SPEAKER_00: You don't want any one party controlling a platform technology, you want it to be open source, you want closed source solutions, private company solutions, open source solutions, you want a range of opportunities. And remember, the last time we had some founder say, trust me, I'll get us some regulation. That was SPF. And he was, he was going to be the one who got us federal regulation for crypto. And he's on trial. That's a very moment for a bunch of Meshuggah.

SPEAKER_06: This Week in Startups is brought to you by LinkedIn Marketing. To redeem a $100 LinkedIn ad credit and launch your first campaign, go to linkedin.com slash thisweekinstartups. Vanta. Compliance and security shouldn't be a deal breaker for startups to win new business. Vanta makes it easy for companies to get a SOC 2 report fast. Twist listeners can get $1,000 off for a limited time at vanta.com slash twist. And CLA. Innovation takes balance. CLA's CPAs, consultants, and wealth advisors can help you get from startup to where you want to end up.

David Friedberg: All right, everybody. Welcome back to this Week in Startups. It is Monday. It is Madra Monday. Yes, that's right. Every Monday, Money Madra joins us, Sonny Madra of Definitive AI. And we do our AI demos. And this week, Sonny, you told me you were having some deep weekend discussions. We know Silicon Valley and the tech industry on the weekends. That's when everybody goes deep. They don't have meetings, but the back channels start to light up.

SPEAKER_12: And the back channel was lighting up this weekend, talking about what topic.

SPEAKER_15: Well, open source, right? There was a, you know, a couple of big threads that kicked up this weekend and, you know, um, the tech start flying. And then I think, you know, we have a special opportunity to have a guest with us today.

SPEAKER_18: Oh, okay. Who do we ask? Yeah. We've got the legendary, the goat, Bill Gurley.

David Friedberg: All right. Bill Gurley, uh, I think maybe third time. No, only second time on this week in startups.

SPEAKER_11: You were last on 2017, uh, episode 722, uh, for folks, uh, yeah. Uh, so welcome back. Um, Bill Gurley, of course, uh, from benchmark and, um, many great companies. Uh, you're particularly passionate about open source. Why, uh, in relation to AI bill?

SPEAKER_25: Well, I mean, obviously there's a lot of answers or answers to that question. Um, but most recently when I did the, uh, regulatory capture speech at your conference, um, I was mentioning at the very end that, that oddly, some of the, the, what you might call the early incumbents in, in AI software, where we're running out and promoting the idea that open source should somehow be curtailed or kneecapped. Um, in particular, it's quite notable that all of the loud voices are either executives at these companies and, or large investors at these companies. And, and, and because of the way these companies have raised money, the, the largeness of those investments are in the tens and hundreds of millions of dollars. So quite a bit at stake. And, um, um, there now seems to be a growing populace. And, and this is what led to the conversation, Sonia, we're having that are, that are worried that these companies are basically, you know, at the very start of this AI movement, trying to cut off their biggest competition whatsoever. And the thing that would probably unleash the most innovation and the most prosperity, which would be if open source models, um, were, were prevalent.

SPEAKER_31: And, and this is more Sonny's world than me, so I'll let him hum it. But I do believe that the, the stats about the performance of these open source models is actually quite compelling.

SPEAKER_25: And what I'm hearing internally from our portfolio is this specifically Lama too, but Sonny sent me another one today, um, are really starting to gain market share amongst the startups that are using these tools. So it would be really unfortunate if they found, and when I, when I spoke at your conference, I mentioned, you know, this guy, George Stigler, who had won a Nobel prize in academic mission, who had talked about how companies use regulatory capture. And he said the two things they try and do is store competition and protect pricing. Clearly, if they were successful in getting governments to block open source AI, they would achieve both of those goals.

SPEAKER_00: Yeah. Yeah. And so that great talk, by the way, if you haven't seen it, just do a Google search for all in summit, Bill Gurley, and, um, the name of the talk, 2851 miles about regulatory capture. So Sonny, when we look at the space, the open source space, let's get to, you know, maybe the, the tail of the tape, some metrics here on how these open source projects are faring against, uh, the large language models that are available by API call, right?

SPEAKER_39: And I guess, uh, paradoxically open AI, which is now the most closed of all the AI companies, but has open in the name and started under the mandate that this technology was too powerful for it to be closed. It needed to be open to everybody has gone exactly close and has gone exactly regulatory capture with, uh, Sam Altman begging, literally, uh, begging, uh, for the government to get involved and to regulate.

SPEAKER_42: So maybe we could talk a little bit about, um, you know, the, what, what are the leading, uh, open, uh, projects? Yeah.

SPEAKER_43: So let's just kind of level set, like, you know, who the players are, because, you know, to Bill's point, there's many different people that are showing up. Uh, and, and also this is kind of a good lay of the land because you see here, some startups, some real incumbents and, you know, some, some new folks that really kind of risen to the top. And so just level setting both anthropic and open AI, you know, uh, I'd say kind of leading models in the space, definitely closed. Uh, Microsoft is kind of on this unique straddle there. They're open sourcing lots of unique pieces of content, uh, the additional frameworks that you need around, um, uh, like LLMs, but they're gone all in with open AI. Google started this whole thing, obviously the, you know, the reason others can get there is the papers are open source and people have written about them, but we haven't seen an open model from Google in, in a while. Although, you know, they do support open models in their, uh, vertex platform and then meta and then Databricks. Uh, and there's another one we'll bring up mistral as well today. Uh, the, these, these folks are fully open, uh, commercial use case is different. The commercial use, uh, differs based on their licenses, but in general, they're sharing everything and we're seeing really fast innovation. So that's the level set.

SPEAKER_31: Um, in terms of just two quick things to add to what Sonny just said.

SPEAKER_25: It's, it's remarkably ironic that, that open AI when Elon backed it was open source and moved away from it.

SPEAKER_31: Um, and then second, as, as Sonny mentioned, I've been reading a ton this weekend to try and get a lot deeper in this world.

SPEAKER_25: And everyone points to this Google paper attention is all you need as the thing that allowed these LLMs to become successful and to really progress. So you did, you do have an academic paper with all Google people on it that actually led to the technology that's being used by anthropic and open AI. Um, so I, I think that's ironic also.

SPEAKER_50: Anyway, sorry to interrupt, Sonny. No, no, no.

