shape of a development team or like a product development team or holistically changes based on what's needed.
00;21;28;19 - 00;21;36;20
Geoff Nielson
Do you see the shape of the the team changing, or just the same people doing a little bit more, allocating their time differently?
00;21;36;22 - 00;21;59;13
Chris Lattner
I mean, I it depends completely on the team. I could see situations where you need Wayfarer engineers. I think that's possible. I also see the situation where the role shifts slightly from maximizing production of code, typing like that's no longer the most important piece to much more strategic thinking, right? So this is where it's like the shift to product manager.
00;21;59;15 - 00;22;17;07
Chris Lattner
Like so what should we build? What does success look like? What are the trade offs? Rebalancing. Two customers are asking for two different things. How do we balance them? What do we do about it. Right. These these questions I think are inherent. And I think that that's something that that skill set of thinking, not just about what we're doing, but why it gets rewarded.
00;22;17;10 - 00;22;42;16
Chris Lattner
Now, I still think you need people that are building building the code that are working on the system. I also think that the tools around AI need to keep up. So continuous integration, the whole software development lifecycle, code review, things like this, these all need to evolve. And I think that again what we're seeing is we're seeing one part of the problem go really fast because of the progress in AI models and coding and things like this, but the whole rest of the equation needs to rebalance.
00;22;42;17 - 00;23;00;12
Chris Lattner
And so we're still in the midst of that. And I don't think that I don't think we know what the end state looks like. Some people are saying nobody should ever look at code. And then therefore a lot of things don't matter. I don't personally believe that, at least not for all use cases. Other people are saying that, you know, none of the stuff will ever work at scale.
00;23;00;15 - 00;23;06;02
Chris Lattner
I also don't believe that. And so there's got to be a middle of the road answer here.
00;23;06;04 - 00;23;31;25
Geoff Nielson
Digital disruption has partnered with McLean and company to bring you the human centric AI program. If you're a knowledge worker, you've probably noticed that your colleagues are all using AI, and it's not always pretty, but it's a solvable problem. It's not a tools problem, it's a people problem. And yet we keep training on tools. McLean and company's human centric AI program is a judgment first learning experience built to drive AI impact at scale.
00;23;31;27 - 00;23;46;14
Geoff Nielson
Three self-paced learning paths, plus lid expert led sessions and content that evolves as AI does. Better judgment, better outputs, better outcomes. Check it out at the link below to learn more.
00;23;46;16 - 00;24;15;27
Geoff Nielson
I wanted to ask. There's been a little bit of like, tension I picked up on in a couple of the different things you said. You know, you mentioned earlier more coders than ever. You also mentioned on specific teams, maybe you need less coders, you know, based on your most reasonable outlook. Do you still see the market, I guess for coders growing, like, does the economics of being able to write code more cheaply just increase the demand for for people doing it?
00;24;15;29 - 00;24;35;27
Chris Lattner
Well? So I mean, I can't I can't tell you the future, but I can look at the past. So I coding is not something that happened yesterday. It's been amplifying this year. I can't so it's it really kicked off late last year around this time. And then it's just been amplifying. Right. And so you could look at the the job statistics things like this.
00;24;35;27 - 00;24;59;20
Chris Lattner
You could look at hiring. You can look at OpenAI and anthropic are still hiring people. Right. So among the most AI companies in the world, they still have bugs. They're still hiring people. So I don't know what the future looks like. But you can look at some of these proxy questions and, you know, yes, some people are running around and I know some big companies did layoffs and said, oh, well, AI is the reason we don't need as many engineers.
00;24;59;22 - 00;25;17;08
Chris Lattner
I again think that is a little bit of hype and BS. I think that maybe they were just overstaffed to begin with and they used AI's in excuse. Right. And so so I don't really know, but I haven't seen that fundamental shift. Also, there's this question of when the coding gets cheaper. Do you just need more software? Right.
00;25;17;15 - 00;25;36;12
Chris Lattner
And so does everybody become a coder. And one of the things that I love about the AI tools is that I see people that naturally would not, like, dive into building an app that do now, right. And so many people are actually willing to build things, and I don't know if they identify as software engineer. They probably reject that because, you know, coding was always too hard.
00;25;36;12 - 00;25;50;07
Chris Lattner
They could never figure it out. But they're building apps or building programs or building interactive web experiences. They're doing all kinds of stuff that maybe they wouldn't do before. And so by that metric, maybe we're going to have ten times the number of programmers, right.
00;25;50;10 - 00;26;16;29
Geoff Nielson
That that's an interesting, if not scary point. And the reason I'm saying that is I'm thinking back to those poor souls doing the maintenance of these production environments. And like, I'm sure there's lots of people that please, for the love of God, like don't push your like misunderstood code, you know, business users into our production environment. So I'm sure there has to be some sort of like guardrails or like, I hate to say safety first, but.
00;26;17;01 - 00;26;35;25
Chris Lattner
I agree with you. I agree with you. But not everything is large scale production systems. I happen to like working on large scale production systems, and it sounds like you do too. Right. But not everything is like sometimes it's just a dashboard that I need. Right. And again, let me let me self-identify. I'm a systems programmer, low level hardware, software boundary in person.
00;26;35;28 - 00;26;53;23
Chris Lattner
I never would have built a dashboard. Like that's just not something I don't know. I'm not an expert on all the front end technologies that day and things like this. And now, yes, I will like I'll get a nice little visualization thing for performance or whatever, whatever it is. Right. And so that's that's empowering. That's awesome. That that's more software that would have existed.
00;26;53;25 - 00;27;02;09
Chris Lattner
I'm not going to ship that to the customer. But but it's still really cool. And again, this is part of the experience of building a program being a programmer.
00;27;02;11 - 00;27;30;01
Geoff Nielson
So so sticking with the, I guess, the larger scale development environments and maybe getting a bit more technical. I mean, you've you've built mojo, you've spent a number of years on it. You've talked about, I guess, this prediction about moving toward a more heterogeneous AI compute environment. And, you know, frankly, it just sounds like an environment, a systems environment that's just getting more complex rather than less complex.
00;27;30;03 - 00;27;36;00
Geoff Nielson
Can you unpack that a little bit in terms of a what's happening and be what the implications are?
00;27;36;03 - 00;27;59;06
Chris Lattner
Absolutely. So if you take a step back, we see AI is happening all around us. Coding is one aspect of that, but also ChatGPT and consumer experiences. New new agent tenants, like all kinds of cool stuff is happening. But how does it work? Well, what you need is you need compute. You need accelerators, you need GPUs, you need specialized chips to be able to run all this AI.
00;27;59;08 - 00;28;27;19
Chris Lattner
And the reason for that is just the workloads for neural networks are very specific. They're very dense computations. Lots of matrix multiplications using specialized computers can give you a lot more performance and power savings and other other benefits for running these kinds of workloads. So in this space. The AI world grew up on Nvidia GPUs. And there was these technologies like TensorFlow and PyTorch literally enabled a lot of the research over the last ten years to happen.
