How Ricursive Intelligence’s Founders are Using AI to Shape The Future of Chip Design

Anna Goldie and Azalia Mirhoseini created AlphaChip at Google, using AI to design four generations of TPUs and reducing chip floor planning from months to hours. They explain how chip design has become the critical bottleneck for AI progress -- a process that typically takes years and costs hundreds of millions of dollars. Now at Ricursive Intelligence, they're enabling an evolution of the industry from “fabless” to "designless," where any company can create custom silicon with Ricursive Intelligence. Their vision: recursive self-improvement where AI designs more powerful chips, and faster, accelerating AI itself. Hosted by Stephanie Zhan and Sonya Huang

• 0:00Speaker 00

  Right now, we can't have much of co-design between chips and models because of this asymmetric design cycle for chips. Because it takes so long, it takes so much time to design chips. The cycle is there's a mismatch between how fast we can create the next generation AI methods and how fast we can build the next generation chips. But if we can make our chips much faster, then we can enable this co-design and co-evolution of workloads, applications, and chips altogether.

• 0:49Speaker 01

  This week on Training Data, we explore the biggest bottleneck holding back AI, compute, and the design of chips themselves. Our guests are the founders of recursive intelligence, Anna Goldie and Azalia Meherzani, the team behind Google's pioneering AlphaChip project that helped design four generations of TPUs. They're now applying AI to the entire chip design process, transforming the industry from fabulous to designless. Anna and Azalia paint a picture of what a Cambrian explosion of custom silicon means. We'll explore the holy grail topic of recursive self-improvement, how AI-designed chips unlock radically creative new chip designs, and why democratizing chip design could accelerate the path to superintelligence. Enjoy the show.

• 1:35Speaker 01

  Thank you so much for joining us. We're so excited to celebrate the announcement of recursive intelligence and congratulations on embarking on this new adventure.

• 1:43Speaker 02

  Thank you.

• 1:44Speaker 00

  Thank you so much.

• 1:45Speaker 01

  To kick off, you've said that chip design is the compute bottleneck to advancing AI. Can you share and paint a picture of what's happening now? What are all the bottlenecks to chip design today?

• 1:57Speaker 03

  I mean, I think when we were saying that, we were kind of alluding to this motivation that we had originally with our moonshot, which was this observation that, you know, neural networks, these concepts that have been around for decades. But the AI that came out of it wasn't that effective until we had like more powerful computer systems and chips. And we thought that now that we had very powerful AI systems, we could use those to kind of tackle the bottlenecks in chip design and, you know, create basically more effective compute. So if you see these like scaling laws, basically the more compute you apply to training a model or inferring with a model, like the more intelligence you get out. And we're seeing that there is this mismatch, like we're using, say GPUs are like originally designed for graphics processing.

• 2:42Speaker 03

  But we're somehow repurposing them for crypto and then for training neural networks models. I guess they're good at like large matrix multiplies. But like if we could more better co-optimize the models and the hardware, we could get more effective compute. And then we could push ourselves out on that, those effective scaling laws.

• 3:00Speaker 04

  You were the co-creators of AlphaChip at Google, which I think is used in four successive generations now of TPUs. Maybe take us back to that project. How did it get started? What were the key results you drove?

• 3:12Speaker 00

  Yeah, so we started the project in 2018. And we had some earlier version of placement that placement work that we were interested in solving. And that was not about chip placement, it was about compiler and basically mapping neural networks to chips. So we did that project, we got great results, we published and we were like, what are the like highest impact projects that now we can kind of push this further and have real world impact? And that's where the chip placement came in the picture. And we started the project working very closely with the TPU team at Google.

• 3:55Speaker 00

  Because back then we didn't know much at all or at all about how chip design is, how the process is. So we worked very closely with them from very early on. And at the time, it's interesting because at the time, AI was like, was still like very talked about, but nothing like the scale was nowhere near what it is today. So we had to really work through things and showing with showing them data and like. Constantly like iterating over our approach and hearing them, what they needed from us, listening to them, iterate through the process and eventually we bent from a research concept on chip placement all the way

• 4:39Speaker 00

  to something that was actually used in product and we could tape out and all that.

• 4:45Speaker 02

  I remember one of our very first meals together, when I, it hit me that you were so customer obsessed and had so much empathy for the customer. And I asked you how and why? And you said, well, I've been working with the TPU team as our customer internally for so many years. So I think that really kind of gave you that other perspective. What were the TPU team's early reactions to what you had created?

