Pod Insights Training Data Fireworks Founder Lin Qiao on How Fast Inference and Small Models Will Benefit Businesses About this episode In the first wave of the generative AI revolution, startups and enterprises built on top of the best closed-source models available, mostly from OpenAI. The AI customer journey moves from training to inference, and as these first products find PMF, many are hitting a wall on latency and cost.
Fireworks Founder and CEO Lin Qiao led the PyTorch team at Meta that rebuilt the whole stack to meet the complex needs of the world’s largest B2C company. Meta moved PyTorch to its own non-profit foundation in 2022 and Lin started Fireworks with the mission to compress the timeframe of training and inference and democratize access to GenAI beyond the hyperscalers to let a diversity of AI applications thrive.
Lin predicts when open and closed source models will converge and reveals her goal to build simple API access to the totality of knowledge.
Hosted by: Sonya Huang and Pat Grady, Sequoia Capital
Mentioned in this episode:
Pytorch: the leading framework for building deep learning models, originated at Meta and now part of the Linux Foundation umbrella
Caffe2 and ONNX: ML frameworks Meta used that PyTorch eventually replaced
Conservation of complexity: the idea that that every computer application has inherent complexity that cannot be reduced but merely moved between the backend and frontend, originated by Xerox PARC researcher Larry Tesler
Mixture of Experts: a class of transformer models that route requests between different subsets of a model based on use case
Fathom: a product the Fireworks team uses for video conference summarization
LMSYS Chatbot Arena: crowdsourced open platform for LLM evals hosted on Hugging Face
00:00 - Introduction
02:01 - What is Fireworks?
02:48 - Leading Pytorch
05:01 - What do researchers like about PyTorch?
07:50 - How Fireworks compares to open source
10:38 - Simplicity scales
12:51 - From training to inference
17:46 - Will open and closed source converge?
22:18 - Can you match OpenAI on the Fireworks stack?
26:53 - What is your vision for the Fireworks platform?
31:17 - Competition for Nvidia?
32:47 - Are returns to scale starting to slow down?
34:28 - Competition
36:32 - Lightning round
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 We thought replacing other frameworks as library with PyTorch is to be simple. It's just swap the library, how hard can that be? But we realized it's, we thought it's just a six month project. It turns out to be a five year project for us to support entire Meta's AI workload building on top of PyTorch because we have to rebuild the whole entire stack from scratch, from ground up because we have to think about how to load data efficiently, how to distribute inference in PyTorch efficiently, how to scale training efficiently. And then we end up rebuild the whole entire inference and training stack on top of PyTorch.
0:46 Speaker 00 When we left, it was sustaining more than five trillion inference per day.
0:51 Speaker 00 So that's a kind of massive scale by two to five years. And the fireworks mission is to significantly accelerate time to market for the whole entire industry, compressing it from five years to five weeks or even five days time to market. So that's our mission.
1:24 Speaker 01 Joining us today is Lin Tiao, founder and CEO of Fireworks. Lin is an AI infrastructure heavyweight who previously led PyTorch at Meta, which is the backbone of the entire global machine learning ecosystem. She's taken her experiences at PyTorch in order to build fireworks and inference platform for generative AI. We're excited to ask Lin about the market trends behind AI inference and how she plans to support and even accelerate the market shift to compound AI systems at Fireworks.
1:54 Speaker 02 We're thrilled to have Lynn, CEO and founder of Fireworks with us today. Thanks for joining us, Lynn.
1:58 Speaker 00 Thanks for having me.
2:00 Speaker 02 Well, we're really excited to talk about a lot of things with you today, from PyTorch to the small model stack that you're building, to what you're seeing in terms of enterprises building production deployments. But before we get there, can you maybe say a sentence or two on what you're building at Fireworks?