SPEAKER_39: And, and going deeper into the playing field, what's your analysis of what's happening here?

SPEAKER_00: What I hear often, uh, people say is when you're in the lead, you're closed. And when you're behind, you go open. So iOS was in the lead and they, and they kept it very closed. Google search their algorithm. They're in the lead. They have a, a dominant position. They'll never open up the search, uh, algorithm, but then you have somebody like maybe meta. They feel like they're pretty far behind on, uh, language models. So they open source it. And the report was they internally leaked it accidentally on purpose.

SPEAKER_53: All right, listen, when you're selling to business to business buyers, you really want to get your pitch in front of decision makers. Why? Because upper level execs are usually the ones making purchasing decisions. Duh. The problem is high level folks can be really hard to find and target on most social media platforms, but on LinkedIn. Oh my God. They know all of the CTOs, all of the CFOs, all of the VPs of finance, engineering, HR, recruiting. All those titles are sitting there waiting for you. And now let's just talk about the funnel. LinkedIn is about to hit a billion members. Did you know that? 950 million members at this point in time. There are 180 million of those 950 who are senior level execs. There are 10 million C level executives in that 180 million senior level execs, which are part of the 950 million members. I am a C level executive. I am on LinkedIn all day long because LinkedIn equals business. Business equals LinkedIn and LinkedIn ads are built specifically for B2B marketers. LinkedIn generates 2 to 5 times higher return on ad spend than other social media platforms. LinkedIn equals business. Business equals LinkedIn. When people are on LinkedIn, they're ready to do business. It's that simple. So make business to business marketing everything it can be. And get $100 credit on your next campaign from me, your boy, J. Kyle. I'm sending you the hundy linkedin.com slash this week in startups to claim your credit. That's linkedin.com slash this week in startups. Terms and conditions apply because they've given you the hundy.

SPEAKER_42: So maybe some, some historical information here on when do companies choose open versus closed.

SPEAKER_25: And to your point, Google was behind in cloud services and led a movement to open source Kubernetes. So there's a company that has played both sides of, of the aisle, depending on where they are.

SPEAKER_39: When you, when you look at this field, open AI and anthropic. These are startups. Uh, have you ever seen startups? Be, uh, asked for regulation this early, uh, and this often and be so opposed to open source. Is this a new trend? And what are you attributed to?

SPEAKER_62: I mean, Bill, like there's more your area. You answer that one. And then I can chime. Yeah.

SPEAKER_30: Well, it's funny because I, I have certainly never seen it.

SPEAKER_31: I've never seen it, Jason. I've never seen this early a, and maybe it just speaks to the wild success of open source.

SPEAKER_25: I mean, the number of venture backed open source companies today versus 30 years ago is just amazing. Um, it is a very disruptive way to get your technology out there quick and fast. And, um, and it's, you know, one of the other things that Stigler talked about, a phrase I mentioned several times, he said, when you have this type of blocking regulation, you end up with a net loss to society.

SPEAKER_31: And I, I'm one that firmly believes that open source is amazing for society because when you have technology locked up with patents, it's harder for ideas to spread. It's harder for ideas to spread across borders and to other countries among all these different smart people that can get out and innovate. And so I have never seen something this early. Obviously there's a ton at stake. There's a, a ton. Like I mentioned, these companies have raised money at an unprecedented level. So yeah, you could call them early state startups, but you could also, I mean, who they've raised billions of dollars each, you know, and only maybe in the ride sharing market did you have that happen so quickly. And so there's a lot at stake. There's clearly a lot at stake, but I've never seen startups proactively pitch governments and not just ours, but several governments around the world. I've never seen that. I also find it really suspect that there aren't like technology lead, like academicians out there leading this charge. The people that are leading the charge, calling for the regulation and calling, and some of them raising this question of whether open source should be allowed, are the incumbents. Are the one either the incumbents are their backers. Um, and some of them, you know, Mustafa, uh, uh, for, uh, from inflection, our podcast said, I know it looks odd. Me being the one that's, that's asking. Yeah.

SPEAKER_69: And, uh, yeah, it does. It is, uh, it is odd.

SPEAKER_70: And it would, it reeks of protectionism, uh, pulling up the ladder behind you.

SPEAKER_00: Um, and with Microsoft and Google, they seem to be, uh, maybe trying to dance along the line here, Bill. They want to have cloud computing, uh, sir. They both have major cloud computing services, Azure, Google cloud. They want to offer these things, but they also have, you know, a proprietary use for this. Obviously Google search and bard and the chat interface are going to overlap. Microsoft trying to get Bing to break out using AI in the office suite. So they see that as a competitive advantage.

SPEAKER_42: What's your, how would you handicap Google and Microsoft's behavior? Yeah.

SPEAKER_31: Well, I mean, I think to a certain extent, it was interesting on the chart that I don't know who produced the chart that Sonny put up, but it had Microsoft neutral. I, I could find myself believing that, I mean, they had to do a rather convoluted deal to get access.

SPEAKER_25: Um, to the technology that they're, you know, using with open AI. And I wouldn't be shocked if they're comfortable with a hedge on that, um, primarily because they already control these creative products, um, that they now believe will be enhanced with AI. And, and I don't know that whether it's open source AI or someone else's AI that it really impacts them because it, it's the lock in they have on the product. So I wouldn't be surprised by that. And like I mentioned, Google's played this both ways.

SPEAKER_31: So, you know, they, they pseudo open sourced Android because within open source, there's different dimensions on how open it is and whether you've really committed to a third party.

SPEAKER_25: Like the Linux foundation that, that runs the regulatory, uh, aspect of it or not regulatory, cause I don't want to confuse it with the government there that, that, that runs, you know, how it's had the, the, the, many, the most open projects have a third party. That, that keeps open source foundation. Yeah. Yeah. There's other, there's other than the open source foundation, but that's the largest one, um, that, that manage the process. And Android is not like that. So anyway, I had to do that quick aside, but Google's done. Whereas Kubernetes wide open, uh, Linux foundation manages it. And so they, they've been all over the map. And, uh, as you mentioned, they once posted a paper, we love open source, but it's not right for search, you know?

SPEAKER_82: Yeah. Well, once you get that lock in, that's when you don't want to open source it because people can then, uh, build competitive products.