• 5:10Speaker 03

  I mean, they were like super skeptical. For example, like Azali and I, we were researchers, like we had read a bunch of research papers and we're like, okay, there's this half perimeter wire lanes, that's what academics report results on. So like, we made this RL agent that could optimize half perimeter wire lanes for a placement. And then we were kind of excited about sharing that with them. And we showed these results and they were like, actually kind of like angry at us, like, why are you showing us these results? Like, we don't care about half perimeter wire lanes, like we want like routed wire lanes, congestion, like horizontal and vertical congestion, timing violations, power consumption, area, like. And so, you know, we we just kept listening to them and we were like, okay, so what is the thing that matters?

• 5:55Speaker 03

  I think another part is just to make your customer feel like they're part of it too.

• 5:58Speaker 02

  Yes.

• 5:59Speaker 03

  So for example, we worked with them to create the cost functions that they cared about, approximations of those. So like Mustafa, who was on the TPU team, developed these very fast, like congestion cost function. And then we could optimize against the end density and then wire lanes as well. And then we show results on those and then we run the commercial tool and we show that this actually correlates to good results on the metrics that they do care about.

• 6:25Speaker 02

  Yeah.

• 6:25Speaker 04

  So I want to talk about floor planning and I come from an EDA family. You know, and floor planning is, it is the crown jewel of EDA. Can you say, just for our listeners that don't come from the chip design industry, what exactly that is and how your technical methods help solve it?

• 6:41Speaker 00

  So floor planning is a process where we place the components of a chip onto the silicon. And the complexity of this process comes from the large size of the input problems. Like for a block of a chip, not the entire chip, like a single block in a chip, the graph that we are optimizing has, could have millions of nodes. And all of these nodes need to be placed and routed. And we need to make sure that all these constraints that earlier we were talking about, like PPA, power area performance, are optimized and all the physical constraints are met.

• 7:22Speaker 00

  So the placement problem has to do with how we can do this such that all those constraints given by these, like very small technology node sizes given by TSMC and others are met. So we are dealing with a combinatorial optimization problem that's large scale, but also all the metrics are hard to evaluate and measure.

• 7:45Speaker 04

  Got it. And this traditionally was done together with EDA software, like a cadence service synopsis, right? How does your method differ from the classical methods?

• 7:54Speaker 00

  So what we developed here was a learning-based approach to the problem, where we would train a reinforcement learning agent that would try out different ways to approach the placement problem. And it would learn through interactions with this environment that we built to learn from the positive placement and also the negative examples, and iteratively improve itself. So one of the main differences between our approach and all prior approaches were this ability to learn from experience, which would enable the model to self-improve.

• 8:34Speaker 00

  Just like a human expert that becomes better as they solve more instances of the problem and harder problems, our agent was able to kind of exhibit similar behavior. And that was a very major kind of delta and change between what we had and prior approaches.

• 8:57Speaker 04

  Yeah. And I remember with, I mean, if you take AlphaGo as an analogy, Move37 was kind of shocking because it was just so different from how a human player would move. Are you seeing something similar happen with floor planning and chip design?

• 9:08Speaker 03

  Yeah, we saw these very strange like curved placements.

• 9:12Speaker 02

  Ah.

• 9:12Speaker 03

  So there are donut shapes as well. I think humans would tend to make the macros very, so macros are memory components. And they're larger and they make them very aligned and then they would put logic maybe in the middle. And then the wires would connect all of these components, but. If you make the shapes curved, you can reduce the wire lanes, which can reduce the power consumption and timing. Violations, but it's just the complexity of making this curved placement was like mo- beyond what. Humans would have wanted to take on or it felt risky to them. So yeah, these aliens.

• 9:42Speaker 02

  So.

• 9:44Speaker 02

  What was the moment you realized that this had the potential to transform the entire end to end process, not just the floor planning stage? And when you decided that it had the potential to really be a company?

• 9:56Speaker 00

  There, there were a few like milestones that uh first was like showing that it works, that it gets the superhuman results. On some blocks. And then later on, the chip was actually taped out and it came back and it was actually working, which is. A big, a big uh kind of milestone. Uh because it can be really, really costly if for some reason our AI missed something. Um so those were big moments, uh the real impact and also on the algorithmic side when we saw this kind of self improve. Improvement properties and the fact that models were, our models were becoming better as they solve more problems like that, that is like very.