2:15 Speaker 00 Yeah, so we started Fireworks in 2022. And Fireworks is a SaaS platform, first and foremost, for general AI inference and high-quality tuning, especially using our small model stack. We can get to very low latency for real-time applications, very low cost for sustainable business growth, and customization, automated customization for tailored, high-quality enterprises. So that is Fireworks.
2:44 Speaker 02 Wonderful. I want to maybe start with the PyTorch story. PyTorch is kind of the foundation upon which the entire AI industry runs today. And you and Dima and some of your other co-founders were integral and leaders of that project at Meta.
3:02 Speaker 00 So think about PyTorch as a programming language for digital brains.
3:08 Speaker 00 And it's designed for researchers to very easily create those digital brains and experiment with it. But the challenge of PyTorch is it's very fast for people to create various different deep learning models, the digital brains, but the brains don't think fast enough. So that's a challenge I took on to address while I was head of PyTorch.
3:32 Pat Grady? And you mentioned, you've mentioned before that most of the companies that are trying to build something similar to what you're building at Fireworks have chosen to be framework-agnostic, whereas you very much made a big bet on PyTorch. Can you say why make the big bet on PyTorch? And what benefits that brings to your customers?
3:50 Speaker 00 Um, that is really based on what I see uh when I operate PyTorch and Meta also across the industry. And I clearly see a fernal effects that PyTorch is because it started that as a tool for researchers. Uh it start to dominate dominate the top of the fernal for model creation. And then the next stage of fernal is people doing applied production work. They take those research models and test it out for production setting and try to validate that the hypothesis. And then feed into production. So that's clear fernal effects that's happening. Um, and as PyTorch is did that for researcher, it take over the top of the fernal.
4:33 Speaker 00 And it's really hard for people to rewrite into other frameworks for production. And naturally, it just flow down towards the bottom of fernal. And that's how PyTorch become dominant. And I'm start to see um more and more models, especially more nascent models are all built in PyTorch and run in PyTorch in production. Including the general AI models. That's why we only buy on PyTorch and we don't want to distract ourselves to support other things.
4:59 Pat Grady? So researchers researchers like it and it flows downstream from there. What do researchers like so much about PyTorch?
5:05 Speaker 00 Simplicity. Uh simplicity scales. And that's kind of lesson learned. We um through the journey of PyTorch and Meta. Uh and also building on the community, it is a relentless journey to focus on simplicity. And we have a constant seeking how to make the user experience simpler and simpler and hiding more and more complexity in the backend. For example, when I started this journey at Meta, there are three different frameworks. Um Coffee two for mobile, Onyx for server side production, PyTorch for researchers, too complicated. And the the mission is to not reduce three frameworks into one to simplify. But it's actually a mission impossible.
5:49 Speaker 00 After I consolidated all three teams and there's no consensus to how to simplify and build this one stack. And we took a very idealistic approach and take the PyTorch frontend and take the Kappa two backend and we said we're gonna zip them together. It seems simple, but it's very hard to do because these two frameworks were never designed to work together. And the integration complexity is even much higher than build a framework from scratch. So too complex. And then we said, forget about it. We're gonna all in on PyTorch. Keep its beautiful, simple frontend and rebuild the backend. So we built Torch script, that's PyTorch one dot O.
6:33 Speaker 00 So that's really like the key focus on simplicity wins over time. The other interesting thing is we thought replacing other frameworks as library with PyTorch is to be simple. It's just swap the library. How hard can that be? But we realized it's, we thought it's just a six month project. It turns out to be a five year project for us. To support entire MATA's AI workload building on top of PyTorch because we have to rebuild the whole entire stack. From scratch, from ground up. Because we have to think about how to load data efficiently, how to distribute inference in PyTorch efficiently.
7:15 Speaker 00 How to, you know, scale training efficiently. And then we end up rebuild the whole entire inference and training stack on top of PyTorch.
7:26 Speaker 00 When we left, it was sustaining more than five trillion inference per day. So that's a kind of massive scale. By two to five years and the fireworks mission is to significant accelerate time to market for the whole entire industry. Compressing it from five years to five weeks or even five days time to market. So that's our mission.