SPEAKER_00: And as we've seen in search, that space has not seen any changes in 20 years. Like that has been a locked, you know, box where nobody's innovated for 20 years ago. A number of people have tried. I tried myself. It's very hard to get any kind of a foothold in search.

SPEAKER_25: But I don't think, you know, those two aren't at the forefront of this, as you mentioned, it seems to be a battle between these extremely well-funded startups. And, and, and really more of a community. I I'd say the people that are on the other side of this, based on what I saw going around this weekend, it's more of a community. I mean, it involves, you know, like Jim Zimlin, uh, who, who, who runs the Linux Foundation, who's been out talking. At your conference, I spoke to Steven Wolfram, who, who, who told me he thought it was ridiculous that someone would try and ban open source here on a safety reason.

SPEAKER_31: And so that's what I say when, if, if I might be more open to listening, if I thought it were, you know, some broad group of technologists and big thinkers that were making this argument, but all of the arguments are coming from the people with the most to lose.

SPEAKER_92: Yeah. And that seems crazy. Um, Sonny, you want to just give us an idea of how, well, yeah, I mean, you know, we're going to lose Bill in a couple of minutes here, but like one thing I'd love to get his thoughts on.

SPEAKER_43: Cause it, I think it really reinforces the point here is that, so, you know, last week we saw a big funding announced around a mistrol. It's like a European based group that, um, you know, raised a hundred plus million dollars to Bill's point, a lot of money. And, you know, they released their model open source open for commercial use as well. And you can see a couple of key points here. One, you can see that there's 7 billion parameter model uses again, half the amount of memory. And down here, even just against other open source models, right there, there are 7 billion parameter model is outperforming, you know, lama to 13 billion. Again, this is just against open source, but like the rate of innovation is moving so quickly here that if, if some of that regulation were to come into play in this, these folks couldn't put this out there and they had to do it. You know, through some kind of regulator, which has been, you know, like the FDA or some of that, those are the ideas that we've heard put out there. I think we just wouldn't see that. And I'll just add one more thing. Like last week, Jake, how, you know, we even demo GPTV and, uh, you know, a week later we get lava and lava as an open source implementation of like a vision model in the same example that we did there. Um, you know, we have it and I think, you know, to Bill's point, I can't get my head around, um, you know, what it is around open source. That's bothering people. We've seen open source in operating systems. We've seen it in databases, right? We've seen it in mobile phone operating systems. You know, you get things get patched quicker when the code is open, you can find vulnerabilities.

SPEAKER_15: And so the arguments that around safety, no one is providing the sort of the background as to, you know, what is it that is not safe here?

SPEAKER_11: It's fairly obvious what's happening here, Bill.

SPEAKER_00: Uh, the people who have the lead are using job destruction and the fear of AI from science fiction and this, you know, uh, could get out of control. The, the demon could be unleashed. They're using that in order to maintain their lead because they know full well that large, uh, numbers of startups or midsize companies embracing an open source project would lead to the demise and would absolutely evaporate the lead of open AI.

SPEAKER_97: And this really is about open AI and Sam Altman. Let's call it what it is. Sam is the one who's leading the charge.

SPEAKER_25: Although, although Mustafa and Reid Hoffman have been perhaps even more vocal or at least openly vocal on podcasts and whatnot. So it's, it's not just that, but once again, hundreds of millions, billions at stake. Yeah. I mean, I agree with Sonny. I mean, I, it's ironic, but Linux is the most stable, most secure operating system that's ever existed. And I think, you know, I, I go back to the, um, some of the original thesis of why open source would work and, and more eyes, the better, the more, the more transparency, the better.

SPEAKER_31: And the notion that the people that it's just so ridiculous for someone to say this stuff, super scary. Like, like you should be really afraid, but let me do it. You know? Like I'm, I'm, I'm, you should trust me, but it's super scary, but it can, it can do really good things, but it's scary. It's scary. Let me take care of, you know, help me be the only one that gets to take care of it. And, uh, yeah, that's sad. I, I hope there, there's quite a few people stepping up. I hope this doesn't happen. One last thing I would mention before I have to go. I think the cat's out of the bag. So you're not going to, uh, stop there from being open source, uh, in parts of the globe. So if you shut it down in a particular region, that region is going to fail to innovate relative to the other regions that are out there.

SPEAKER_25: There's a, there's a really cool piece of open source technology called risk five, which you guys have talked about a couple of times in the semiconductor space. And I believe they intentionally moved the, the, um, governing body outside the U S because they were afraid of the restrictions. The U S was putting on semiconductor technology. And I think it was smart that they did that. Um, and, and risk five is going to be wildly successful, you know, regardless of what happens to us, uh, regulation.

SPEAKER_31: And so I think governments need to be particularly careful that if they take a first step move here, they're going to put their own.

SPEAKER_25: You know, society in a worse place in their own entrepreneurism and a worse place than others around the globe.

SPEAKER_105: I mean, and we've seen this, uh, play itself out with social networks and the impact they have on, uh, you know, society writ large.

SPEAKER_00: The fact that so much power was consolidated and oh, trust us, we'll, we'll be the, we'll protect you. It doesn't work. Uh, the person who's making a profit and has a profit mode of the incentive is too great, uh, to act in the public's interest.

SPEAKER_42: Whereas a group of people working on the project together, they keep each other in check. Yeah. That's like, it's a governance thing, right? Bill.

SPEAKER_36: Yeah. Undoubtedly. And this is, this is an important message to spread immediately. Yeah. So I appreciate, I appreciate you guys having me on. Yeah.

SPEAKER_00: All right, Bill. We appreciate you checking in here, uh, and a great continuation of your talk from All In. Uh, everybody that's Bill Gurley. Follow him.

SPEAKER_70: Thank you. On Twitter where he's super active.

SPEAKER_39: If you're a SaaS or services company that stores customer data in the cloud, then you need to be, uh, SOC 2 compliant. You knew that from a third party and you need that third party to close big deals. And if you want to get compliant easier and faster, you need to use Vanta, V-A-N-T-A. Vanta makes it so easy for you to get and renew your SOC 2. On average, Vanta customers are SOC 2 compliant in just two to four weeks. Prepare that to three to five months without Vanta. And Vanta can save you hundreds of hours of manual work and up to 85% of compliance costs. This is a total no brainer. And Vanta does more than just SOC 2 compliance.