• 10:42Speaker 00

  Cause when we get to that point, then there is no stopping for the AI. They can see more data points and solve more instances of the chip. Optimization problem than any single human can ever do. So that's, that's the moment that you're like, yes. So there's a lot more potential here, but um AlphaChip was targeting a. Module, a, a part, a part of a physical design. Uh but chip design, as you know, there, there are many more to uh like stages to it and many more components. And right now, uh we think that.

• 11:19Speaker 00

  Everything from the state of AI and the capability of AI, both on LLM side and other like graph optimizations and other approaches. And the way that we can really scale up these algorithms and enable synthetic data, very large scale distributed computing. Everything is ready to tackle the end to end chip design optimization problem. So that's why we're excited about doing the company right now.

• 11:45Speaker 04

  Super cool. How do you think about getting the right training data for this market? I think it's, you know, it is kind of the binding constraint to AI in many ways. And is this a market where you can have, you know, enough synthetic data or self-play to be able to bootstrap?

• 11:59Speaker 03

  We're excited about synthetic data. So obviously there's some open source data, but it's not that interesting or meaningful. And, you know, customers are willing to share data with us, but we don't want to train our models on that because we want to keep their data private and siloed from each other. But synthetic data is where we think there's the most promise. Like as Ali and I, we've been working on synthetic data approaches and for LLMs in various code domain tasks across like Claude and Gemini. And we see a big opportunity to like get a scale of data that would go far beyond what any customer could ever share with us. Like orders and orders of magnitude more.

• 12:36Speaker 04

  With the TPU program, um, I guess at what moment did the chip experts stop doubting you? And as each successive generation came out, what was the progressive impact of what RL was able to drive in those designs?

• 12:50Speaker 03

  I mean, I think that our approach with the TPU team is like every week we would show them data over and over again.

• 12:56Speaker 04

  Every week?

• 12:56Speaker 03

  Yeah, for every week for like a couple of years. Yeah. Cause I mean, the stakes are very high here. Like if there's something wrong with this layout, like the setup on the TPU team is like this. There are human, the, the TPU is quite large and complex. So they divide it up into dozens of blocks. And then each block is owned by a human or a team of humans. And that's their responsibility to get this block right. If anything goes wrong, like it's their fault, right? And so they would generate their own layouts in collaboration with commercial tools. And then we would show them our layout, some AI generated layout. It would look weird. It would be curved. And they would, you know, they would have to say, okay, like I'm picking this AI generated layout over my own layout.

• 13:39Speaker 03

  And I'm taking responsibility for that. I think, yeah, it requires a lot of trust. We have to be better in every single metric for them to choose that. And we saw across each successive generation of TPU that we were being adopted in more and more of the chip block and more of the area, and also that we were getting increasingly superhuman performance. And we published an AlphaChip blog post, I think September last year, we showed this curves.

• 14:04Speaker 04

  Can I say a word on that? Yeah, on the blog post and the superhuman performance?

• 14:08Speaker 03

  Yeah, I think, like, just that in across every single generation that we were used, which I think there were three generations shown in the blog post, but actually we were used in another generation after that was published. Every single one, we were seeing more of the area and more, like, increasingly superhuman performance.

• 14:27Speaker 00

  Yeah, so basically the delta between AlphaChip layouts and the baseline layouts are also

• 14:32Speaker 03

  Growing.

• 14:32Speaker 00

  Growing, which is like, it's a property of AI and how it scales with data.

• 14:38Speaker 03

  It makes sense, right? We're, AlphaChip was trained on more and more TPU blocks, so it gets better.

• 14:44Speaker 04

  Why is the company called recursive?

• 14:46Speaker 03

  Um, because we're recursive, we're AI for chip design and chip design for AI. Uh, so recursive self-improvement. In terms of like the name, the spelling of the name. Um, so the name itself is recursive, so the initials R I recursive intelligence are the first two letters of the company's name.

• 15:05Speaker 04

  What does recursive self-improvement like, like, why, why does that matter for, you know, the broader AI race? Like, just say a word on, I think this is, this is a root node problem, just say a word on it.