7:47 Speaker 02 Maybe when you look at the open source standards, there's a lot of people that are trying to do it on using VLLM or Tensor or TLLM. How do you think about how fireworks compares to what's in the open source?
8:00 Speaker 00 I really like both projects and because my heart is in open source based on my PyTorch experience. I will say both project are great projects for the, for the community.
8:13 Speaker 00 I think our biggest differentiation is, first of all, fireworks off the shelf is faster than both of the offerings. And second is, we're building a system, we're not just a library. And our system can auto-tune towards our developers or enterprise workload to be much, much faster. And to be much, much higher quality. And that cannot be achieved by just a library. And we're building all this complexity, going back again to our journey of PyTorch.
8:50 Speaker 00 We're providing a very simple API, but hiding a lot of automation, the complexity of automation, complexity of auto-tuning behind the scene. For example, when we deliver our inference with high performance, high performance here means low latency and low cost. We handwritten CUDA kernels. We implemented distributed inference across nodes. And disaggregated inference across GPUs, where we chop models into pieces and scale them differently. We also implemented semantic caching, where given the content, we don't have to recompute.
9:31 Speaker 00 And this, we capture application workload patterns specifically, and they're built into our inference stack.
9:40 Speaker 00 We, yeah, we have many other, many other optimization we are, we have been specifically designed for different use cases. That is not like general purpose or horizontal. So that is being encapsulated. We also have complex optimization for quantization. You can think about, oh, quantization is just one technology, how hard that can be. But you can quantify so many different things. You can quantify KV cache, you can quantify ways, you can quantify communication across GPUs, across nodes. And you know, different performance gain and quality trade-offs. We also automate like quality optimization.
10:24 Speaker 00 There are many things we're doing behind the scene to deliver a very simple experience to the app developers. So they innovate, this concentrate their cognitive bandwidth to innovate on the application side.
10:37 Pat Grady? You, I like your comment earlier about simplicity scales. And as you're talking through everything that you've built to make this such a simple and delightful experience for your customers, it reminds me of the idea of conservation of complexity, you know? Like, the amount of complexity required to deliver any given task can be neither created nor destroyed. It's just a question of who takes the complexity.
11:00 Speaker 00 That's right.
11:00 Pat Grady? And it feels like yours is a business where you have embraced an enormous amount of complexity to make life simple for your customers. And actually, my question is about your customers. So where, where in the AI journey of your customers, where in their AI journey do they say, wait a minute, we need something better?
11:16 Speaker 00 Hmm.
11:17 Pat Grady? And then what brings them to you?
11:18 Speaker 00 Yeah, so we've seen pretty consistent pattern that last year, people all start from open air.
11:27 Speaker 00 Because they are in the heavy experimentation, exploration mode, many startup, they are, they have some creative ideas, application product ideas, and they want to explore product market fit. So they want to start with the most powerful model.
11:43 Pat Grady? Yeah.
11:43 Speaker 00 Where open air provides.
11:46 Speaker 00 And, and then when they, they feel confident they hit product market fit, they want to scale their business. And then the problem comes in because, as I mentioned, most of the general application there be to see consumer, personal, developer facing. It requires very high responsiveness. Low latency is critical part of product viability. Without that, they don't, it's not viable product. People are not patient enough to wait for half a minute for a response. That's not gonna work. So they are seeking, actively seeking low latency. And then another key factor is they want to build a sustainable viable business. They cannot bankrupt quickly. And the weird thing is.
12:29 Pat Grady? Not in this market they can't.
12:30 Speaker 00 The weird thing is if they hit a viable product, that means they can scale quickly. And if they're losing money at a small scale, they're gonna bankrupt quickly, right? So. Bring down the total cost ownership is critical for them. So that's why they come to us.