00;28;27;19 - 00;28;46;10
Chris Lattner
And they're very, you know, video did a really good job of leading into these projects, helping catalyze the ecosystem. And as a consequence of this, like, there's so many Nvidia GPUs going into data centers and I'm sure many, many people will see the next gigantic warehouse being built and filled with GPUs. And they're wondering like, what the heck is going on in there?
00;28;46;12 - 00;29;11;09
Chris Lattner
Now? Nvidia does a great job, but they have many different customers made different use cases. They need to be able to run well and guess and statistical whatever and render triangles and do all these things that a GPU was originally designed to do. And they're amazing at AI, but they're not really purpose built for it. And so what we've seen over the last few years is that we see actually AI specialized accelerators.
00;29;11;09 - 00;29;36;18
Chris Lattner
And so Google has the TPU, they call it AWS has training. There's many of these different specialized accelerators that are saying, let's give up on graphics. We're not rendering triangles anymore, at least not for this workload. Let's really specialize and be very dedicated to AI workloads in particular. Now, the problem with that ends up being that you get a lot of performance and a lot of cost benefits and other things like this by specializing chips.
00;29;36;25 - 00;29;56;16
Chris Lattner
But how do you program them? This is a big open question. And so one of the things I worked on, back in 2017, almost almost a decade ago, is the Google TPU and bringing up the Google TPU. It's a very specialized chip. It's very advanced, a whole bunch of different ways. The architecture looks very different than a CPU.
00;29;56;22 - 00;30;26;09
Chris Lattner
It also looks very different than a GPU. And so at Google, we had to build an entire software stack just to enable this chip. You fast forward to where we are today. It's nearly ten years later. It's not just Google this building. These things. Lots of different companies are building these. And also you get into as AI is exploded, you have, you know, Apple with laptops and you have so many different companies that are building AI powered and accelerated technologies that they all need a software stack to go into this.
00;30;26;11 - 00;30;44;08
Chris Lattner
Now, the problem with that is that there is no standard. Everybody has built their own thing. Some of them have done a better job than others. Often they're not all that great, honestly. But for application developers, what we want is we want choice. We want to be able to scale across a wide range of different chips. If you're in the data center.
00;30;44;10 - 00;31;01;04
Chris Lattner
Yeah, it would be awesome to be able to use Google TPUs. It'd be awesome to be able to use Qualcomm or other systems. If you're building a smart tablet or a pendant or some other device like you don't want to have to completely rewrite your software, you want AI to work. And so this this is a huge problem that we have.
00;31;01;04 - 00;31;15;06
Chris Lattner
And it's this combination of not just the applications moving fast, but it's all the hardware moving fast. It's all the software that has to run on it. This entire ecosystem is a gigantic mess. Right. And so this is a huge issue that we're all struggling with.
00;31;15;08 - 00;31;32;16
Geoff Nielson
So where do I I was going to ask where Mojo sits in the stack. But maybe just before I do that let's let's rewind a little bit and like tell me about Mojo. Tell me about like the vision behind it. And I guess like the motivation for why you would build something like this in the first place.
00;31;32;20 - 00;31;51;16
Chris Lattner
So let's get into a little bit of a personal story. So this is kind of, mojo. And a lot of what modular is building today is, is my life's work, right. This is not something that I woke up one day and decided to do over three, three, three weeks or something. And so across my my career, I spent the first roughly a decade at Apple building this fundamental technology called LVM.
00;31;51;18 - 00;32;15;04
Chris Lattner
LLVM is a compiler. Technology enables you to talk to CPUs, and LVM kind of took over the world in a good way. It powers Swift and Rust and Julia and and so many different systems. C++, like all these languages, the power, every phone, every data center, pretty much everything out there that has a CPU in it. Across that time, I built the Swift programing language.
00;32;15;04 - 00;32;34;27
Chris Lattner
That was a nice and weekends project that I started at Apple. It did well and so I was developers got to move forward. But about a decade ago I became really interested in AI. And so I took a phase shift. And so for the last decade, I've been trying to figure out this AI thing. And as I mentioned, I worked on Google TPUs, got the software stack to scale.
00;32;35;00 - 00;32;52;09
Chris Lattner
It turns out that's really hard, a novel chip without software, you have to build an entire new software platform. And then I did it again. And about supporting like 4 or 5 different chips, even within Google. But each of them had a different software platform. And managing that was a nightmare. I then went on and worked at startups.
00;32;52;09 - 00;33;12;14
Chris Lattner
I saw many different peers and colleagues and other people that are working on PyTorch and TensorFlow in these systems, and they were never designed for this plurality of hardware. But that was a really big issue and people were always outcome driven. How do I enable the next demo? How do I enable get get this next feature work, get this next, next thing into my product?
00;33;12;14 - 00;33;39;06
Chris Lattner
And nobody is taking a step back and thinking about that software platform. And so about four and a half years ago, I started a startup company called modular by co-founder Tim Davis, and we said, hey, AI is a big deal. That was a little bit controversial four years ago, but I think now the world's agreeing, and we need a new software platform that can actually scale into all this different compute to unlock all the capabilities we want.
00;33;39;08 - 00;33;56;04
Chris Lattner
Right? We don't want after right software. Every time we want to be able to, to experiment with the new chip to see if it might be better without having to rewrite all of our software just to get there. And what we realized is this is not something you can do in three months. This is not something that is an easy problem.
00;33;56;06 - 00;34;17;08
Chris Lattner
And so we decided, let's let's actually do it. Let's lean in, let's raise a bunch of venture capital money. Let's let's build an incredible team of hundreds of people. And let's lean into building the fundamental software platforms and take this new technology that's been evolving over the last decade and uplevel it all. And so what that adds up to is it adds up to several new technology contributions.
00;34;17;08 - 00;34;44;28
Chris Lattner
One is Mojo. Mojo is a new programing language. It's Apache two. It's it's a very cool thing. Mojo's cool because it looks like Python, but it runs like rust or C++. And so it's super fast. It's but it's easy to learn like Python. It's very important. And unlike C++ or rust, it runs on all the chips. And so you can run Mojo on a TPU, you can run on an Apple GPU, you can run it on an Am.
00;34;44;28 - 00;35;06;21
Chris Lattner
Dear Nvidia data center or consumer accelerator. You can run on all these very exotic things as well as CPUs. And so it's really the first language that can span the entire realm of compute that everybody's looking at and provide a unified, easy to use experience for doing that. As you lean into this further, you realize, well, it's not just about being able to program a chip.
00;35;06;21 - 00;35;25;06
Chris Lattner
You want to actually be able to run models. And so we have another layer called Max, which is then saying, hey, let's actually take this and make it so you get the full AI developer experience and make it easy to scale across this for for models and workloads people care about, for building workloads and all the other things you want to do with AI, the.
00;35;25;09 - 00;35;51;15
Geoff Nielson
The the vision behind like just being able to have this work across all these different stacks and standards makes makes complete sense to me. Is this, you know, as someone who's not deep into this, is this additive in a stack? Does it replace, you know, some of the existing, you know, coding languages you use right now, like, how do you how do people who are using this and making the most of it actually integrate it?
