text stringlengths 0 2.68k |
|---|
**Adam Stacoviak:** That's hilarious. |
**Anurag Goel:** And it was quite an experience. But I think at that point -- Render was actually just four people when we won in 2019. And we launched the thing... And when you launch it, once you have customers, things just feel very different. Once you have people who want more things from you, then you have to get ... |
But that's the other thing, how do you grow this thing? And I think we were lucky in that we kept growing, despite not having any marketing or sales. So especially in the early years, it was an afterthought, and building the product and allowing people to scale on the platform and giving them what they want was front a... |
I think now we've raised all this money and we have 80-ish people in the company, and we're growing fast. We're hiring fast. And my goal now as CEO has switched to repeating myself over and over again to make sure that we're all aligned around the same kind of thinking around what we're doing, why we're doing it, who w... |
**Adam Stacoviak:** Yeah. |
**Anurag Goel:** And that is a lot of what I spend my time on - hiring, and making sure that we're all moving in the right direction, that everyone knows why we're doing what we're doing... Because it turns out -- you know, all the people who joined early on, they knew what we were doing. They didn't need to be told. T... |
**Adam Stacoviak:** Wow. Is that -- so it sounds like you must have gone through a period where... I imagine you don't write code too frequently today. You don't write a lot of code, I imagine. |
**Anurag Goel:** I wish I did. I wish I did. |
**Adam Stacoviak:** If that's true, then why did you do what you've done? |
**Anurag Goel:** No, no, no. I wish I did in the sort of -- |
**Adam Stacoviak:** \[42:11\] I understand what you mean. I'm being facetious, to some degree, and somewhat rhetorical in that response. It's just a joke, but... |
**Anurag Goel:** No, it's a good question. I wish I wrote -- so here's the thing. I wish I wrote more code on the weekends. Just on the weekends. I don't wish I wrote code day to day. But my weekends are spent doing other things. |
**Adam Stacoviak:** What do you do on the weekends? |
**Anurag Goel:** I'm spending time with my wife, man. |
**Adam Stacoviak:** Okay. I get that. I mean, you never know, though. You never know. You could be doing something like a side project just to keep you warm, just to keep your skills refreshed and whatnot. |
**Anurag Goel:** Yeah, so I need to get into that. For some reason, that hasn't -- I think I'm just in a state of trying to recharge on the weekend. I actually end up playing video games on the weekend, real-time strategy... |
**Adam Stacoviak:** Oh, man. Do you build machines? Do you build your own gaming machines? |
**Anurag Goel:** I did build my own machine, and I still use it. |
**Adam Stacoviak:** Okay. How long ago? Recently? |
**Anurag Goel:** No, that was 2016, when I built this startup... Well, if you can call it that. It was a one-person startup that I ended up selling to another company. But I built the first GPU-backed Jupyter Notebook online that you could start with a single click. And it had all these data scientists who were really ... |
But then Google came in, and they were like "Hey, we'll give it to everyone for free." And to be completely honest, I didn't care as much about data scientists as I cared about developers, application devs... And so then I started working on Render, and I was like "I don't want to do this thing anymore." So then I sold... |
**Adam Stacoviak:** So how does that relate to you building your own gaming PC? |
**Anurag Goel:** Oh, so I needed to run deep learning on my own GPUs. And so it wasn't intended to be a gaming machine. It just became a gaming machine later. |
**Adam Stacoviak:** Because it had no more use. |
**Anurag Goel:** Because - yeah, it had no more use. |
**Adam Stacoviak:** You're like "I've got this GPU, this massive GPU..." |
**Anurag Goel:** I had two of them. I built a two-GPU machine. |
**Adam Stacoviak:** Yeah, it's kind of crazy how that swings back again, because -- and we could probably touch on this to some degree when we get to your focus now with Render on AI, and what that means in your announcements, because I have some questions around that... But there's this new resurgence to -- especially... |
Finding a GPU, whether it's an older gen like a 3090, which is still relatively useful, 4090, which is one of the most earliest models, and then now 5090s - those are all at the top tier of NVIDIA's RTX-level GPUs. Good luck finding those things, right? Good luck finding them, and if you do, they cost just as much as t... |
\[45:57\] It's kind of crazy that it swings back again. You were playing with that, and maybe instead of Render -- I guess maybe now you're best positioned to maybe pick that back up again, because I imagine part of this innovation you're talking about in your announcement post, which I'm sort of paraphrasing... Let me... |
**Anurag Goel:** So it's actually not that. |
**Adam Stacoviak:** Okay. My assumption is incorrect. |
**Anurag Goel:** Well, it might become that, we don't know... But we don't have enough evidence to say right now that that is what we should do. But I think it goes back to what I was saying earlier about people changing how they're building applications... And we realized early on in 2022 that our customers were not t... |
We did a bunch of customer research, because in 2020 we were like "Well, should we go buy a bunch of GPUs?" And you and I both know of companies that are peers that went and did that, and then realized a bit too late that they shouldn't have done it. But we knew early on that this was not the path for us, and I'm not c... |
So when I talk about what we're doing for how people are building apps in the AI era, I'm thinking about the ways in which we can make it easy for people to run a lot of distributed compute, or compute that requires a lot of retries, or compute that is based on a lot of different conditions. |
**Adam Stacoviak:** Temporal then. You're trying to build what Temporal's built. |
**Anurag Goel:** I can neither confirm nor deny that statement. |
**Adam Stacoviak:** Well, that's what you just said. I can confirm that by what you've just said. |
**Anurag Goel:** Well, I think there is an opportunity for us to build something that makes that kind of problem a lot easier to solve for application developers. |
**Adam Stacoviak:** Yeah, it's necessary. |
**Anurag Goel:** Exactly. |
**Adam Stacoviak:** Yeah. Well, if you don't, your application falls over. You've got something waiting, so your application is blocked. You need to have that resilience. |
**Anurag Goel:** Exactly. Exactly. |
**Adam Stacoviak:** And the stack doesn't necessarily -- you have to build it yourself. Or - not an ad. They're a sponsor of ours. I'm aware of how their platform works... But they may actually sponsor this episode. I don't even know, because we don't know which episode is being sponsored when we produce it, but... |
**Anurag Goel:** And that's totally fine, because I think Temporal is a great product. But again, it requires a lot of wrangling, and it requires a lot of like figuring out. And Render is really good at making complex things easy, and giving people an interface that they like to use, and that they love talking about...... |
\[50:16\] Again, model training is not where we would focus... But you can imagine -- you still can't run serverless inference in a way that is... I mean, there are places where you can go do it, but the other thing we've realized for Render is people generally want everything in the same place, to the extent that they... |
**Adam Stacoviak:** It's kind of funny, your next innovation is like the oldest tech, basically... |
**Anurag Goel:** Yeah, the thing that AWS built first. |
**Adam Stacoviak:** Yeah. I mean, it's what they're most known for. They've wanted so much that S3-compatible is the way. I'm sure that when you build your object storage, you're going to make it S3-compatible, because you have to. |
**Anurag Goel:** Absolutely. Yeah. But again, it's not -- so the way we think about product is not to build technology for the sake of it, it's to build technology to solve the very real problems that people have. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.