12:47 Pat Grady? So it sounds like, I remember you had this insight a year or so ago and we spoke about, you know, training tends to scale in proportion to the number of researchers that you have, whereas inference tends to scale in proportion to the number of customers that you have. And in the long term, probably going to be more customers of AI products than AI researchers out there, and therefore inference is the place to be. It sounds like, it sounds like you're kind of, the customer journey sort of begins as people are going from training into inference. What, what sort of applications, what sort of companies are at that point where they're starting to really go into production?
13:23 Speaker 00 There, there's so many ways to answer this question. It's a very interesting question. So first of all, my hypothesis when I start a company is we're gonna take our startups first because they are most tech advanced. They, there will be ton of startup built on top of GNI. Then we'll go to digital native enterprises because they are tech forward. And then we'll go to traditional enterprise because they are like tech conservative. They want to observe and, you know, adopt when the technology and the product ideas are mature. So that's kind of my hypothesis. And it totally blew my mind what's happening right now because we have a lot of inbound of startups.
14:08 Speaker 00 We are working with digital native enterprises. We're also simultaneously working with traditional enterprises, including health insurance company, healthcare companies and banks. And especially for those traditional enterprise, usually I adjust my pitch to be very business oriented because hey, you know, that's kind of my, maybe my bias and kind of want to strike a meaningful conversation with them. But they quickly dive into very low level details, technical details with me. And it's very, very engaging.
14:40 Pat Grady? What are the people doing like at a traditional enterprise? Who are the people that you're engaging with?
14:45 Speaker 00 So we.
14:47 Pat Grady? Like is it innovation, innovation person, AI person? Or is it more business line leader, somebody who owns a production application?
14:54 Speaker 00 Yeah. So I think it's start to shift, we are more engaging start with CTOs.
15:01 Speaker 00 They kind of, I feel like this business is shifting towards.
15:06 Speaker 00 Innovation driven business transformation. And that's why can we we encounter more CTOs than like the CLOs or other CISOs. Um. So that's kind of interesting shift. Um. But yeah, across the board, I I think there are multiple fundamental reasons that why that's happening. That's my hypothesis. One is um, all the leaders realize current Gen I wave is similar to the cloud first shift or similar to the mobile first shift. It's gonna remap the landscape of industry. Startups are growing really fast and the incumbents feel threatened. If they are not innovating fast enough, they will be obsolete. They will it will be irrelevant.
15:49 Speaker 00 But also across the incumbents, they are heavily competing with each other. They're competing how fast they kind of transition their business to be to create more revenue, to be more efficient using Gen I. So that's one phenomenon. The second phenomenon is general AI is different from traditional AI. I would say kind of this is like very different. Traditional AI is given a lot of power to hyperscalers, right? Because traditional AI is you always have to train from scratch. There's no concept of foundation model you build on top of. Um. And that means you have to go off and curate all the data. And the data rich company, usually they are hyperscalers. And it you need to have a lot of resource investment to train your own models and so on.
16:35 Speaker 00 So that is before Gen I. It's less affordable. It's concentrated in hyperscalers. Post Gen I, because of the this concept of foundation models, people build on top of foundation models and you you don't train from scratch. It's it's not meaningful. It's all the same data. It's all the internet data you can crawl. Uh. It's a more or less similar model architecture. It's a waste of resource if you train from scratch. Instead, you fine tune. You tune uh based on your uh small data set, high quality small data set. So it becomes a small data, small model problem. And it makes it so much affordable to everyone uh to assess uh to assess this technology.
17:17 Speaker 00 Um. And that's why everyone is jumping in to embrace it.
17:22 Speaker 02 How many of your customers are using you for fine tuning versus just using a base model? And what do you think goes into building a great fine tuning product?
17:29 Speaker 00 It really depends on the problem they're trying to solve. We actually see the open source model is becoming better and better. The quality difference between open source model and the closed source model are shrinking. And my prediction is going to converge at the same model buckets, same model size.