• 15:15Speaker 00

  Yeah, so A, uh, chips are the fuel for AI. And scaling laws are driving much of the progress in AI, whether it's on pre-training, post-training, test time and all that. And so the faster we can make chips that are more custom or better designed for the AIs that we run, uh, the faster we enable, uh, this efficiency, more efficient kind of compute. And that kind of bends the curve for our scaling law. So that means we get to the next generation of AI uh, faster and we can gen, create, design better AIs faster. And so this is a type of recursive self-improvement that we want to see because our AIs then

• 15:59Speaker 00

  Can help our chips become better and we can design them faster and so on. So that is the recursive self improvement loop that we are going after.

• 16:10Speaker 02

  One of the visions that you have is to transform the industry from just fabulous to designless. What does that mean?

• 16:17Speaker 03

  So fabulous was basically this concept of it used to be that people thought that no serious chipmaker could exist without their own fab. This was like, obviously before like this multi trillion dollar companies like Nvidia. You know, TSMC basically created this whole new world where we could have these incredibly valuable fabulous companies. So we think that there's an opportunity to create incredibly valuable companies that don't need to have their own in-house design teams. So many companies, they serve these models at massive scale. They're spending like, you know, this, I think a hundred billion dollar plus on AI inference alone like this year, and it's growing rapidly. So companies would benefit from like maybe custom chips to serve their models or train them.

• 17:01Speaker 03

  But that requires enormous teams of like hundreds or thousands of human experts in-house. And we don't think that's necessary. So we want to move towards, yeah, designless paradigm.

• 17:13Speaker 02

  It's fascinating. We're seeing it come true with obviously Google and TPU's, Amazon with Trainium, also OpenAI and Broadcom and, and maybe even Tesla.

• 17:25Speaker 04

  Do you think the future is then, you know, even today, there's almost like model, model, application co-design? Do you think there's going to be chip application co-design and even individual companies will have multiple chip architectures as a result?

• 17:38Speaker 00

  Yes, we think that there is, we are going to see more and more co-design across the deep learning or AI stack from modeled and data to software and all the way to the chips. And we are going to enable that and co-design is really the secret or the path to more efficiency and more performance going forward.

• 18:03Speaker 00

  And uh but right now, we can't have much of co-design between chips and uh models because of this asymmetric uh kind of design cycle for chips, because it takes so long, uh so mu- it takes so much time to design chips. The the cycle is, there's a mismatch between how fast we can create the next generation AI methods and how fast we can build the next generation chips. Um but if we can make our chips much faster, then we can enable this co-design and co-evolution of workloads, applications, and chips altogether.

• 18:41Speaker 02

  One of the things that I loved that we talked about early on was that um the value you bring isn't just on reducing the cost um it takes to actually design a chip or the speed at which you can accelerate the chip design process, though that is transformative in itself. It's actually to unlock completely new potential applications and maybe custom silicon as well alongside it. Um can you share a little bit about that? What is the Cambrian explosion that you might expect?

• 19:10Speaker 03

  I mean, we think computing is going to be increasingly ubiquitous in every aspect of our lives. So like, obviously, there are things like AR, VR, there's like maybe chips in space, even like hearing aids, like there are experiences that aren't possible unless you can serve them at like sufficiently low inference, like latency, or at like low enough power. And we think that like custom silicon can enable these applications.

• 19:35Speaker 00

  Yeah, I do think that AI is going to be everywhere in every kind of experience, every every aspect of industry and life going forward. And there are chips that would enable these AIs to run. And given the scale, we we would want them to run as efficient, uh very efficiently and uh low power, high speed, all of that, and custom silicon is really gonna enable that. Um and we want to enable custom silicon for for basically any workload that is being run at sufficient scale.

• 20:12Speaker 02

  Since we have the ear, whoever's listening on the podcast, who might some of your ideal customers be?

• 20:18Speaker 00

  Um. There are a range of customers that we are envisioning. The, the, in the first phase of the company, when we are, when we are building ways to accelerate, dramatically accelerate the chip design process, our customers would be chip designers, like Nvidia, AMD, ARM, MediaTek, and all, all of these companies that are already designing their chips, and all of them would want a faster design cycle, and we can help them with, with this technology. But we, we don't want to stop there.

• 20:56Speaker 00

  We, we want to enable any customer that has a workload or is family of workloads that they want to run, that they're running it at sufficient scale, and they would benefit from a, from custom silicon. We want to enable them to have that without having, having teams of hundreds to thousands of chip designers in their companies. Like, so we, I think that kind of goes back to this Cambrian explosion of chips that we can enable.

