Latest / Systems and Power / Is Big Tech BUYING Elections? The AI Race and Open Weight Models - Niel Pal | Systems & Power #17
Transcript
- Moses Parente: Alright, welcome back to the Systems and Powers podcast, guys. Thank you again, Neil Powell, for joining us. Welcome. Niel Pal: Happy to be here. Moses Parente: Of course we got lots of to talk about, especially with AI news, big tech news, and especially midterms are coming up and a lot of people are talking about is big tech buying out our elections? Niel Pal: Yeah, I know that there's been a lot more investments that they're making into the space. I think you saw in some of the all-in podcasts, Jensen, very friendly with the Trump administration. So yeah, I mean it is interesting. I don't know like to what extent they're making these direct investments. I'm sure it's through like different packs, but yeah, it's definitely something to keep an eye on. Moses Parente: I think as long as I mean, a lot of their main goals are basically like, hey, we need to win the AI race against China. So it's all about we need legis people in the legislation who are as pro AI as possible. And largely Republicans have been very historically Niel Pal: Definitely. Moses Parente: anti regulation f so in terms of the AI boom. They just want to take advantage of this as much as possible and just I guess focus on making sure that they have at least as much pro AI legislators in the Senate and the House of Representatives to make sure that hey they can pass some stuff. But with the right now on I think on polymarket right now the Democrats are slated to win at a ninety percent chance Niel Pal: Yep. Moses Parente: to take the House of Representatives I think they have an eleven point lead as of right now as we're talking. Niel Pal: Yeah, I mean with the inflation that's happening right now and yes, maybe somewhat over corruption, it makes sense that the Democrats are leading. Again, also too it makes sense that these big tech companies are leaning into politicians that are pro AI, specifically for data centers. I still don't fundamentally understand why Democrats are so opposed to building out data centers, but maybe a conversation for another day. Yeah. Moses Parente: That's that's definitely gonna trigger a lot of the more populous people on the left. Niel Pal: Yeah. I think one thing that was pretty f I guess unique for me was just like this rampant there's a huge problem with AI that open AI was trying to spread and same with anthropic, that we need to do safe, you know, AI build outs. I think largely that is a dog whistle against open weight models. They don't want you using the models from China. That's why we have to all care about regulation. And of course, right, the ones that are gonna be cozy with the US government, which is not gonna be the Chinese models, they're the ones that are gonna get pushed by the government as something that's safe. Moses Parente: Do you do you think that we should be, at least in the United States, be more open to using open weight models? Niel Pal: Yeah, so I mean it's obviously way cheaper, it's a lot more efficient, and if you have your own data, right, which data's now is such a huge market, with that data you can fine-tune it and it's much more efficient to run on your existing systems. So I think yeah, we should be using open weight models. It's a little concerning that we have to fully rely on China and there's no real equivalent US-based one. I know NVIDIA is trying to make some moves with their acquisitions, trying to get in that space, but by and large China's definitely the leader. Moses Parente: Why do you think currently, like I guess at on the in the US they're a lot more opposed to like using open weight models? Niel Pal: Well, I mean in China, right, those are the existing models, right? And the government wants you to use it, they get more data, training data obviously, so it's helpful. it ends up just being the best possible option, and which companies like Coinbase, their CEO, actually mentioned in a post. The majority of their workloads workloads are now using Moses Parente: Yeah. Niel Pal: open weight models, right? More companies are gonna follow suit. But again, it is tricky, right? And you will always especially for like I believe DeepSeek you mentioned was now banned for some federal government kind of workloads. Yeah. Moses Parente: Yes. Niel Pal: There will always be a concern, especially with China being your main provider for these models. Moses Parente: So in terms of open open weight models, you said you think that more companies are gonna be following suit, to kind of like Coinbase as a as you mentioned. Do you think that this will be something that would lead to the US losing the war war with AI or no? Niel Pal: So I wouldn't say that. In terms of it being used, yeah, these open weight models are gonna used more. What that's gonna do is actually lead, if you look at the inference numbers, people are gonna be using AI a lot more, but their costs of tokens are gonna go down. Which just naturally leads to more building and great products being built in the US. So I think that this may be not so great for Anthropics, bottom line, and OpenAI's bottom one, hence why they're pushing so hard for we need safety. There's a huge problem going on. Right? It's scary for them, but they're massive multi-billion dollar businesses. And in the long run, I think it's actually gonna be much better for the US economy. Moses Parente: I was talking with some some people about the upcoming Anthropic IPO that's gonna be happening in November Niel Pal: Yeah. Moses Parente: with some people from the Manhattan Institute a while about a couple of weeks ago. And they were making the argument about why that anthropic they're pushing so hard for AI safety, largely because we want because put all a lot of these other companies following all these compliance rules are going to be extremely costly. But because Anthropic is so big, they're because enforcing all of these compliance rules on all these other smaller companies, it's essentially creating a mon a monopoly where they're at the top and everyone else is at the bottom. I would love to get your stance on that. Niel Pal: I mean I agree. Yeah, I think like when you're a large company with those resources, you have those relationships already built with the US government. I