SPEAKER_53: They also automate up to 90% compliance for GDPR, HIPAA, and more. You can't afford to lose out on major customers. We all know that. Listen, it's a hard year. Last year was hard. You can't lose those major customers because you don't have your compliance dialed in. Just work with Vanta. Get your compliance automated and tight and tight is right. Lock down those big deals. Here's the best part. Vanta is going to give you $1,000 off. That's 10 hundies. Get $1,000 off at Vanta.com slash twist. That's Vanta.com slash twist for $1,000 off your SOC 2.

SPEAKER_00: Okay, so let's go deeper into the actual models here. Yep. Because the demos are great. You glossed right over the demo of what you called lava. Yeah.

SPEAKER_116: Which I'm going to pull that back up. I think we should stop.

SPEAKER_00: We should start there because this is I think where the rubber meets the road. If you're wondering why a friend of the pod Sam Altman might not want competition. Well, if you look at open source last week, we had the multi modal chat GPT for we were playing with it. I have it on my phone now.

SPEAKER_118: It's extraordinary. It's amazing. You take a picture, you upload a picture and you ask it to do things with the picture.

SPEAKER_84: One of the things we have to do is to make a hamburger recipe based on a hamburger. That's sunny.

SPEAKER_97: Um, was interested in, but there is an open source project called lava. Yeah.

SPEAKER_122: Large language and vision assistant. Got it.

David Friedberg: Got it. So this is built on top of a llama. I take it or it's its own project.

SPEAKER_43: It's its own project. Got it. And, um, it it's, you know, built to mimic the spirits of multimodal GPT for. Hmm. And, um, you know, just kind of touching a little bit on, you know, the Microsoft being yellow in that chart that we talked about earlier, you can see here, it's a, uh, put together by researchers at, uh, was university, Wisconsin, Madison, Microsoft and Columbia. And so the kind of really interesting group. And I think it, you know, does justify that, you know, Microsoft is still supporting, you know, open source, which, which is important here. And, um, you know, it's, it's a great paper. Uh, we won't spend too much time, um, uh, looking at it, but I, I suggest people to go look at the, the GitHub URL, which is here. But, um, you know, this doesn't have the most type in L L a V a, um, large language and

SPEAKER_70: visual, uh, and vision assistant.

SPEAKER_39: And so this interface looks very similar to the chat GPT for multimodal interface. You said, give me instructions on how to prepare this, this being a brioche bun.

SPEAKER_00: Uh, weak egg yolk hamburger. That looks absolutely delicious.

SPEAKER_134: And this is the same image we used last week.

SPEAKER_136: Correct. And so how did it do versus chat GPT for?

SPEAKER_43: Yeah. Yeah. So I'll say, look, chat GPT, uh, you know, V did a better job in terms of, uh, describing what it was and then the instructions. But, you know, for me, what I'll say here is the fact that this is available and it's open and we can build from, and it's one week later, that's that, that's that race that's starting to collapse now. Right. We're not like 100% equal a week later, but being, uh, you know, suit my grade on this would probably be like, you know, in comparison to that, it's like, uh, you know, maybe a B because it doesn't, here's, here's the results right here. Yeah.

SPEAKER_141: And then a, you would give this, this student a B, which means, Hey, this student applies themselves.

SPEAKER_00: And if you were, let's say, uh, a startup, would you tie your wagon to chat GPT for a closed system with deep ties into Microsoft? Or would you fork this and start building with your own hooks into it and wrapping? What would you advise a startup you've invested in?

SPEAKER_70: What would you advise them to do? Sunny? Or would you have them split the baby and, and do it?

SPEAKER_43: It would be like on a use case by use case. Right? Like, I think it really depends on like what, what value you're trying to build. Like if you're really want to, you know, have the whole stack in your control, because that's how you can provide value. I think I would do this. Like I would use, you know, lava, but if it's just a small feature within my product, then I would use chat GPT. Cause I'm going to get there quicker and I don't have to worry about the infrastructure and scaling costs. And so the reason I didn't do this one live, it probably takes like a two or three minutes to run. Sure. Because they don't. And this goes back to, you know, the funding. They don't have the billions of dollars behind it to have, you know, huge farms for inference. And so this thing, it runs a bit slower. Um, and you, you can try it out live yourself as well. Jake, I'll drop the link for us here. Um, and you know, I think they don't have the 10,000 GPU cluster. Yeah. I has, um, exactly. And, but you know, there's going to be cases where for your business, it's important that you build this and you, you build some IP around it. And so I'll go back to a point that I made a couple of maybe episodes ago, which was. The API for GPS, right? The API for GPS on a cell phone been around for a long time. Apple, you know, made it slightly better, but. Companies were built when they built a lot of infrastructure around that. Right. And saying, Hey, like, you know, I'm going to build my app there and I'm not, this just one part of what I'm trying to do. So I think you have to kind of look at in some places, you know, how important is this to the core of what you're trying to build? But I, I do want to reiterate, like, look how fast, how quickly this is happening on the backs of it. And I do think it's always great to have bill here earlier. This is why there's so much noise around slowing this down, because if you're this company, you know, being valued at, you know, whatever, 20 billion, 50 billion, 100 billion, these ridiculous

SPEAKER_93: valuations right now to have an open source model right on your heels. It's gotta be really challenging. I'll put that back to you as an investor.

SPEAKER_39: It wants to buy shares in open AI at 90 billion for common shares. I take it with a cap upside.

David Friedberg: You would probably want to monitor these open source, uh, solutions and say, well, if they're

SPEAKER_00: getting 90 times revenue or a hundred times revenue, whatever it is, uh, for the current valuation of open AI, is that a good bet or are these open source projects going to create downward pressure and the downward pressure would provide is the API calls. And what chat GPT for opening? I could charge for access to their language model is going to go down. And because you could just fire up your own open source solution on your own servers. And this is where Microsoft, um, Google, and of course that we didn't talk about Amazon's position here, which is, I think they also invested in anthropic. Anthropic. Yeah. Amazon's been very clear that they would like to be a neutral third party in all of this. And so Amazon will make money no matter what they're going to have every language model on AWS. And their interest is in continuing to lower the costs continually and just make up for it in scale. Uh, but this could be, you know, a road to nowhere.

SPEAKER_118: This could make open AI be like Seagate, you know, it's like a hard drive provider. Like there's just not a lot of value there.