00;35;51;18 - 00;36;12;06
Chris Lattner
Yeah. So it completely depends on what you want to do. And so modular is named modular because it plugs into all the things. And so you can take Mojo and integrate into your Python code. If you're a Python programmer, often you get to the point where you say, okay, I'm writing more and more and more code. Well, suddenly performance is a big problem.
00;36;12;08 - 00;36;32;14
Chris Lattner
What do I do? Well, I have to rewrite a big chunk of it in rust or C++ or something like that. And now I fall down this deep, deep rabbit hole of complexity with with mojo, you can say, look, I'll just switch this into Mojo. I keep pythonic syntax. I get all the benefits of a language like C, C++, but now my Python code, not just because thousands of times faster.
00;36;32;16 - 00;36;56;19
Chris Lattner
That's pretty cool. And it's not saying you have to switch everything over to Mojo and you go whole hog. It's just saying you can incrementally adopt something where it's useful and it reduces pain and just make sure like better, right. If you're coming in from the AI side of things, you can say, hey, I've got AI models and I want this, this very fancy Kimmie three model or minimax model or quantum model or things like that.
00;36;56;20 - 00;37;16;24
Chris Lattner
I want to run it on an Apple GPU. I want to run it on a train from W.S., I want to run it on and not just in video. Well, suddenly now you can use Max, and now you get portability. You get a full solution, you get a full stack that can actually take you into new places and make it much easier to experiment with these chips.
00;37;16;26 - 00;37;33;14
Geoff Nielson
I like the I like the flexibility of it. It's, you know, the way you kind of describe it is, is pretty compelling. One of the pieces, you know, I'll ask you a deliberately difficult question here, but whenever somebody says like, we've got a new universal standard, right. There's there's six standards and now we've got a new universal standard, right.
00;37;33;15 - 00;37;37;00
Geoff Nielson
Like, you know, the challenge you run into and certainly, you.
00;37;37;00 - 00;37;40;00
Chris Lattner
Know, now you have a seventh one, right. Correct.
00;37;40;03 - 00;38;01;22
Geoff Nielson
Correct. And so, you know, I'm going to qualify the back half of this question by saying, you know, one of the other pieces that of course, you're aware of is modular, you know, has you've since sold it to a hardware vendor in the space. Right. Qualcomm is now here. So so how do you avoid being that seventh standard versus staying true to the word modular.
00;38;01;25 - 00;38;25;19
Chris Lattner
Yeah. So that's a great question I'm glad you're going there. Please bring more hard questions. The reason that the modular stack is different than the six or a dozen prior standards is it's actually portable. Like this is this is not a thing. Nobody else has done that before, nobody else. So the modular stack today runs on six different chips from six different hardware companies.
00;38;25;21 - 00;38;54;05
Chris Lattner
Right. And so this is Apple and Vedere MD Qualcomm. Google AWS right. These companies on the business Battlefront don't really get along right. It turns out that any one of these companies are all lovely in different ways. And they have amazing engineers. They're not incentivized to build something that runs on surveillance chips. Right. And so to first answer your question like this is structurally different from that perspective.
00;38;54;10 - 00;39;13;14
Chris Lattner
Now you ask another good question, which is hey modular just joined Qualcomm. Qualcomm's building some of those chips. Why isn't it just going to make it to us chip. And this is something that we have to prove it. But I'll explain how we think about this. Qualcomm makes money selling components into every level of the data center. Right.
00;39;13;15 - 00;39;30;15
Chris Lattner
And so it's not just making a GPU. It is building really cool data center accelerators. There's a whole roadmaps for fun. But what it firmly believes in is that the data center is heterogeneous. There's going to be chips from startups, there's going to be chips from big companies, there's going to be chips from lots of little companies.
00;39;30;21 - 00;40;00;07
Chris Lattner
And the the chips are not all just the GPU. It's also the networking and the the timers and the switches and the optical interconnects and all of the things that go into building a data center. And there's a ton of silicon that goes into these things. Qualcomm makes a lot of it, by the way. Right. And so what Qualcomm wants is it wants choice and wants to neutralize the software landscape so that shifts can thrive and compete on their own merits so that people do have choice and go forward with this.
00;40;00;10 - 00;40;17;06
Chris Lattner
To make an analogy. So when I was at Apple, I was building this LLVM compiler technology that also helped make the entire world work, and pretty much everybody builds on top of it's the same thing. Like Apple realize they need the software to exist. They're going to invest a huge amount of effort into it and drive it into their products.
00;40;17;06 - 00;40;36;09
Chris Lattner
And Apple got a huge head start, by the way, by doing that. But you don't have to do that with the exclusion of everybody else. And so when Google wanted to pick up LVM for Android, you know, Apple's like, okay, well, you have compilers too, but our product value is at the iOS level, right? It's not at the compiler level.
00;40;36;15 - 00;40;54;14
Chris Lattner
And so, you know, this is a new way of thinking about open software for pretty much everything. Open source is not a novel concept, right? There's lots of reasons that people contribute to open source. This is this is also kind of a new thinking for Qualcomm, because Qualcomm hasn't had a lot of the open source DNA historically.
00;40;54;19 - 00;41;08;16
Chris Lattner
And so this is partially Qualcomm proving. Yes, it is developer first. Yes we can do things like yes it is like returning more credibility with the developers and kind of using modular as a way of saying like let's actually prove this to the world through that. So.
00;41;08;18 - 00;41;30;25
Geoff Nielson
So so I, you know, I'm a big fan of open source, and I know in some ways, like open source becomes like a contentious conversation. And I don't know, it's it's I don't know if this is too strong a statement or an obvious statement, but it feels like in some ways, like the capitalist motivations of big Tech are to like, quash open source and try and make things more proprietary and like lock them in here.
00;41;30;29 - 00;41;39;23
Geoff Nielson
And so, I mean, do you see that as a risk? Like do you see the future becoming more open source or less open source in this arena?
00;41;39;26 - 00;42;02;15
Chris Lattner
Well, so I mean, I can agree with you, I think that there are a lot of players in the industry that are looking at, we win through other people losing, right? I'm sure across your career, you may have noticed engineers or leaders or managers or other people that that lean into, you know, I and my career can be propelled by making somebody else look bad.
00;42;02;17 - 00;42;23;00
Chris Lattner
Right. And that can be very effective. Let's not let's not kid about it. Like people can make money on that. In the business battlefront, people can progress through their careers by doing politics, things like this. The people I've seen that I like, that I try to model are people that are successful because they work with other people and lift them and achieve things together as a team.
00;42;23;02 - 00;42;43;14
Chris Lattner
Right. And this is how I've been successful. This is how I built large scale open source communities. This is also I've been successful in my career as a leader. And so there's different strategies. And Qualcomm comes with that second, that second approach. They've always driven standards. They've always worked with other partners. They have a huge matrix set of relationships.
00;42;43;21 - 00;43;07;04
Chris Lattner
You know they're selling chips into meta devices or glasses or things like this. Right. And so they're working with those partners and always has been. And so that's part of their culture or part of their DNA. And so dominating everybody and driving out other people. Yeah. There's certain people that have that instinct. I think Qualcomm's more thoughtful about this and realizes that, yes, actually it's a diverse world with a lot of different devices.