• 21:29Speaker 04

  I'd love to talk about the kind of existing incumbents in chip design. So like, I grew up in a family of Cadence. Mom and dad both worked at Cadence, and I think chip design is a duopoly between Synopsys and Cadence. And I think they're adding AI to their product suites. How do you see your company playing out versus the incumbents adding AI?

• 21:54Speaker 03

  I think that we see ourselves as like sort of coming from the opposite direction. Like, we're a frontier AI lab. We come from like, you know, Google Brain or like Anthropic, those kind of like backgrounds, and we want to kind of rethink or reimagine how chip design can be done. And we, we think that fundamentally, in order to be able to, this isn't like a point solution thing where we want to replace some module one at a time with AI. We want to like reimagine and co-optimize different stages, and we think that an AI approach is necessary here.

• 22:25Speaker 04

  What do people from the chip design industry think of you guys? I would imagine there's a range of like from excitement to extreme skepticism, extreme excitement to extreme skepticism.

• 22:34Speaker 02

  Or extreme fear.

• 22:38Speaker 04

  Yeah, what do the chip design folks think of you all?

• 22:40Speaker 03

  I think a lot of them are excited to work with us as like potential customers and we're excited about them too. And a lot of people are excited to come join us and work together. But yeah, of course, like what we're doing is very ambitious, so I'm sure many people are skeptical too.

• 22:55Speaker 04

  Let's talk about that for a little bit, like there was, you know, there's some internet uproar. You know, micro niche internet communities love getting spicy and there was like some spiciness around AlphaChip and like people from media saying like, ah, it's not, it's not really real for XYZ reasons. Like what, what do you think was behind that and what do you think was valid in that criticism? And what, what do you think is the important thing that they weren't realizing?

• 23:18Speaker 00

  So whenever AI goes to a new field and does something disruptive,

• 23:26Speaker 00

  we would see reactions like that. And this is not necessarily just, just for in our case, it appears almost in every other.

• 23:35Speaker 03

  The bitter lesson.

• 23:37Speaker 00

  And we think the bitter lesson is part of it. Like we are true believers in the bitter lesson and it's usually a little challenging to kind of like

• 23:49Speaker 00

  accept that a whole new technology or new way of looking into problems that people have been, have worked on for decades is just coming and is solving things more end to end. And, and these people, like in this case, Anna and I and the AlphaChip team, where we were coming from not, from not an EDA background. At least we had not worked on that, that, that problem space. And now we could come, come up with problems that were, with solutions that were. Being productionized and all that. So, so there's always this kind of a reaction in the beginning.

• 24:28Speaker 00

  But somehow the true kind of impact of our work was was much bigger than just the problem that we solved. And I can mention some of those, for example, this bringing, like doing reinforcement learning, looking into this graph neural net optimizations were applied to many other problems across chip design, including a best paper award in at DAC in 2023.

• 24:57Speaker 03

  Nice.

• 24:57Speaker 00

  The first author of that paper is now is now our teammate at Recursive. And that was for physical design, we saw there its adoption in other stages of chip design, like synthesis and so on. The work brought a lot of like attention to AI back in 2020 in in the chip design field. And that was, we think it's like one of the most impactful thing as a result of our project. Right now there are conferences, like we, we actually, we are involved in one of the LLM aided design that are just, just focused on AI approaches to chip design, which back in the days was not, not a thing.

• 25:38Speaker 00

  So, so we are, there are so many positive things happened as a result that you're grateful for that and.

• 25:46Speaker 03

  I had a funny interaction at like ML for Systems Workshop, which is this another conference that we had started in like 2018. And the person was like, oh, AlphaTip inspired my entire PhD thesis. And I was like, oh, that's wonderful. And I was like, so I'm just curious, like, how did you first come across the paper? And he was like, oh, because of the controversy.

• 26:03Speaker 00

  Yeah.

• 26:05Speaker 03

  Very amusing. I think the thing that surprised me a bit was I thought that the people who would be upset by the work would be like physical designers whose jobs were at risk. Right, and those, and you know, those physical designers were skeptical. It required a lot of data for them to be able to change their mind and accept that our method worked. But they actually weren't the people that were upset. It was people who had developed prior methods in the field. And I think in retrospect, that makes sense. There's something painful about like you pour like your like time, your like your human ingenuity or like soul a bit into these methods. And then people come from outside your field. And then they're using sort of, in some sense, simpler approaches, like they're using things at scale with data and compute.