SPEAKER_43: You know, in the cloud today, you know, for the most part, obviously, you know, some people do do this specifically, but when you're asking for a server, you don't really care whether it's, uh, you know, Intel or AMD or, you know, some kind of a thing that's virtualized running, you know, arm underneath it. Right. And so, um, especially, you know, now if you have specific use cases that require. That type of, uh, you know, like a certain instruction set that only Intel provides for performance, you'll ask for it. But I think, and those clouds have all those options available.

SPEAKER_155: All right, everybody, Steven Estes is a principal at CLA Clifton Larson Allen.

SPEAKER_157: This is a professional service provider that specializes in CPA tax consulting and wealth advisory. Welcome to the program, Steven. Thank you for having me. So at what point should a startup seek out professional accounting and tax services? I think there's a couple of tipping points, right?

SPEAKER_160: One of them is really as soon as funding and equity-based comp come into play.

SPEAKER_161: So whether it's a 500 case, you know, seed or pre-seed or a $10 million Series A, I mean, as soon as you're, you're raising and you're looking to hire talent and give equity to those people, you should have quality advisors for both legal and tax. And like I said, another, another marker really is that foreign activity, right? Having a foreign subsidiary or foreign founders, you can really find yourselves in hot water real quick if you don't know what you're doing in that regard. Because all companies have to play by the same rules in the international sandbox as whether it's a startup or Coca-Cola.

SPEAKER_162: Get started right now at claconnect.com slash tech. Let them know your boy, Jake, how sent you.

SPEAKER_163: Claconnect.com slash tech to get started right now.

SPEAKER_165: Let's go back to Mishra for a second. I thought this was fascinating as well. Yep.

SPEAKER_15: I just ran this live while we were sitting here. So, you know, so the folks that were watching, you could see that's the timeframe.

SPEAKER_165: Yeah. But okay. So, uh, suffice to say, you run it on your own servers, you could go faster and it's gonna just get faster every week.

SPEAKER_00: And then, you know, these open source projects can sometimes have such a diversity of talent in them. And because they don't have a command and control structure, they're gonna have more interesting insights, right? That's the nature of open source projects. They might have these four or five, you know, developers in South America and then these 12 in, you know, Japan, and then these six in Ukraine. And they all have some other use case and they contribute to the model in a way that's different than if Microsoft or Google or open AI or building software. They don't need to ask for permission to work on some part of the of the model or the open source project. Is that correct?

SPEAKER_171: Yep. Yeah. I mean, not at all. It's permissionless.

SPEAKER_00: Yeah. Yeah. Now there might be some permissions that occur when you do want to, you know, actually commit those changes.

SPEAKER_43: Exactly. Yeah. So to go now, usually how these projects are governed is, you know, they, they, you know, like Bill was mentioning, they have organizations that are built to, you know, kind of decide, like, you know, where the direction of the project is going. And so those then directions, then basically the developers on the open source project use those high level, like, think about that as roadmap and vision and say, oh, we want to build this next. And everyone aims towards it, but nothing stops you from taking it. If you don't agree with that and doing something and then some, you know, submitting it for commitment into the project and either they can take that or you can fork it.

SPEAKER_173: Many projects have been forked, right? And so, yep. Um, that's, that's always your option.

SPEAKER_70: If, if the, if the core project doesn't want to go off on your side quest, you can just make a copy of any, you fork it and. Exactly.

SPEAKER_82: You, you then start a new project. And we've seen that happen over and over and over again. Okay. Let's get back. Yep. M-I-S-T-R-A-L. M-I-S-T-R-A-L.

SPEAKER_177: M-I-S-T-R-A-L. Yeah.

SPEAKER_70: Got it. Okay. Now you mentioned how many, uh, parameters. Let's explain that for, uh, civilians listening.

SPEAKER_00: Yeah.

SPEAKER_131: Yeah. So the way to, the way I kind of put this into like, into simple thinking is more parameters

SPEAKER_43: equals, um, like parameters are like neurons in a brain. And so if you look at a small organism, um, it doesn't, you know, has a very small brain, doesn't have a lot of neurons. I think it's generally considered that humans have somewhere between 40 and 80 billion neurons. Mm-hmm. And so, um, when we think about it, the more, um, neurons exist, the more data that this has access to that, not, not from like what's what, well, it started from what it was trained on, but that data is then processed and held in kind of these weird fragments that we've talked about before. And so the more of that that's there, the more it's able to basically reason, thank, provide you really, you know, incredible results like we've seen. Um, and so these folks at Mistral have, have done an incredible job of creating a smaller model that performs like the big larger ones. That's, you know, probably in due to some of the technological, um, approaches they've taken and also perhaps even the training data and, um, and they shown that in, uh, like knowledge and reasoning and, uh, across these different tasks, uh, which, which, you know, math code, uh, which is really incredible. I thought this was very exciting.

SPEAKER_00: So the exciting thing here is, you know, people are taking different approaches. They're making, um, you know, these language models more efficient and faster, smaller, better, cheaper, all of those things. And how many language models, uh, is the open source community grinding on right now? And when I say grinding, like making daily progress on like, you know, of the, you know, it was, it could be some projects that are abandoned, et cetera. So like listing all the language model projects on GitHub or wherever hugging face, that's not productive, but let's say major ones that are getting daily updates to them. Right. Cause daily updates would, would be a sign that this thing is cooking.

SPEAKER_43: So how many are there, there's probably under 10 that are like, uh, and so what we really have to do is maybe take a step back because the space is playing out in a really unique way. Um, the folks that are, there's a less set of people that are going after really, really large models models like open AI that are general intelligence models, right. Where more of the energy is going as smaller run in, you know, kind of more confined, uh, you know, compute requirements. And that area has a lot of models, like too, too many to count. And what many folks have realized is that it's a better approach to go after a smaller model, um, that's, you know, tuned or trained for specific tasks than to try to compete with, uh, general purpose models. Yeah.

SPEAKER_00: And so that means that when we talk about this competition and regulatory capture, if you're open AI, if you're anthropic and you want to lock all this down, if you're a Reed Hoffman or Mustafa, uh, or Sam Altman, you know, Greg, whatever you're saying, Hey, slow it down. Trust us. We're going to protect everybody. You then have, they're under attack that, Hey, I'm going to make a verticalized one. That's just for audio. Just for video, just for, um, code, just for, you know, a specific language or a specific culture, whatever the vertical is, you're going to see, um, you know, these large language models be attacked by a thousand cuts, uh, by verticals and by thousands of people contributing to an open source model is the goal to do smaller, more nimble models like, uh, and have the best model with the least parameters. Is that like, uh, an attack vector here that people are trying to make these things smaller and more efficient and cheaper to run?