00;43;07;07 - 00;43;29;18
Chris Lattner
Everybody's focusing on the data center GPU, and that's a really important topic that a lot of people are centering into. But everything has AI in it, Qualcomm's and most of those devices. And so it wants a really vibrant AI landscape. And and it realizes that it should be catalyzing this. And I think it's a very progressive mindset. I think this is this is one of the reasons why I was excited to join.
00;43;29;20 - 00;43;54;14
Geoff Nielson
No, I respect that. And it's, you know, personally aligned with my values as well. Like, I'm, you know, I'm of a like mind that that's the, you know, you can grow the pie versus just trying to steal share from from other people as well. One of the, I guess, one of the realities of Mojo and how it's been rolled out is and again, I'll deliberately make this a little bit hardball just for fun.
00;43;54;14 - 00;43;56;24
Geoff Nielson
And because, you know, we're having a good time here it is.
00;43;56;25 - 00;43;59;04
Chris Lattner
And that's, that's way more interesting. So bring it.
00;43;59;05 - 00;44;21;14
Geoff Nielson
Yeah, yeah. You will have fun. So so you know, Mojo was deliberately, as far as I can tell, initially rolled out like very bare bones with a lot of like, quote, trust me, it's going to be awesome. And, you know, kind of deployed feature by feature. It's, you know, it's on the roadmap. And, you know, it's only very recently that it's crossed kind of that 1.0 threshold.
00;44;21;16 - 00;44;38;20
Geoff Nielson
So I guess, you know, two things that are interesting to me is number one, it seems like that was a very deliberate strategy on your part. It's not accidental that that happened. So so why did you choose to do that? And then what's special about 1.0 that should get people's attention.
00;44;38;23 - 00;45;00;16
Chris Lattner
So let me so the answer, the short answer is lessons learned from Swift and the battle scars that I accumulated across that journey. And so let me tell you some of the mistakes I made. I'll take. Obviously, there's shared guilt that I'll just blame myself for the, the bad things. So swift the history of that is. I started Swift as a nights and weekends project in 2010.
00;45;00;18 - 00;45;21;02
Chris Lattner
Nobody at Apple knew about it for about a year and a half was just something I was building. I eventually told told my management about it and their heads kind of exploded. Right. Three years in. So 2013, it became a priority. Like, okay, let's actually put significant engineering on it. And then we launched publicly in 2014.
00;45;21;04 - 00;45;42;22
Chris Lattner
And so that was four years of development. Admittedly, part of it was just a very small team of one person, but but four years of development and then it launched as a 1.0. Now, when we announced Swift publicly, only 250 people in the world knew about it total. Most of them were in my team. And so it was basically my team, plus some senior execs, basically.
00;45;42;25 - 00;46;01;12
Chris Lattner
And so the problem with that is that if you build something in isolation, you could have really well meaning people. You can try really hard, but you're isolated from the use case and you don't really know, and you don't really have the validation that comes with real users. Right. And so when Swift launched, Swift one was a hot mess.
00;46;01;15 - 00;46;19;14
Chris Lattner
Like it was super painful. We were breaking people's code all the time. It was we had rushed features. There was a whole bunch of things that were missing that we wanted to add so swift to, for example, at an error handling so you could throw exceptions. Not the biggest like moment, but it's something that radically reshapes APIs.
00;46;19;19 - 00;46;38;23
Chris Lattner
And we just didn't have time for Swift them. So we got there. We we open source mojo years later we we we did all these things but or sorry, we are swift years later. But it was like six years between the start of Swift and and being opened. And the relationship with the developer community was very much very different.
00;46;38;23 - 00;47;03;04
Chris Lattner
And it was it was not great, let's say now Swift, a successful I'm very proud of it. The team has done an amazing work even after I've moved on to the projects. They've done a great job of bleeding, so forth. But when you start to do this again, you don't want to make the same mistakes. And so one of the things I noticed is that you do want both a small design team, because particularly early in a project, you want to move fast.
00;47;03;05 - 00;47;20;01
Chris Lattner
You don't want to be legislating everything. You want to have a small team of people that can sit around a lunch table together and talk about something, and then just make a decision you don't want to committee. But you also want validation. And so you want to be engaging with people. You do want people using it, and you do want to actually work with that kind of community.
00;47;20;03 - 00;47;40;26
Chris Lattner
And so with Mojo, what we did is we made it very controversial at the time, decision to open it really early. And so we we said. World. There's mojo 0.1 not 1.0. It's out there. You can use it. We want to talk to you. We want to get your feedback. But no we're not looking for your opinions to join the design team.
00;47;41;01 - 00;48;02;04
Chris Lattner
We want to keep the design team very small. And as it got more mature, we said, okay, well now let's open up the standard library. Well, standard library now being open source, that was a step. Then you say, okay, now we'll take patches. And so what this was was just saying okay, well enough of it. Is there enough of it stable that we can actually kind of afford to have design discussions with other folks.
00;48;02;11 - 00;48;20;25
Chris Lattner
And it's not out of malice. I do admit that I'm a little bit of a control freak, but so there's some aspects of that. But it's about being very deliberate about managing this, managing expectations, trying to communicate clearly and saying, you know, this is what we can we can support. And so to your point, we open source mojo completely.
00;48;20;25 - 00;48;44;18
Chris Lattner
We had 1.0 and we didn't want to open source it before the 1.0 because open source brings growth, creates more adoption. A lot of people told us we will not adopt it unless it's fully open source. And so we said, okay, cool, let's get it to the point where it's stable 1.0 and then open source it. And unlike the Swift thing where we released it, every adopted it and breaking people's code and it was just a gigantic mess.
00;48;44;21 - 00;49;07;21
Chris Lattner
And so we managed it very differently. Now, if you look at the timescales, Mojo total getting to 1.0 and open source was less than four years. And so in internet time and AI time, it felt like forever to everybody. But it turns out that building an entirely new program in the language, from scratch, with an innovative new design and the ability to target all hardware.
00;49;07;26 - 00;49;30;17
Chris Lattner
This isn't just building Swift, this is building something much bigger and much harder in some ways. So went actually quite fast. Right? And so this is where, again, everybody's going to be grumpy. The people on Reddit did not approve of the approach. And and that's cool. But our goal wasn't to appease the people on Reddit. Our goal is to build something that was really thoughtful, that can move the world forward, that could have innovative design points.
00;49;30;17 - 00;49;47;07
Chris Lattner
And so that was what I was, what I was going for. Now, I'm also very happy that it's fully open. Like I love the community and the reaction to that. And so I'm very happy we made it. But I think that the the making it, making it together with the world in the open, but not not trying to pull people into it early was a big piece of it.
00;49;47;08 - 00;49;56;22
Chris Lattner
And I think that was way better than keeping everything in the Apple style black box, where it's just like, completely secret and nobody's allowed to know about it until you it onto the world.