think Dario and you know, maybe Sam to as well, right? They're both friendly with the Trump administration. They're going to want that sort of lock in where we want to establish some sort of like safety and all these compliance requirements. Their other competitors won't be able to keep up with that. And naturally that allows them to charge more compute and more tokens for other customers. Moses Parente: 'Cause one thing that I've always thought about 'cause I I'm per I've I've been I'm on the board of Isaac New York. We've always been very put a lot of heavy emphasis on guardrails and I've been a lot involved in a lot of cloud projects in the past where we did do a lot of parts on when we were talk when we were talking about like and it putting safety when you're doing your deployments and all of that. And I wanted to ask like, okay, w should do we n is this all like do we actually need a balance for all of this? Or this is all s something about a part of a much bigger play? Because I I definitely Niel Pal: Sorry, Moses Parente: do think we need at least some sort of smart I genuinely believe we need some smart regulation that that that's going on. I think that having at least some sort of rules for everyone to follow is something that is beneficial and kind of like having a referee in the room. But then Niel Pal: Absolutely. Moses Parente: we don't want it to get to the point where a company is trying to be cast a larger monopoly on everyone else, you know? Niel Pal: Hundred percent, right. And I think it's like a g a give and take, right? no real solid answer, at least on my end. We'll just kinda have to see how the market plays out. But I understand the game OpenAI and Anthropic are trying to play here. But at the same time their point does stand. We need to have some level of guardrails and just like a way of setting up the rules so we can continue to build in a safe manner. I mean we've seen in these past what is the year ChatGPT came out? Twenty two I believe. Moses Parente: Yeah, like around November, I believe. Niel Pal: Wow, I mean think about that. It's been less than four years and like what a revolution it's been. You know, Moses Parente: Yeah. Niel Pal: who knows what these next four years look like and what kind of impact that has on the US economy, how we operate as businesses. So it's important to set up some right guard rules. I just don't know like to what extent that, you know, these companies are pushing for. Moses Parente: I also think it's very I I think it's a little bit ironic now. I mean I see from the the the Democrat side, they're pushing a lot of populist messaging heading into the midterms against data centers, against basically building in terms of big tech. And I understand why a lot of them are doing that. They're doing that in red states where there's a lot of land and places where people are trying to put up data centers so that way they can and It it it's like look, we need data centers in a way because we need to We need that to r to r to run our A AI models. But at the end of the day, like you can't they wanna have their cake and eat it too. They want the benefits of AI, but essentially Niel Pal: Yep. Moses Parente: they don't wanna have to pay up the upfront costs. And I do understand that it's good to look work with the local government and that it's good to check out like what the community needs to put these data centers in safely. But it's gonna be f I think a lot of that that anti-AI messaging is largely just populist slop so that way they can get elected and win as many seats as possible heading into the midterms. Because I guess when I mean the Democrats have been out of the out of power for the House for the past four years and now they smell blood when it's coming to when you see the Republicans, you see Trump's unpopularity, they want to do what they can to take the House back. Niel Pal: Yeah. I mean it goes back to the broader message of just like income inequality. And largely I do believe like that's some of the frustration we're seeing with you know, even back like a couple years back, Meta, right? They are capitalizing off your attention and this company continues to get more powerful. There ever was a competitor to Instagram. Look, that company's not gonna survive. Meta's gonna throw an insurmountable amount of money and they're gonna get bought out. Like these companies are so powerful. that there's a reason why you can understand in the general populace, people are against big tech, they're against these technology revolutions that are happening. Because they're not necessarily being part of it. I get it, I totally get it. But I do think that may not be the right message in terms of like what Moses Parente: Yeah. Niel Pal: is gonna be the future. For example, like you know me, I think we may be related to this aspect. I'd like to tow the line between center left, a little bit of center right. And there's a guy named Brad Gerstner, who's partnered with the Trump administration on a thing called Trump accounts. that I think is a phenomenal thing he's done, where he allows every single American, once they're born, they get a thousand dollars from the US government. And that sits until I believe age twenty, where it's able to compound. And if they want to withdraw it, they can. Right? Not a lot of money. A thousand dollars is nothing. But simply making that investment in our young people, transferring some of that capital from our these wealthy, you know, billionaires to just the general population allows people to become an owner in the economy. I think that ultimately could be the message we're trying to fight for. Moses Parente: you just reminded me of a point back when I remember Andrew Yang was talking about universal basic income. And that was kind of like what made him popular when he was running for president for the Democrat nomination. And well, a lot of people like to call that, that's socialism or whatever. But I mean we have Social Security and and that's basically socialized retirement. And I i in a sense of seeing that, a lot of people are now have are are talking about like how AI is now pr doing the whole universal universal basic compute and now that's le it could possibly be leading to universal basic income. Would love to get your your take on that with but like how AI is now I guess transforming and probably