SPEAKER_43: Yeah, I, I think that's the ultimate goal because what we've already seen in the last year and, you know, with the rise of Nvidia stock was these really, really large models require, you know, compute and energy, like less people talk about the energy, but compute and energy.

SPEAKER_15: When you have like a cluster of 10,000 or a hundred thousand GPUs, the amount of energy it's using is really substantial.

SPEAKER_11: I wonder what one of those costs to run a day, like one of those GPUs, you know, an H 100 per day.

SPEAKER_138: If it's actually doing work, it's doing jobs.

SPEAKER_43: So maybe, uh, maybe the guys can look at this, but honestly, I think it was like the amount of

SPEAKER_15: electricity more than like a house. It was like something really substantial that, uh, we can probably research in the background here.

SPEAKER_00: It might cost a thousand dollars a month or something, you know? So if these things cost a thousand dollars a month to run. Yeah. Yeah. And you've got a thousand of them. That's a million dollars a month. It's 10 million a month in energy costs for 10,000 of them, which I think is what some of these big clusters are now doing. Yeah.

SPEAKER_203: So it's not, it's not, that's not nothing.

SPEAKER_00: You know, it's like 120 million dollars a year in electricity costs on top of these things costing 40 K each or whatever they cost. Yeah. Yeah. Yeah.

SPEAKER_149: The energy costs might be 25% of the yearly costs. So over four years, it might be the same. Yeah. It's crazy.

SPEAKER_43: Yeah. So I was just like, I'm looking at it here while we're doing it, but it basically is saying like, um, an H 100 card runs at 700 Watts. Right. And so think about that 24 seven, that turns into like a really, really big night. That was basically turns into, I would say my guess is close to a megawatt a month.

SPEAKER_82: Yeah. So we have to then figure out what a megawatt a month cost. And here's from Reuters.

SPEAKER_00: Um, running chat GPT is very expensive for the company. Each query costs roughly four cents, according to an, an analysis from Bernstein. If chat GPT queries grow to a 10th, the scale of Google search require roughly 48 billion

SPEAKER_84: worth of GPUs initially in about 16 billion worth of chips a year to keep operational, which is just crazy. But I think this is going to plummet, right? It's going to go down 50%, 90% a year. So, uh, and yeah, I, I, I get the sense this is going to be, you know, if it's 4 cents a query, it's going to be 0.4 cents and then 0.04 cents. And then nobody's going to even think about it. Kind of like storage, uh, became so de minimis.

SPEAKER_00: But the energy cost is pretty crazy. Um, there's been reports that everybody's making chips. So open AI supposedly is looking into making chips.

SPEAKER_118: We heard last week that part of, I think the anthropic deal was that Amazon was making their own chips and they wanted anthropic to use it. Obviously apple makes the M one, the M two, whatever's the, a, whatever they're up to a 14, 15, 16, whatever's in the, the iPhone. Uh, so we now have, everybody's going to make their own chips.

SPEAKER_149: I heard Google is going to make their own chips too. So now we have Google has been doing it for a long time, right?

SPEAKER_178: They have the TPU, right? Which is, uh, yeah, they've been doing that for a very long time.

SPEAKER_00: There seems to be some intense, uh, ramp up of this because not only can't you get Nvidia chips, there's a line out the door for them.

SPEAKER_70: Um, people maybe don't feel great about the pricing of them. So now everybody makes their own chip. So that is something I didn't anticipate.

SPEAKER_23: Um, and so here's the headline from Friday, Microsoft to debut AI chip next month that could cut Nvidia GPU costs. I don't, and that's from the information.

SPEAKER_131: So I started my career in, um, actually making chips, but on, you know, for networking equipment.

SPEAKER_43: And this happened in networking, uh, at the late nineties, uh, everyone needed their own chips to basically interface with the immense growth of the internet is on the sort of the core side of the internet with optical, you know, with fiber optics, right? And so everyone was making chips there and there was a time and place, you know, somewhere between say 97 and 2001, where there would have been like hundreds of chip companies making, uh, front end chips to interface with, you know, optical transceivers that ultimately all consolidated down to maybe like three companies. Uh, and, and it happened for the same reason, like no one could get these chips. Right. And then no one could develop them in ways that each particular vendor required. And I think we're just kind of seeing that movie again here, and we'll see an explosion, which will bring the cost down. And then ultimately I think it'll consolidate as well. But those are the waves that we see in tech anyways, right?

SPEAKER_220: Where we see like expansion and consolidation and expansion again.

SPEAKER_00: Uh, and so a massive amount of investment and then it becomes commoditized or so cheap that people may be some number of players bow out and let.

SPEAKER_221: Or roll up and yeah, exactly.

SPEAKER_00: That's another way. Yeah. I mean, and we, we saw that I predicted this in GPUs because we saw it with fiber. People were building out so much fiber that they overbuilt. Yep. And then I think Google and some other providers bought up a lot of that fiber. Um, you know, pennies on the dollar, uh, that had been overbuilt. So we might be in the overbuilding overbell phase. Um, what other demos do we have for this week?

SPEAKER_43: Um, yeah, so I just had like mist roll, which, you know, we, we can run through that one. Actually, like we just, we, I had the paper there, so I can pull that one up. And so this is their playground on, on hugging face as well, where we spend some time. Uh, all right, let's, uh, give me a, give me a drop in here.

SPEAKER_178: J Cal and I'll send you this link too. If you want to try one, give me a question here. I got this question.

SPEAKER_225: Well, you know, um, how about what are the best, uh, restaurants in Napa?

SPEAKER_00: That's it. I mean, this is like some live information who knows, uh, where they got this information. Um, but it's screaming fast, even on hugging face. I mean, instant answer. So that was, um, that's the first thing I'm noticing is that this is. And, um, yeah, uh, it, it of course, uh, got it exactly right. Bouchon, French laundry, soul bar. I mean, ad hoc. Yeah, these are all great ones. Uh, but tag, I know a bunch of these.