• 26:48Speaker 03

  And you're outperforming, it's like a little, it's painful.

• 26:52Speaker 04

  Totally.

• 26:52Speaker 03

  But you know, everyone, we can all, you know, build on top of this work.

• 26:56Speaker 04

  Yeah, we can all adapt.

• 26:57Speaker 03

  Yeah.

• 26:58Speaker 04

  Can I ask you the opposite question? So I love the bitter lessness to answer. I then have the opposite question, which is, well, the large language models became so good at coding, for example. Why won't they just naturally become good at chip design? Why does there need to be a specific chip design frontier lab?

• 27:19Speaker 00

  Yeah, so chip design has a lot of components and some of them are language and related to language and code. But actually, a big chunk of it is not related to language and code. And it has to do with this different properties and graph structure of the of the chip. And all these constraints that appear, like these are like really, really hard, large scale, computational optimization problems that required a specific, like the custom way of, like specific kind of approaches from an AI perspective.

• 28:00Speaker 00

  And our approach here is to apply the right method to the to the to every problem. And L L M's, we love them so much and we're gonna use them extensively. We are gonna build L L M's that are really useful for some stages of the process. But they're not gonna be sufficient for all of this.

• 28:22Speaker 00

  Um. So we are gonna have our own AIs and specific optimizations for different stages. And those are really important because uh unless we design them, those those uh kind of modules that are really fast, really optimized, we can't iterate around them with L L M's or with other methods. So so our approach is gonna be a very hybrid uh kind of multifaceted uh AI that is gonna after this. Uh but we are true believers in AGI and

• 28:57Speaker 03

  bitter lesson.

• 28:57Speaker 00

  bitter lesson and we're we just want to be on the frontier of that with better chips.

• 29:03Speaker 02

  What do you think a world with AGI looks like? It's so hard to imagine the world in ten years or even five, but what what do you imagine for whenever that future is?

• 29:15Speaker 03

  I guess AGI, like, what does that mean? Human level for everything? Like, maybe you just have many employees that are just effectively, you know, a bunch of compute powering them. I think eventually, like, you know, maybe we can all take a vacation, but.

• 29:30Speaker 00

  It's just uh so it's gonna be extreme, like, uh cases of productivity for every single human being. Um and as a result, we can create a lot of uh economical value uh at a scale that is not possible today and hopefully distributed at a scale that is not possible today. So uh I I have a very optimistic view or view about future about the future where we when we have AGI. It's just lots of good things are gonna happen and if you're careful about it, we can distribute the value to to all humans. And and that's great.

• 30:12Speaker 02

  We're gonna have data centers in space.

• 30:14Speaker 03

  Yes, why not?

• 30:15Speaker 02

  And is that gonna require, actually, is that gonna require custom silicon? Because it's like a different thermal footprint, right?

• 30:20Speaker 03

  Yeah, that's great.

• 30:21Speaker 00

  Yeah, there are other, there are different requirements in space from temperature, from like a kind of resistance of the chip and so on.

• 30:28Speaker 03

  Yeah, even like working on the Pixel phone, they have different corners. So like, the phone has to be robust to different temperatures, like different voltage settings. And so like, yeah, you just have to increase in many more corners, I guess, for these chips.

• 30:41Speaker 04

  Yeah, that's a good example, actually, of like where the different properties of, you know, different heterogeneous compute, you have different chips, different use cases. I think chips in space, right now, like the whole data centers in space, you know, field is structured around how you make existing GPUs perform actually in space. But I think if everything for the pre-raid for your company works, there's going to be specific compute for space. Yeah.

• 31:06Speaker 00

  Customization for space, in this case.

• 31:08Speaker 04

  Yeah.

• 31:08Speaker 00

  Yeah.

• 31:10Speaker 02

  So you've already become incredible talent magnets in just a couple of weeks since incorporating the company. We have Jiwoo, Echan, Dan, Ibrahim.

• 31:22Speaker 02

  What else are you looking for in terms of talent?