00;49;56;24 - 00;50;24;27
Geoff Nielson
So no, that's that's awesome. And I'm chuckling to it's just trying not not trying to appease Reddit. And I'm glad because I think that's an absolute fool's errand. But you know that that aside I really like the approach. It makes a ton of sense to me. And, you know, as I said, having a background in product development and innovation, like, yeah, I love the idea of piloting beta testing, getting something into people's hands and seeing, you know, what resonates with them, what they what you think they're going to care about, what they actually care about.
00;50;24;28 - 00;50;38;22
Geoff Nielson
And I'm like, what? Did you have any specific lessons learned there or any surprises in terms of like redoing the roadmap, things that you thought would be a would matter that didn't or like, what were the big kind of light bulb moments for you?
00;50;38;27 - 00;50;57;21
Chris Lattner
Yeah, totally. Well, let me tell you one brilliant thing we did and then one colossal mistake we made. Love it. So a brilliant thing we did. Unlike Swift, when Swift launch again, 1.0 error rate can use it to build apps. There was one app that was a test app that was built, and so there's basically no internal usage.
00;50;57;24 - 00;51;18;24
Chris Lattner
And we launched it and said everybody should use it, right. So zero to hero with an unproven technology. It was a gigantic mess. And so with Mojo, quite the opposite. We didn't build it for other people to use. We built it because we needed it. And so we we were using it all through the painful development process. Million lines of code, tons of internal users, a lot of people complaining very directly to the design team.
00;51;18;24 - 00;51;39;29
Chris Lattner
And so, I was brilliant because launching something without users is, I think you probably have seen is, is is really quite different than having dogfood it yourself and drink your own champagne kind of a thing right now. I think we did a great job there and I'm very proud of Mojo. We did make one huge mistake and this is something that I didn't.
00;51;40;02 - 00;51;56;18
Chris Lattner
Maybe I'm a little bit tone deaf. Like, this is me being not not aware of how humans interpret things. But when we launched Mojo as a 1.0, we said, look, it's a member of the Python family. It's still very early, but we think this can grow into something that feels like a successor to Python.
00;51;56;20 - 00;52;19;08
Chris Lattner
Okay. And so that was part of the message. And I still think it is true. But I didn't realize that the the need, the desire, the the the lust from the world for a successor to Python was so great that people missed the nuance thereof. It will eventually get there years ahead from now. And so people flocked in and said, okay, it's going to be a successor to Python tomorrow.
00;52;19;11 - 00;52;39;16
Chris Lattner
And so that caused a huge amount of confusion, a huge amount of product confusion, confusion with the design team about like, is this what we're supposed to prioritize or not or, or things like this? And so that was that. That caused a lot of confusion and that was a mistake. What we did is we then said, okay, well, let's radically pull the messaging back and say, stop talking about Python.
00;52;39;20 - 00;52;58;23
Chris Lattner
It's a member of the Python family. Yes, but it's really about GPUs. It's really about accelerators. It's about AI programing, it's about low level performance and things like this. And just now, as we're getting past 1.0 and it's becoming a more full featured language and things like this. Now we're saying, okay, it's it's getting better for systems programing in the next six months.
00;52;58;25 - 00;53;24;25
Chris Lattner
We need things like async await to fully land and be really good. Well it'll be a pretty credible competitor to languages like rust. And so we're building up from that direction. And probably in two years it'll be pretty interesting as a Python for type of thing. Right. And I think that part of my bet here is also, it doesn't need to be literally the same as Python, because we have AI tools, and tools are amazing at translating code and moving things around.
00;53;24;26 - 00;53;40;07
Chris Lattner
But what we want is we want to feel like Python should feel like what Python should have been, should bring in these new features, that should bring in these capabilities, that should solve the packaging and deployment problems, that should solve performance, obviously, but also enable this new category of compute.
00;53;40;09 - 00;54;10;02
Geoff Nielson
I totally get that from a user perspective. Like the the pitch there makes complete sense. And you know, the framing as a successor to Python. I'm curious about this. I don't know if anybody else is curious about this, but I'm curious about the dynamics in the like in the programing language industry, if I can call it that. Because when you say like, we're going to be the successor to Python, like I have to imagine Python is like, hey, man, like we're, we're, we're Python over here.
00;54;10;02 - 00;54;24;24
Geoff Nielson
Maybe like, can't Python also be the successor to Python? Or when you start talking about rust, like is the industry collaborative? We're using the word competitor. I don't know if you have relationships with anybody building these other tools. What's what's that dynamic like.
00;54;24;27 - 00;54;31;08
Chris Lattner
So so it's very funny the way you position it the the programing language industry. So the big PL giants right.
00;54;31;10 - 00;54;33;22
Geoff Nielson
Yeah. All the money programing language.
00;54;33;24 - 00;54;51;17
Chris Lattner
Yeah. All all the people making all the money selling programing languages. Right. Just actually it's a small knit group of people that generally know each other. Right. And so I know Guido for example, and, and we talk when Mojo's first coming on the scenes, we gave Guido a preview before he even launched it as one example.
00;54;51;20 - 00;55;10;06
Chris Lattner
I know and have been working with the rust people and the Julia people. Obviously, I know the Swift people pretty well, right? And all these different things. Right. And so on the one hand, we're all trying to build beautiful things, right? We all want to make our communities our worlds better. So for example, is doing great things and it's expanding the targets.
00;55;10;07 - 00;55;31;25
Chris Lattner
They recently announced Android support, for example, and things like this. And so, and so that's the natural thing to do. And I love it. Right. And I want all these things to keep getting better. And I'm not a like cold people back kind of person. That's never that's never been my approach. Right. On the other hand, each of these things are beholden to their origin story.
00;55;31;28 - 00;55;55;00
Chris Lattner
And so Python has gotten way faster, right? It's it's made tons of technical progress, way more capabilities. But in my opinion, it will never be a fast language. It will never compete with rust or something like that. And there's technical reasons for that, that I'm not going to go into it, but I'll just make a claim that, you know, in the 30 years of Python, if people could solve that problem, they already would have already pretty important, right?
00;55;55;02 - 00;56;12;29
Chris Lattner
Same thing with Swift. Swift is a great language, I love it, lots of blood, sweat and tears put put into this, but it was never designed to run GPUs or accelerators or TPUs or things like this. It just was not designed for that. And so what I, what I look to when I'm building a system is I look to just how to build the best thing.
00;56;13;01 - 00;56;32;27
Chris Lattner
That is always been my North Star. I don't really care about technical novelty. This is why I did not do super well in academia. Like the finding. Finding the the novel contribution to science was never the priority for me. I'm an engineer. I want to build great things people love. I want to get them into production. I want to solve the problems and go.
00;56;33;04 - 00;56;57;08
Chris Lattner
And so the the thing that makes Mojo really special is that it can talk to all of the hardware. It's portable, it's open. It is a very different animal in the kingdom here of programing languages. And what it's doing is, is standing on the shoulders of giants. It's learned a tremendous amount from Rust and Swift and many, many, many Cuda, many, many other systems that have existed before.