SPEAKER_118: Um, and if you asked it, um, uh, what, uh, add, uh, put those in a table and add the average

SPEAKER_70: cost of dinner.

SPEAKER_118: I wonder if it has anything there.

SPEAKER_236: I don't think so.

SPEAKER_118: Uh, you know, that would be like more live data, but it did pretty good here, you know, depending on what. I would think so.

SPEAKER_236: Yeah. I don't think it'll have that, but like, yeah.

SPEAKER_43: So, uh, yeah, but you know, that's a tough one, I think even for chat GPT, but what I really liked about this one, and I'm glad you did this question was like it, and this is what we're going to get as we put these out there. It, first of all, kind of says, Hey, in order to provide the best, but here are my assumptions, right? Then it talks about, Hey, you can define best. And here's like a combination of food quality, good service and that, and then target audience and research how it got there. I really, really am a fan of like, like you said, the speed. You could run this on your laptop if you want to.

SPEAKER_178: Right. And I think which is really incredible here. Yeah.

SPEAKER_165: So you could give this language model to, uh, an astronaut, you know, going to Mars.

SPEAKER_00: And if it lost contact with the internet, it would still be able to give those answers, which is just for people to think about this. If you were stranded on a desert island, having, uh, one of these language models fully trained on a laptop, you would be able to have like an increasingly, um, impressive conversation that could wind up saving your life. If you asked that how to start a fire on a desert island, what would it say? Yeah. Yeah. I mean, actually ask it, you know, I'm, I'm trapped.

SPEAKER_82: If you were trapped on a desert island, how would you, on a desert island? How would you start a fire? Now, let me bring it back.

SPEAKER_00: If you didn't go to survival school and you had this, all of a sudden you've got this bot with you on some foreign, uh, location. I wonder if it will actually even help you build that fire. Uh, you could, you find a piece of glass is one technique. Rubbing two sticks together is another, uh, obvious, uh, technique trying to find flint or stones or metal that could spark something. Um, gather material, collect dry leaves, small branches, of course, that's kindling. Find a suitable location. Okay. Create a fire pit, prepare the tender. Okay. This is all the same stuff. Kindling, prepare fuel, create a fire layer structure. Okay. Ignite the fire. There are several methods to start a fire, such as using flint and steel, magnifying glass and, or a battery. I forgot the battery one. Um, so that's interesting. So tell it to, um, explain to me the techniques in number 10. Let's see if it understands that we're asking it about its existing answer. Cause even steps one through nine are preparing a fire.

SPEAKER_42: Uh, putting the wood together, but not actually sparking the fire. What we care about is actually starting it. So I wonder if it will be able to teach us how to start a fire.

SPEAKER_211: Oh yeah. Here we go. Flint and steel. This method involves striking flint against steel rod.

SPEAKER_84: Okay.

SPEAKER_00: Magnifying glass on a sunny day focus. Sunlight, uh, onto the tinder using magnifying glass battery and steel wool. That works. Yeah. Uh, lighter matches. Yeah. If you have them.

SPEAKER_84: Um, it's pretty good. Pretty good answer.

SPEAKER_105: Um, and what else do we have in the, uh, in the, in the demos? Everybody loves the demos here. Yeah.

SPEAKER_43: So last week, we kind of, uh, right at the end, we got into, uh, WhatsApp and the model and the AI. And I think, uh, one of the things I wanted to correct, so I wanted to pull WhatsApp back up because, you know, it is available in the group chat. So I was getting a second here. Yeah. So you can do a group chat and include an agent. So, cause that was one of the reasons you didn't give it, uh, uh, uh, even a higher grade last week.

SPEAKER_178: Hmm.

SPEAKER_00: Yeah. And I loved it last week. I was super impressed. Um, and I just thought, you know, if my wife and I had a, a chat going and we had slash Gordon Ramsey and we could, you know, ask Gordon, Hey, here's what they, you know, here's what we have. What should we make? Hey, we got. You know, some salmon and we've got some pasta, uh, you know, we've got butter and cheese. And it was like, okay, yeah, make some far folly with salmon in it. Here's a recipe for you. Yeah.

SPEAKER_178: So, um, this is our AI, you know, twist, uh, WhatsApp group that I created. Let me get this over to the side. And in here we have the meta agent, right? And so, um, and if you can see here, you can do this sort of an at, and it allows you to add meta AI and say, I am thinking about booking a trip.

SPEAKER_239: To Napa. What are the top five hotels?

SPEAKER_256: Okay. Here we go. Let's see.

SPEAKER_239: And what's great about this is like, I could then ask a follow up to it. To respond back in a second. Maybe they've, they've gone slow as well. Now, there we go.

SPEAKER_258: Well, no, here we go. Uh, yeah. Our Bosch, uh, Meadowood, four seasons. Yeah.

SPEAKER_186: I mean, it's not great. It's not terrible.

SPEAKER_258: Yeah. Not, not, not great. Not bad. What's a fun activity for the three of us to do while in Lake Tahoe over Christmas.

SPEAKER_105: I mean, it's pretty obvious what you do in Lake Tahoe over Christmas, right? Uh, I consider ice skating, uh, taking a sleigh ride, uh, enjoying a festive atmosphere. Yeah. Um, yeah.

SPEAKER_264: We could have also gone skiing, but we're snowmobiling, not a great answer.

SPEAKER_122: Not a lot of snow at Christmas sometimes though, but maybe it knows.

SPEAKER_00: Yeah. Uh, how about more answers? Additionally, you can explore Emerald waters of clear kayak court, not in the, not in the winter. You could do that. Uh, Lake Tahoe over there. So, uh, you can see the, the meta AI is like a very rudimentary AI, I think. I'm not sure who, um, I'm not sure which model that is.

SPEAKER_42: You think that's Lama? It must be, right? Is there a language model?

SPEAKER_268: Yeah. It's, it's a Lama, right? Yeah, it's definitely that one. Yeah. Yeah.

SPEAKER_23: Yeah. I think that leaves a lot to be desired right now. I think vert, this is where I think verticalized AI is going to be much better.

SPEAKER_00: When you ask about travel, you really need a travel AI. That's just that, right? Um, so I, I think that that's where these models are going to wind up is that they'll be very verticalized and fine-tuned ones by vertical. And if you ask about food, it should really be narrowing you down and not just giving you the generic one, it should be just like when you search now for recipes on Google, it doesn't just give you 10 blue links. It really is thoughtful and they have a lot more information on it.