17:46 Pat Grady? If you go open and closed, will converge?
17:48 Speaker 00 The open and the closed will converge.
17:50 Pat Grady? Do you think there will be a time lag where closed is always six months ahead? Or do you think they'll just be neck and neck?
17:55 Speaker 00 For same model size, especially like seventy, between seven to seventy billion, or even within a hundred billion model size, the quality will converge. That's my prediction. We'll see, we'll see after a couple of years and we will come back to this podcast and see how it goes.
18:14 Speaker 00 So, so the key here is customization, right? Given like, if this trend is true, then the key differentiation is how we customize those model towards individuals' use case, towards individuals workload.
18:30 Speaker 02 And is it easier to customize an open source model than a closed source model? Or how that comes in?
18:35 Speaker 00 So I would say it's easier. It's just the open source model tend to have a much richer community. And there are a lot more people working on building on top of those models. For example, a Lama 3 model, it's a very, very good base model. It is very strong in instruction following. It follow instruction very well. So it's very easy to it, to align the model to solve a specific problem really well. And for example, we have been investing in function calling strategically as a direction. We can talk more about that. It's a own topic by itself. But we find like fine tune, fine tune a function calling model on top of Lama 3 is so much easier
19:17 Speaker 00 compared with fine tuning based on mixture models or previous Lama 2 and, and other models. So that's just kind of, the base model, open source base model is becoming very, very strong in instruction following, in logical reasoning, in many other base capabilities. So it's very easy to morph it to become a high performance model for solving specific business task. That's the power of small models.
19:42 Pat Grady? If we think about just open source software, open source infrastructure software, 20 years ago, open source was thought of as a fast follower sort of thing, you know, Red Hat being a canonical example. And then more recently, open source is not the fast follower, it's actually the innovator. And you think about Mongo or Confluent or some of these other great open source businesses that have been created. Do you think there are areas in the world of models where open source is actually going to lead closed source and is actually going to be ahead of the proprietary models?
20:16 Speaker 00 So I think the dynamics is very interesting right now because the proprietary closed source model provider, they are betting on very few models, right? So OpenAI's LM models are like maybe three models, right? Or you can think about that as one model, because what is a model? Model is the model architecture and data, training data, right? That defines a model. So I'm pretty sure they use all the models, they have more or less similar training data, model architecture is more or less similar, so it's kind of scaled all parameters and so on. It's not just OpenAI, I think Anthropic or Mistral and also all these kind of model
20:54 Speaker 00 builders, they have to concentrate the effort to focus on a specific like model segment. That's kind of the best ROI.
21:07 Pat Grady? Yeah.
21:07 Speaker 00 That's a bit small. But open source push a different dynamics because it enables so many researchers to build on top of it. So that's kind of the small model phenomenon. It's smaller and it's easier to tune, easier to improve quality, easier to focus on specific problem space. So it enable thousands of flower to blossom, thousands of flower blossom. So that's the direction we believe in is to solve an enterprise problem, a thousand flower blossom is much better for enterprise.
21:48 Speaker 00 Because you know, you just have so many problems and I bet at the given, at any problem space, there is a solution for you and we further customize towards your use case and your workload. And what you get is better quality. Much lower latency for real time application, much lower cost for business sustainability and and growth. So so we believe in that direction.
22:17 Speaker 02 Maybe to that point, have you seen your customers so far are able to match the quality that they, you know, got with OpenAI when they move over to the the fireworks stack? And like, how are you enabling the, what I call the small but mighty stack to compete?
22:31 Speaker 00 Yeah, so it really depends on domain. So for some domain, actually, people don't even fine tune. They use an off shelf model as is, and it's already very, very, very good. For example, in the domain of coding code palette, code generation, transcription, translation, OCR, it's just phenomenal. Those models are really, really good. So that's kind of off shelf and ready to go. But for some areas, it requires business logic, because every company is defining what is good differently. And then, of course, like off the shelf model will not work off the shelf, because they don't understand the business logic.