00;56;57;10 - 00;57;20;26
Chris Lattner
And dispassionately, what we're doing is we're assembling the best technologies, the best, most modern techniques, and pushing the world forward. And a thing that is a little bit controversial is that we bring great ideas from academia. And so we have new features like linear types and and dependent types and things like this that only exist in functional functional programing languages that we bring forward.
00;57;20;27 - 00;57;37;29
Chris Lattner
And we make it so they're accessible and useful for people. This is something that again, it's been 1015 years ago. But Swift did the same thing and it's era and it brought enums into the world. And everybody was like, oh my gosh, enums are crazy. But that's just because the people writing C and C++ code had never seen them.
00;57;38;02 - 00;57;48;15
Chris Lattner
The academics had had them for a long time in functional languages. And so it's not a novel idea. It's just about engineering, design, architecture. And that's what that's what I focus on.
00;57;48;17 - 00;58;13;22
Geoff Nielson
No, I love that. I love that perspective. It's it's refreshing and inspiring, especially when we're inundated so much with like, these are like, I guess, ego driven stories of like the frontier AI organizations or so much of big tech like you just so obviously bring a love of the game to this, Chris. Like you've got that that engineer mindset.
00;58;13;22 - 00;58;33;19
Geoff Nielson
And it's just nice to hear about working in a space where that is the you know, it's not cutthroat. It's not. Hey, how can we squeeze every last dollar out of this? It's hey, how can we build something really cool and useful? And if we do that, you know, good things will follow. So.
00;58;33;20 - 00;58;51;21
Chris Lattner
And, Jeff, like you, you're hitting the nail on the head here. But let me go back to. This is not just a thing. I spread it out as a startup to make money in six months. Right. This is my life's work, right? I am very honestly and earnestly passionate about everything here. Like I love it, right? And so it's cool.
00;58;51;21 - 00;59;08;25
Chris Lattner
I get paid to do the things I love, but but also what I've found. And go back to some of your earlier points about programmers and motivation and how the world works is that there's different kinds of people in the world in general. And so one of the reasons that I have a string of technologies that have powered the industry.
00;59;08;26 - 00;59;33;11
Chris Lattner
So LLVM, clang, Miller Swift, like built stuff for hardware design. Like there's a lot of things that that have gone on to be pretty big. Things that have helped build other people's systems is because I approach it in this kind of principled way of saying, look, it's going to take more than three months, right? And people will be annoyed, people not to come doing it the right way.
00;59;33;13 - 00;59;55;08
Chris Lattner
That's cool. Like, I don't I don't actually care about them. What I care about is focusing on building the team in the right way. Inspiring, inspiring people, getting people to understand what success looks like. It turns out that that's actually really hard. You say, like, let's build a new programing language to solve all of compute. Okay, that's that's an epic industry scale problem, right?
00;59;55;09 - 01;00;09;14
Chris Lattner
That's not a little thing. That's not that's not a problem that normal people take on. And you go into that with a huge amount of ambiguity of what does success look like? Can we even do it? How are we going to compete against the titans of the industry that have been working on this for decades? Right. These are all big questions.
01;00;09;14 - 01;00;34;29
Chris Lattner
And so what you need to do is you need to break these things down. You need to be very disciplined. You need to have milestone based development. You need to be true to what is actually important to succeed. And then most importantly, be willing to do the work. And what I've seen so much of is a lot of people want to do, and we're surrounded by, by the way, these cheesy demos with these massive over claims.
01;00;35;02 - 01;00;54;25
Chris Lattner
Look, I've solved AI software with agent coding and now all all software is dead. It's like, okay, cool. Well, there's two different theories in the world, right? It could be the agents have destroyed all software value and that nobody's gonna be writing software. Or it could be that you fast forward a few years and we're still writing software, and we're building on platforms, and we need stable platforms.
01;00;54;27 - 01;01;16;18
Chris Lattner
And so to the people that are saying software is dead. I think that the people that are investing in it could be that they're right, in which case nothing matters. Or an alternate view of the world is that maybe platforms still matter and good technologies and architecture do matter. And if so, the people that continue to focus on that are actually going to be, you know, pushing the things that matter.
01;01;16;20 - 01;01;32;09
Chris Lattner
And so I don't know the way the world works, but I choose to stay very focused and make sure that we build things. I try to build it with great people I like to work with. These are fairly old school principals, but but this is what I've seen work to build successful products and teams and architectures and infrastructures.
01;01;32;09 - 01;01;37;00
Chris Lattner
And I think this is how the world actually works well.
01;01;37;00 - 01;01;55;17
Geoff Nielson
And I have a couple of reactions to that. One of them is that it just kind of seems like you are. And I'm not just trying to blow smoke here, but like, you are the right guy at the right time to be building this, right. Like, I don't know, I hope this doesn't come across the knock. I don't know that you could have built this 20 years earlier in your career, right?
01;01;55;18 - 01;02;15;01
Geoff Nielson
Like, you've got the experience. You've got the credentials to say, hey, this is a big problem. It's going to take a while. Trust me, I can make this work. That's right. Right. That's right. And you're not in any one silo where you're incented to not make it work as well.
01;02;15;03 - 01;02;33;04
Chris Lattner
Well, so. So there's two different things going in there. Let me agree with you that me raising money to build modular is much easier than somebody else doing it, right? And yeah, because you're starting to start of a journey. All that there is is ambiguity, right? And so in a face of huge ambiguity, how are you going to solve this?
01;02;33;04 - 01;02;52;11
Chris Lattner
Going to make money selling a compiler. Right. I heard that a number of times. Right. Some people believed in me because they they believed that I would figure it out. Right. And somebody else early in their career, that would not have happened. Right. And so you're you're home percent right about that. But the other piece of it is picking the partners, picking the use cases being strategic.
01;02;52;17 - 01;03;08;05
Chris Lattner
And a lot of that is the experience that I've accumulated across making many mistakes and, and, and, and learning certain lessons in different environments. And so I think that's something I do benefit from. And that's kind of what the investors, you know, we're transitively betting on as well.
01;03;08;08 - 01;03;29;26
Geoff Nielson
It makes sense. And the second piece I wanted to get at is it just I mean, I can tell just by listening to you talk like you have almost like an allergy to bullshit and to hype, right? Like, like just a complete disinterest in, like, as you said, like flashy demos that don't actually make sense. And so, like, we're so just inundated by that right now.
01;03;29;28 - 01;03;43;19
Geoff Nielson
Like how how you know, like how do you, how how should business leaders and tech leaders that are just flooded with these stupid demos, like, how do you how do you separate the signal from the noise?
01;03;43;21 - 01;04;01;25
Chris Lattner
Well, so a lot of my disposition on this is because I've been in AI for ten years. So most people have been paying attention to AI really for last three years, maybe four years, something like that. And so, let me share with you a is been full of a lot of over claims and BS for all ten years.