• 31:25Speaker 03

  I mean, we're definitely building out on the technical side. So every stage of tip design, we're looking for top talent, but also on the LLM side. So this is like, we're a frontier AI lab. So you want people all the way from like pre-training, mid-training, post-training, RL training, you know, experts in evaluations, data. And also on the non-technical side, we want, you know, operations specialists who can help unlock us, you know, like recruiting, like Chief of Staff, like that kind of thing.

• 31:56Speaker 02

  What do you hope to have accomplished in a year?

• 31:58Speaker 03

  Oh, we're going to release our first product by a year, but yeah.

• 32:03Speaker 00

  Yes.

• 32:04Speaker 03

  Strong commercial partnerships.

• 32:06Speaker 00

  Yeah, strong partnership and we want to have our first product in that timeline that we can offer broadly, not just to our partners, but more broadly to other chip designers.

• 32:20Speaker 02

  Anything you're willing to share in terms of what that first product might look like?

• 32:23Speaker 00

  Yes, it's going to accelerate the process. It's going to tackle the long holes in chip design, and it's going to be more end-to-end than the products that we have access to today.

• 32:34Speaker 02

  Yes. How do you see the role of a human engineer evolving in a world where the design process is automated? What do they end up doing instead with their time and talent?

• 32:46Speaker 03

  I mean, I think there's like, our company goes through various phases. So like the answer to that question maybe changes over time. But, you know, one way to think about it is like, you know, these engineers can come work with us and help us reimagine the process. So that's one way.

• 33:02Speaker 03

  The other thing is like, it's sort of like as humans start working with like Claude code or cursor. You know, maybe they're not doing as much hands-on coding, but they're becoming amplified and becoming much more productive through these AI tools. So something like that. Like, for example, even just AlphaChip, we could generate these like Pareto optimal curves of like different trade-offs. Like, every one of these points would have taken humans like weeks to generate, but we could generate many, many of them because it's it just takes so little time and compute for this model. So like, human can like explore like, what is it we really want? What trade-offs we care about?

• 33:35Speaker 02

  So we're gonna vibe code chips, that's what I heard.

• 33:38Speaker 03

  Maybe that's not quite, but yeah.

• 33:43Speaker 02

  So you've worked with some of the greatest of all time. Um, Jeff Dean, Noam Shazer, many, many more, uh, very closely, um, Coakley and others. What are some of the lessons that you've learned from them or experiences that have really shaped you? And and how they shaped your perspective today?

• 34:02Speaker 00

  So I I think with all of these, uh, people that you interact extraordinary. So just setting the bar really high. And, um, all of them are also extremely passionate about what they're doing. And that's how Anna and I feel about this company. Like, we are so excited to solve this problem, no matter what. So that's something that we think is very important to actually, uh, be successful. And, um, the other thing about them is just, they're just really nice people.

• 34:37Speaker 03

  Yeah.

• 34:37Speaker 00

  And that setting that nice c- c- culture, collaborative cu- culture, but when where everyone is heard and, uh, everyone can do their best work. And that's like a very important, important thing that we want to, uh, follow their roots as well.

• 34:54Speaker 03

  That's right. I I think like Jeff, for example, like incredible breadth, incredible depth, incredible speed, but also like very high integrity. Like, he treats every single person with like, respect. It doesn't matter if they're like president of the universe versus like they're an intern or something. Like, it's equal like treatment. And I think that's like, we want to embody that too. And I just remember like Jeff, he actually gave me like a performance review, the corrective feedback or something. He was like, ask for more from other people. And so I wanted to, yeah, expect more from others too. Hmm, I like that.

• 35:33Speaker 02

  I like what you shared. I saw all these photos of Jeff at NURIPS, running with like many people,

• 35:42Speaker 02

  which I think perfectly encapsulates what you just said. He welcomes others in and he, you know, motivates them and spends time with anyone, no matter what.

• 35:52Speaker 03

  I remember Quoc also, he was our manager almost for 10 years. He was telling us like, deep learning makes all of us Renaissance people. Like we can make contributions in many different types of fields. And I think that's true here as well. And he also said, like, have fun in this company. Like actually enjoy. And I think, we think that's important too. We think if everyone's having fun, then we're going to do even better work together. I love that.

• 36:16Speaker 02

  All right. Thank you so much, Anna Zalaya. We're so excited for your first year at Recursive. And we're really, really excited about it being the frontier lab for AI chip design.

• 36:27Speaker 03

  Thank you so much.