23:13 Speaker 00 For example, classification, different company want to classify, hey, you know, some marketplace want to classify whether it's a furniture, or it's a dish, or it's something else. That's completely depends on their domain. Or summarization, you think summarization is a very general task, but for example, insurance company want to summarize into a very special template.
23:39 Speaker 00 Right, so and there there are specific business tasks.
23:47 Speaker 00 Yeah, on many other things, we we just kind of work with across the board, various different problems. And those requires fine tuning. And I want to call out, fine tuning sounds simple, but it's actually not simple at all. So the end to end requires enterprise or developers to collect data first thing to trace. After they traced into label. After the label, they need to pick and choose which fine tuning algorithm to use. There's supervised fine tuning, there's DPO, there's slew of preference-based fine tuning. As in, they don't label absolute good result. They basically say, I prefer this over that.
24:33 Speaker 00 They need to pick whether they want to use parameter-efficient fine tuning like LoRa or full model fine tuning. And for some task, they need to tune hyperparameters, not just the model weights itself. So among these many technologies, they have to kind of figure out when to use what and so on. It's very, it's very deep. And usually those app developer, they haven't even touched AI yet, and this is a lot for them to pick up. And then, once they've tuned and they test it, it's still improving some dimension. If it's still not good in some other cases, and then they need to capture those failure cases and analyze, should I collect more data and go through this cycle again?
25:18 Speaker 00 Or it's actually a product design, right? It's very interesting. Some failure cases, not really failure cases, it's just they haven't designed what the product should react. For example, people are building an assistant to auto-generate content when people type. And if you're in a table and your cursor is in a cell, and what does auto-generate mean? Do you auto-extend what you're typing in the cell? Do you generate more rows or you do nothing? So it's actually a product design. So that requires a PM to be in the loop to think about the failure cases. So with all this complexity, what we want to do is take away the rudimentary stuff. Take away the complexity of figure out which tuning approach to use.
26:05 Speaker 00 How to automatically label data? How to automatically collect data from production? Want to take away all this and keep a simple API for people to use, but leave the design part to our end user. For example, how the product should respond should completely in their realm to figure out and solve. So that we want to kind of create that separation. And we started working in this direction and hopefully we'll announce our product there soon.
26:32 Pat Grady? I love that you're kind of liberating people to not have to think from the tech out and to actually think from the customer back. And sort of use all the stuff that you've built to deal with the underlying technology and really focus on, to your point, the design patterns and the usability and making sure that they're actually solving an important problem end to end in a compelling way.
26:51 Speaker 02 What is your vision for the Fireworks platform? And like to Pat's point earlier on conservation of complexity, you know, we started this podcast talking about how you're conserving complexity for your customers on the inference stack. You just now talked about how you're conserving complexity for your customers in terms of the fine tuning workflows. Like, what are the other pieces that have to come together? And what is your ultimate vision for what Fireworks, the product is?
27:15 Pat Grady? If everything works, five years from now, what will you have built?
27:19 Speaker 00 So, um, so what we, like, the north star for Fireworks is this simple API access to the totality of knowledge, right? So right now, we're building towards that direction. We already have more than a hundred models. We provide across large language models, image generation models, audio generation models, video generation models, embedding models, and multimodal models as image as the input to extract information. So that's kind of one side of the foundation model coverage. But put all the foundation model together, it still have limited knowledge, but because its training data is limited.
28:04 Speaker 00 Its training data has a starting time, ending time. All the information they can crawl on the internet is still limited because there are a lot of knowledge that's hidden behind APIs, hidden behind the public APIs that you don't have access to, or you just cannot get real time information. There are a ton of private APIs hosted with the enterprises. No way anybody will have access outside of the companies. So then what, how do we get access to totality of the knowledge for the enterprises? Is to, is to have a layer to blend across many different models and public-private APIs.