01;04;01;27 - 01;04;39;26
Chris Lattner
Okay. So what you get is you develop a little bit more of coping, coping abilities, bullshit detection. You get to understand more of, okay, well what is actually a fundamental contribution when people react to it, in which way does it matter? And so you get a better filter. Right? And so, I'm not perfect, I make mistakes, I misread situations, I've like, when Jenny, I came on the scene, I could have reacted three months faster and realized that, yeah, actually, this is a fundamental transformation and and leaned into it, but a lot of it is just again, it's the having an exposed to the froth and the churn and the and all this kind of
01;04;39;26 - 01;05;01;06
Chris Lattner
stuff for a really long time. I also will say I love it though, right? I love that there's so much investment in tech, right? AI for all this problems. What what it is causing is it causes huge amounts of talent and huge amounts of dollars to flood into technology. And we're seeing things move really quickly, not just because of AI, but because of that investment.
01;05;01;08 - 01;05;24;08
Chris Lattner
I think that's great. I think it's fantastic for technology. I think it's fantastic for society. Yes, it makes us all a little bit uncertain, and we have to kind of figure out how to deal with that. But if you if you look past it and you realize, okay, I'm still doing the same thing fundamentally that I was doing a year ago, like all this change, all this progress, all these amazing demos, all these claims, all the gadgets, really.
01;05;24;09 - 01;05;46;20
Chris Lattner
I mean, I'm, I still carry around the phone on a laptop, you know, like everything is changing all the time. But but really the technology adoption goes a human progress. And so the uncertainty goes to AI product progress, speed. And so what we need to do is take a step back and just chill a little bit. And b, keep your eye out for what really matters.
01;05;46;20 - 01;06;07;17
Chris Lattner
And this is where the BBS detection is really key, because so many things get launched as the next coming of whatever. But the incentives are for the people pushing the technologies to try to get that right. And and so they're trying to position it that way because again, they're heavily funded. There's all the different incentives that go into that.
01;06;07;20 - 01;06;21;00
Chris Lattner
And so you see it for those structural reasons. And so once you kind of see the system, you kind of understand why things happen the way they do. There's games being played. There's a whole bunch of stuff behind the curtain. Suddenly it kind of makes more sense. So.
01;06;21;02 - 01;06;41;25
Geoff Nielson
Oh, you're you're preaching to the choir on that one. Like I'm somewhere between skeptical and cynical of that entire machine. But at the same time, you know, I, I work with an awful lot of, like, tech leaders and business leaders right now who, you know, we touched on this earlier, but like, things are moving faster than ever before and they're more complex than ever before.
01;06;41;28 - 01;07;07;24
Geoff Nielson
And I feel for I feel for them. Right. Because there's just like a level of stress involved in that and feeling, you know, you mentioned it like you're getting farther behind and. Oh yeah, oh gosh, I'm, I'm three months behind on a I, I'm, you know, 30 months behind whatever. Like everybody has this anxiety that they're behind and, and they're trying to catch up and it feels like, like we're just getting farther behind.
01;07;07;24 - 01;07;41;28
Geoff Nielson
Or maybe we've hit this inflection point where, like, we as mere mortals can, like, never catch up again. And, you know, you said something that I really liked, which is maybe we just need to chill. And I don't know if I don't know how flippantly you were saying that, but but I wanted to like, I wanted to revisit that in the context of this, like macro increase in, in stressors and complexity, like, yeah, what do we actually need to do to feel more in control, if that's even the right framing?
01;07;42;00 - 01;08;04;14
Chris Lattner
Well, so let me take let me take the alternate position on that. Right. So if you are not evolving, whether you're an organizational leader, a big company or even an individual engineer, you are falling behind like we are. We are definitely in a world where situation is changing. You do need to be evolving. You do need an open mind.
01;08;04;16 - 01;08;25;18
Chris Lattner
If you do the same thing you did ten years ago, you won't be obsolete, but you'll fall behind, right? And other people will go faster. They will evolve quicker and that's a big problem. Now again, the BS is when the next star says, oh, we're going to disrupt all software by agents. And it's like, well, you're not the only people have agents.
01;08;25;18 - 01;08;44;26
Chris Lattner
Everybody has agents. So agents are not the secret source for you. They could be alpha against somebody who doesn't adopt them. But what it means is that everybody should adopt the technologies. They should stay on top of things. And we should all evolve as an industry. The chill part of it, I think, is, is therefore not completely right.
01;08;44;26 - 01;09;04;23
Chris Lattner
You shouldn't wait for the world to pass you by, but also just reduce the amplification. Right. And so you if you're watching Twitter, like every day, there's another hype wave and it's very high amplitude and a lot of the stuff doesn't matter. Some of the stuff does. Right. And so this is where I think we need folks like yourself and other people that can actually unpack this.
01;09;04;27 - 01;09;24;28
Chris Lattner
Like we need we need to we need leaders who invest in new directions, not as a freak out reaction approach, but as a more thoughtful about, okay, well, this is something that is important. Let's do a pilot. Let's go explore it. Let's not put all of our eggs in the basket and act like the world changes. You know, if we don't do this, let's do some of these experiments.
01;09;24;28 - 01;09;47;21
Chris Lattner
Let's do that investment. And similarly for engineers, do not try every model or every new feature or every new wizard thingy that comes across your desk, because you'll get nothing done. On the other hand, if you get if there's tools that get adoption and there's people talking about it, and there are people that you trust and and are credible, then yeah, actually don't don't resist completely.
01;09;47;23 - 01;10;03;28
Chris Lattner
Take a weekend and do something and experiment and do a pet project and try something out now and then. And don't worry too much about is it an open I or anthropic or whatever, because they're so close to most mortals? Can't tell the difference. But we should be adopting like and so we should be moving forward. And so I think that's the trick.
01;10;04;04 - 01;10;23;26
Chris Lattner
The other thing I'll be annoyed about is that in this frothy world, with all these over claims, it can be very difficult to identify fundamental long term technology shifts. It's like Mojo, for example. I'm not annoyed about this, but most people don't know about it, and so they don't know that there's even a thing happening. It's because we're not out there saying it cures cancer or something.
01;10;23;28 - 01;10;43;29
Chris Lattner
But but it is a fundamental thing that makes sense. And what will happen, just like other things I've worked on, is it will slowly gain more momentum. More people will have moments. Leaders like Qualcomm will adapt, adopt things really quickly. It'll go across the entire product portfolio. And and so the people that move quickly and see that and understand what that can mean will get much of the benefit.
01;10;44;01 - 01;11;01;02
Chris Lattner
But or the early scheduled benefit and things like this. Right. But the but this is something that the entire world can use. And so, you know, it will get there. But it's also noisy world. And you have to decide how much are you competing versus how much you're focusing on delivering value and just staying focused?
01;11;01;04 - 01;11;26;20
Geoff Nielson
Is it is that kind of fundamentally the approach that guides you? And I'll just qualify this by saying, I mean, you've created a number of very successful, very highly adopted, you know, products, frameworks, whatever you want to call it. So with Mojo and for other people in the space of trying to get what they're doing adopted, like, is there a secret sauce there?
01;11;26;21 - 01;11;37;14
Geoff Nielson
Like, is it just, you know, build something awesome and wait for it to get discovered? Or do you have other tools in your toolkit for, you know, accelerating the take off?