SPEAKER_11: Listen, there's been another amazing episode. Thanks for bill Gurley for tuning in. It's pretty clear. Uh, the community must fight for more open source.

SPEAKER_00: Anytime anybody says, trust us, uh, we will be the sole source. Uh, you should be wondering, uh, uh, if you should trust them and just, uh, listen, you need to fight for open source and you don't want one private company. It's not a dig to any, you know, it's not a dig to Sam. It's not a dig to anthropic. You don't want any one party controlling a platform technology. You want it to be open source. You want closed source solutions, private company solutions, open source solutions. You want, uh, a range of opportunities. And remember the last time we had some founder say, uh, trust me, I'll, I'll get us some regulation. That was SBF and he was, uh, he was going to, uh, be the one who got us federal regulation for crypto and, uh, he's on trial at this very moment for a bunch of Meshuggah. I mean, are you following that? Sonny, you were down the crypto rabbit hole. Uh, yeah.

SPEAKER_131: And, and, you know, uh, very deeply and following it.

SPEAKER_43: And it's really sad to see, you know, I think one of the, you know, I think one of the more, there's a lot of shocking things. So not, not to say, but one of the more shocking things I saw last week is they were running some kind of, um, insurance service. And there was like a counter that was saying, oh, like this much money is insured by this. And that counter was basically like a random number generator.

SPEAKER_268: That was like picking a number between like, you know, 7,500 and 10,000. And like, I mean, talk about explicit like fraud. I, there you go.

SPEAKER_275: Exactly.

SPEAKER_118: I mean, it's deranged. Your point is if, yeah, to say that, like, you know, this, um, was a real business and, you know, um, that they were just in over their heads when they were doing free, meditated things like this. And this is where like all the chats eventually get dumped, all the emails, all the slacks, everything, everybody winds up flipping no matter what the case is. It's very rare that a group of people will circle the wagons. Like even the mob, you know, and the mafia, you know, had a hard time maintaining that. And this is like their entire lives, their families, their traditions were around this. Right. Yeah.

SPEAKER_00: Um, maybe the cartels on the margins can, but not a bunch of dopey kids who've been working together for 18 months. They were committing fraud after fraud. And in this case, they were using a random number generator, you know, with some parameters on it to do the public when they were getting insurance.

SPEAKER_105: Yeah. If I'm understanding it correctly. Um, and listen,

SPEAKER_273: like it takes, you know, like a, a certain amount of evil to do that. Right.

SPEAKER_268: Because you're, you know, there's lots of cases. People have gone sideways, like, you know, Theranos, Elizabeth Holmes. And, you know, at the core, I think she was trying to make a, you know, like a testing machine. It's just, uh, kind of. Yeah.

SPEAKER_70: That's the case of it maybe got ahead of you, right?

SPEAKER_00: You got ahead of your skis. You were a bull artist. You thought, ah, you know, I can, I can fake it till I make it. This wasn't fake it till you make it. This was let's orchestrate a huge crime. Um, and, you know, listen, if you're in the mob and you flip, you get whacked. With these kids, like if they flip, there's nobody there to whack them. It's not like SBF's parents are like some criminal mastermind cartel. They're a bunch of dopey. Yeah. Uh, Stanford professors who, you know, didn't raise their kid correctly. And, uh, you know, they're not going to whack the other members.

SPEAKER_118: They're all gonna flip on each other. It's all gonna come out. I, I told everybody I will bet dollars to donuts that he gets over 30 years. I'm saying over 30 years.

SPEAKER_136: One thing.

SPEAKER_288: We take the over the under 30 year sentence.

SPEAKER_289: I would take the over. Yeah. If I said 50, we should take the over.

SPEAKER_264: Yeah.

SPEAKER_288: I think under, I think. Okay.

SPEAKER_264: So 40 would be the line. We have to think it through. What would you take at 40?

SPEAKER_291: Pardon? We take the over the under at 40 years.

SPEAKER_293: I think the under, I think it'll come right, right around that. Okay, there we go.

SPEAKER_291: Wow. Look at that.

SPEAKER_288: I said a pretty good line. So the line is probably 36 years.

SPEAKER_268: You know, he gets a before retirement kind of thing.

SPEAKER_84: You know, and then I, I think they're gonna Bernie made off him. You know, Bernie made, I've got multiple life sentences. So I set the 30 year, but I actually think there's a chance he gets life. I think this could be a life kind of situation.

SPEAKER_70: Well, you know, there's, it's very rare that you catch somebody red handed. Yeah.

SPEAKER_39: Doing a multi billion dollar, uh, crime. Yeah. There there's a small number of multi billion dollar crimes. In the history of humanity.

SPEAKER_296: Yeah.

SPEAKER_39: It's hard to pull off.

SPEAKER_11: And so if you're going to put, you know, some, you know, uh, kids in a local city who were dealing drugs, 20 years, 30 years, and they were dealing, you know, a half million dollars in drugs, a million dollars in drugs. Let's just call it a million dollars in drugs, a million dollar drug, you know, cartel, you know, in Chicago. And this kid was stealing over a billion. I mean, it doesn't feel proportional, right?

SPEAKER_298: If he, if he were to get any, anything, but a magnitude more, because it's a thousand next, that crime.

SPEAKER_222: Yeah. This is serious crime folks. Yeah. I don't know. Yeah. I mean, you're, you're spot on there. I just don't know how to like do that relative calculation.

SPEAKER_268: I mean, like there's people in, in jail today for marijuana use that, and you know, it's, it's, it's, it's allowed now, right? Yeah.

SPEAKER_301: No, luckily we're, we're, there is a consensus.

SPEAKER_11: Trump, Biden, Obama, before everybody is, you know, aligned that we need to reverse those for nonviolent felons. Um, and really think about those. All right. Listen, another great episode. And, uh, you know, listen, um, he's guilty. I've made my decision already.

SPEAKER_148: The evidence is there.

SPEAKER_11: The jury will make their decision, but, uh, I hope he gets a, a, um, a huge sentence. Uh, and, and that's it. All right. Sandy Madra. They have it folks. Definitive intelligence. If your company is looking for AI and analysis of big data and you need help. Sonny's your guy.

SPEAKER_306: Just email Sonny at definitive.io. And we'll see you all next time. Bye. Bye.