28:48 Speaker 00 So that's kind of the vision. And the tool to, the vehicle to get there is function calling, is the function calling model. Basically, this model is capable of
29:01 Speaker 00 Understanding here are the APIs you want to access and for what reason it can automatically be the router to most precisely call out to those APIs, whether it's accessing models or accessing non-model APIs in the most accurate way. So think about strategically, that's extremely important to build this simplified user experience because then our customer, they don't need to figure out, they don't need to scratch their head and figure out, oh, I need to fine tune to be able to access those APIs and how to even do that myself is kind of a tall order for me to do. So we want to basically, you can think about that
29:46 Speaker 00 because many people are familiar with a notion called mixture of expert.
29:50 Pat Grady? Yep.
29:52 Speaker 00 So OpenAI is providing mixture of expert and mixture of expert becomes a very popular model architecture. The concept is has a router sit on top of few very big experts and each is specialized in doing its own thing. And our vision is we want to build a mixture of expert that access hundreds of those experts. And each of those experts are much smaller,
30:16 Speaker 00 much agile, but with high quality of solving specific problems.
30:21 Pat Grady? Can I ask you real quick, do those experts live in fireworks, in AWS, in hugging face? Like where do those experts come from? They get put together with fireworks as the overarching framework?
30:35 Speaker 00 Yeah, our ambition is those experts living fireworks. That's where we want to curate.
30:43 Speaker 00 Curate towards, curate the models we serve towards that. And that's why today we already have more than a hundred models. And it will take some time to to kind of build this layer in a very solid way, but we're gonna release our next generation function calling model. It's really, really good. A little preview on that. It's, it has multiple layers of breakthroughs. And we're gonna announce it together with demos and example and people can leverage and build on top of.
31:13 Pat Grady? Very cool. Do you, do you see any viable competition for Nvidia on the horizon?
31:21 Speaker 00 That's a, that's a very interesting question. First of all, I think Nvidia is operating a very lucrative market.
31:33 Speaker 00 Any lucrative market invites competition. This is just the economics here. And also from the whole entire industry point of view, in general, industry doesn't like monopoly. So that's kind of another trend, like pressure coming from the industry. So I think it's just, it's not a question whether there will be competition to Nvidia, it's just a question of when.
32:02 Pat Grady? Do you think it's coming soon?
32:04 Speaker 00 I think it's coming soon. I think it's coming soon. I think, I mean, obviously, we can look at Nvidia's competition in multiple segments on the, like, general purpose competition segments GPU that M D is coming up.
32:22 Speaker 00 That's interesting. And I think also in a specific AI segment where the AI model space is stabilized, there's no more like innovation, like the problem is well-defined and this is the model, then customizing will have its own role.
32:39 Speaker 00 So, so I think kind of I will look at the market that way and I, I do think there will be competition coming soon.
32:46 Pat Grady? Can I ask you about that, by the way, because you guys are in this part of the market where
32:52 Pat Grady? you are model agnostic to some degree and it's really about the optimization of those models when it comes to put them into production. Do you, do you think that the returns to scale on the frontier, the, the models that are out on the bleeding edge, do you think the, the returns to scale are starting to slow down? Do you think that we're going to go into a phase where capabilities have started to mature or asymptote? And the race is more about the optimization and tuning and application of those capabilities?
33:22 Speaker 00 I think both will happen at the same time. One is. It will start to stabilize and plot two in the model applicability point of view. And will heavily customized, our strategy is heavily customized towards the use cases and workloads. Yeah. So that's one direction. And the second is I want to caution that, like, because at Meta, I also think for certain period of time that that is a model for ranking recommendation, right? So and and we should heavily like index on on that assumption. But but then after a few years, it's not the case.
34:06 Speaker 00 There's significant amount of model innovation in a seemingly stabilized modeling space. And and that pushed its curve for Meta. I think same phenomenon will happen in the Genai space, that a new model architecture will happen. And we're kind of overdue.