01;11;37;16 - 01;11;54;19
Chris Lattner
Well, so, so what I, what I care about, just a new insight a little bit is I care about not just getting adopted, but it actually being meaningful and sustained. So I don't want to be the next framework that gets announced, gets a hype wave, gets adopted. But for three months and then three months after that, everybody moves on to the next thing, right?
01;11;54;20 - 01;12;12;24
Chris Lattner
That's cool. That could be really awesome and a great experience to build products. And I'm not saying that's a bad thing, but I like to build things that are actually sustainable and will last for a longer time. And so my approach, which I'm not saying I should follow, but my approach is to be very deliberate, be very focused, try to build something with the real contribution.
01;12;13;01 - 01;12;39;16
Chris Lattner
Do that hard work. Don't just make it possible. Make it feel great. Like developer experience really doesn't matter. But it's also about picking partners like pick partners that can amplify, propel, get you distribution things like this modular join Qualcomm both because Qualcomm would open our stack. That's a huge thing, but also would be a great first customer and would utilize it across this entire portfolio.
01;12;39;19 - 01;12;59;05
Chris Lattner
Well, suddenly now you get exposed to all the users of those products and all these different things, and you don't have to be exclusive to the platform. Just like LLVM wasn't exclusive to Apple, right? But Apple reaped a ton of benefit from adopting LLVM before anybody else, and that's what made 64 bit iPhones happen. And Apple be able to enforce its GPUs.
01;12;59;05 - 01;13;27;22
Chris Lattner
And a huge amount of value was accrued to Apple by doing that. And the investment paid off many full. But you don't have to hold people back. And so part of it is that picking partners, picking opportunities, aligning with the people that have the right incentives. But part of it's also Jeff coming on your podcast and just try and get the message out and sharing with people and actually doing some of that outbound, reaching out and talking to people and trying to try to fit that into the busy schedule that you otherwise get for product design and for other problems we have to solve.
01;13;27;24 - 01;14;01;14
Geoff Nielson
No, that's fair. And I realize it's sort of like a meta question given the context of what's happening here. But, you know, I will tell you, like I will say out loud, like I, you know, I'm, I'm a big fan of yours. And, you know, you've got an amazing track record here, which is why you're here, like to hopefully credentials myself a bit like I nonstop have like PR people being like, you have to talk to this CEO you've never heard of or CTO of this company you've never heard of because they're doing this crazy thing with AI and like all day, like, you know, I'm hitting like delete.
01;14;01;17 - 01;14;09;14
Geoff Nielson
Well, and and you but you've got the, you've got the record and you're doing something really exciting, which is why I'm happy to be part of the, the story.
01;14;09;21 - 01;14;31;26
Chris Lattner
You know, I love what I do. I hope that people find it meaningful and and useful. Right. And there's a lot of different ways to get involved. So we're an open source project. We're fully Apache. We take contributions. We have cool programing language and other people that are contributing to the compiler, but mostly it's people that want to use it to build new packages, new new experiences and applications.
01;14;31;29 - 01;14;50;28
Chris Lattner
If this is interesting, you can get involved. We're also opening up even more of the stack. We have a ton of cool things happening for hardware people that are building new chips. We we love working with people in different ways. And so you can reach out to us on our forums on discourse and discord. I would love to see you on GitHub and we'd love to work with you in general.
01;14;51;01 - 01;15;09;29
Geoff Nielson
That's awesome. And just as we sort of move to wrap things up here, Chris, you know, we've covered a lot of ground here, a lot of really interesting takes and insights. Any sort of parting thoughts or capstone advice for technology leaders, business leaders as they navigate, you know, the future of work in the future of tech?
01;15;10;01 - 01;15;28;10
Chris Lattner
Yeah, I think that you have to keep pressure on, right? I mean, I think that the natural reaction for organizational leaders in particular, is you generally end up with large scale teams. I know I have one to you have some people that are that are advocates and are really pushing hard, and you have other people that are dragging their feet and they don't want to evolve and change.
01;15;28;12 - 01;15;49;20
Chris Lattner
And so I think we as an industry do need to move. I think you should encourage the radicals that are pushing for things. You should you should encourage them. You should give them the right structures that they can make organizational change for, because you don't want kind of the system to to prevent progress. And this evolution that's happening all around us, if we're not careful, it can leave people behind.
01;15;49;21 - 01;16;05;07
Chris Lattner
And I think that all of us want to stay on top of things. We want to be investing in the right way. We want to be efficient. But but I think that that inertia needs to be overcome. And so I recommend that you do that, like stay on top of the stuff, use the tools, play around with some of these things.
01;16;05;10 - 01;16;15;20
Chris Lattner
If you're an organizational leader, it may be a long time since you wrote code, for example. But guess what? You can build a cool app. Like anybody can do this, and so you should play around with the tools. It's pretty fun.
01;16;15;23 - 01;16;25;07
Geoff Nielson
Awesome. Chris, I really appreciate you joining today. As I said, it's been a lot of fun. I really appreciate the conversation. I really appreciate the insight. So thank you so much for joining. Yeah.
01;16;25;07 - 01;16;28;11
Chris Lattner
Well, thank you for having me, Jeff.
01;16;28;14 - 01;16;52;24
Geoff Nielson
Most viewers don't know this, but Digital Disruption is developed by Infotech Research Group, a leading advisor to technology leaders around the world. If that's not you, you don't need to care. So skip ahead and enjoy our content. But if you are a technology leader, Infotech helps IT teams get projects done faster, better, and at a lower cost. Infotech provides unlimited access to practical tools and expert guidance.
01;16;52;24 - 01;17;07;23
Geoff Nielson
You need to execute at a fraction of the cost of traditional consulting, no matter the project. From AI strategy to cybersecurity to vendor negotiation, Infotech has you covered. Check it out at the link below and don't forget to like and subscribe!
The Next Industrial Revolution Is Already Here
Digital Disruption is where leaders and experts share their insights on using technology to build the organizations of the future. As intelligent technologies reshape our lives and our livelihoods, we speak with the thinkers and the doers who will help us predict and harness this disruption.
Our Guest Chris Lattner Discusses
The AI Coding Revolution Has a Dirty Secret
Chris Lattner, creator of Swift and LLVM and co-founder of Modular, joins this week’s episode to talk about coding, vibe coding, and why more code doesn’t always mean better software.
We Discuss
The Jensen Huang Scandal Is Bigger Than You Think
What happens when Big Tech depends too much on one person, and we depend too much on AI? This week we’re covering the Jensen Huang controversy, the backlash against AI data centers, and how AI coding tools could weaken critical thinking.
Our Guest David Weinberger Discusses
AI Has Exposed the Biggest Lie About Humanity
Author, philosopher, and technologist David Weinberger joins Geoff Nielson to explore how artificial intelligence challenges our assumptions about knowledge, human judgment, and business leadership.
Our Guest Jeremy Roberts Discusses
AI Data Centers in Space: Stupid or Genius
Are AI data centers in space a practical solution for the future or just an expensive fantasy?