34:26 Speaker 02 So you mentioned, we've talked about competition from other other vendors, other direct competitors. What about OpenAI? Does does OpenAI keep you up at night? Like, they drop prices on their APIs all the time. They're, you know, they're they're making their models. They're also trying to win the better, faster, cheaper race. Like, how do you think about them? And how do you think about, you know, ultimately what you're going to build that's different from where they are going?
34:50 Speaker 00 Right, so again, I feel like for the, they are actually going smaller, right? And cheaper. I think for the same model size, for the same model bucket, whether it's closed source or open source, the quality is going to converge. That that's, again, that's my prediction. Yeah. And and the real meaningful thing is to to push the boundary here is heavy customization or automated customization tailored towards individual use case and individual workload. I'm not sure if OpenAI is has the appetite to do it because their mission is AGI. If they hold their mission. Which is a great mission, actually.
35:33 Speaker 00 But it's kind of solving a different problem than solving an enterprise problem, which has, which basically means there are a lot of problems, a lot of specific problems that is really good for the small models to customize towards. And that's where we want to focus our energy and build on top of open source models, assuming the trend that they are going to converge in quality.
35:56 Speaker 02 Love that. Our prior rule of last time you were here made the point that, you know, in prior technology waves, internet, mobile, it was the people that did all the hard work driving down the marginal cost of running this stuff that actually enabled all the application development on top and all the end use cases that we get to enjoy every day. And I love that you are taking that exact approach with AI where it's still so cost prohibitive for most people to run in production. And by just dramatically bringing down that cost curve, you're actually in the whole industry blossom. So it's really wonderful. Should we close out with some rapid fire questions?
36:31 Pat Grady? Yeah, let's do it.
36:32 Speaker 02 Okay, can you go first?
36:33 Pat Grady? No, go for it.
36:34 Speaker 02 Okay, let's see. Favorite AI app?
36:37 Speaker 00 We do a lot of video conferencing and the nose taker for video conferencing is a game changer for us, whatever it is. There's so many different varieties, but I just love that.
36:48 Pat Grady? Which one do you use?
36:50 Speaker 00 I think we use Fathom. Yeah, our sales team use that. It's really good for training and also summarization, so it can shorten our time.
36:59 Speaker 02 Nice. What will be the best performing models in 2024?
37:07 Speaker 00 My prediction is there will be many, given the rate that every week, every week, there's a new model coming up. And SIS Arena, they keep competing with each other. So this is all good news for the whole entire industry. It's really hard to predict which one. But the one prediction I'm pretty confident is the model quality is will keep improving and keep increasing.
37:36 Pat Grady? In the world of AI, who do you admire most?
37:40 Speaker 00 I I would say Meta. It's not one person. But that Meta's commitment to open source.
37:51 Speaker 00 I I think Meta is the most brilliant. In the journey of GNI. By continuous open source, a series of Llama models. And continue to push the boundary, continue to kind of shrinking the quality differences. Uh so, okay, so what what Meta is doing is is basically decentralized power.
38:12 Pat Grady? Yeah.
38:12 Speaker 00 From the hyperscalers to everybody who has a dream. And to innovate on foundation models, GNI models. Uh I I I think that's that's really brilliant. But that, okay, well, agents perform or disappoint this year? I'm very bullish on agents. It's um, I I think it's gonna it's gonna blossom.
38:40 Pat Grady? That's all we got.
38:41 Speaker 00 All right.
38:41 Pat Grady? Thank you.
38:43 Speaker 00 It's really fun to have this conversation. Thanks for having me. Thank you for joining us.
Previous GitHub CEO Thomas Dohmke on Building Copilot, and the the Future of Software Development Next Phaidra’s Jim Gao on Building the Fourth Industrial Revolution with Reinforcement Learning Page updated October 12, 2026 · transcript and summary are machine-generated