being able to provide that sort of opportunity for the United States. Niel Pal: Yeah, I mean to some degree, like we already do get universal basic compute. If you consider just like a personal plan, right? I mean look, for a twenty dollar a month plan you get on Cloud or OpenAI, it's very subsidized. You get a lot of value. And reality is like they're charging these enterprise contracts, which is how they're getting all this money, right? So you know, I do see we're us getting benefits and ultimately in terms of these companies really growing and thriving. and being adopted by all of American society. There will have to be some more, you know, giving back to society, making sure that it's adopted within these communities, people can see a safe way of getting that out. So yeah, I see that. fundamentally I disagree with Andrew Yang's universal basic income. Now as I've gotten older, you know, I lived in New York, I'm living in SF now. Capitalism is a I miss New York, man. Moses Parente: New York misses you, mad. Niel Pal: I mean, it is what it is. liking SF right now. But I think capitalism is just a very powerful driving force. And to go to the point where we do a universal basic income, to the point where everyone can have a sustained living, that in my eyes is probably kind of like a utopia. It's just in terms of the resources to go around, the in terrible bureaucracy we have in our government, where it's just gonna be misspent. And overall just the will of the people to continue pushing and driving our future forward. I don't think we're there just yet. Moses Parente: I understand your your point. But I do think it's funny that Trump is trying to, I guess, basically just buy out in terms of votes like with the five thousand dollar for everyone. I was like, bro, Niel Pal: Unbelievable. Unbelievable. Moses Parente: h what what is that? Niel Pal: Trump has just like so many sound bites of just ridiculous things. Look, it's hilarious. He's insanely clippable. Moses Parente: He's very clippable. Niel Pal: I don't know, man. Like I I still remember my first interaction with Trump, right? Like this was back when I was 13 years old. I went into like a bookstore in Seattle. And there was a book, The Art of the Deal, by Trump. At this point I had no idea who Trump was. I just know him as a rich guy. That was his brand and his persona. Yeah. Moses Parente: yeah. Niel Pal: First page, there's five typos. Trump does not care what you think. Moses Parente: Well you actually met Trump back when when you were do you were you saw the book or or or that was or Niel Pal: No, no, no. I I mean that would have been awesome to meet Trump, but no, I just read his book in the bookstore. Moses Parente: okay, okay. Nah, it's it's the The Art of the Deal was a very interesting book. I remember I read it like a long time ago. But i Niel Pal: Yeah, yeah. Moses Parente: it it's funny how I think like like you see basically a rich dude from New York who basically he puts his name in gold and has all these buildings and then now he's captured the hearts and minds of basically the Like the people in the South and Southern Republican voters. That's who his base is. And he has basically thirty percent of the country who are willing to vote for him no matter what. One Niel Pal: Yep. Moses Parente: of the truest things I think he said was that if I I can shoot someone in the middle of Fifth Avenue and everyone would still love me. That that's something that is genuinely true. There he has a essentially a cult of personality that is just Niel Pal: Yeah. Moses Parente: willing to go behind him no matter what. Niel Pal: Yeah. Yeah, I mean it is probably amazing just the amount of pull he has. I am a little disappointed on like a politics standpoint. Look, I think like I'm probably leaning a little bit more center left. But you know, given that he went and advocated for Ken Paxton out in Texas and Vivek out in Ohio, to see them both kind of lose in this race, ones where they should be clear cut favorites, I like that's caught me off surprise. Moses Parente: I think it's also well, number one in i if you see on red states, de Democrats seem to be performing there the best. And I think a lot of that is largely because a lot of the like the the war in Iran that's being that's affecting gas prices, a lot of those are affecting people who live in red red states the most, and the people who live there are kind of feeling the the brunt of it. But also some for Niel Pal: Yeah. Moses Parente: a a a take State like Texas, somewhere where if John Cornyn was still running, this would have been a clean sweep for him. But Trump is somebody who wants someone who is 100% completely loyal to him. He doesn't want someone who is ninety-eight percent loyal, ninety-nine percent loyal. He wants one hundred percent loyalty. And Niel Pal: Yeah. Moses Parente: that's what Ken Paxson was offering, and In the Republican Party. So I saw this on ballotpedia. Trump endorsed candidates who run in their repu their respective Republican primaries have a ninety-six percent success rate. And so that Niel Pal: Wow, okay. Moses Parente: that he he essentially has complete control over the Republican Party at that point, and he's willing to let it burn if it means people aren't being loyal to him. Niel Pal: Yeah. And I mean I guess it kind of shows that like, hey, look, Ken Paxson is not the most charismatic guy. Like it's overwhelmingly clear. But for you know yeah. Moses Parente: James Salarico, he's fun. Niel Pal: For Taller Rico to be leading in a state like Texas, you know, even with all Moses Parente: Yeah. Niel Pal: the insults JD Vance is throwing, Trump is throwing, it shows like the power of charisma, which ultimately for me is probably gonna be the most important thing in the politician. You know, it sounds a little bit ridiculous to say, but like The ability to kind of enthrall an audience, to get people to buy into a message. I saw that with the Obama administration, you know, obviously a few presidential elections back. we saw that with Zoran right now. Daniel Lurry, who I just got to meet in SF. Moses Parente: nice. How how was like how was it like meeting him? Niel Pal: He's act you know, he's he's a charismatic guy. he does these like TikTok videos, which find maybe a tag corny. He just Moses Parente: Ha ha. Niel Pal: goes up to like local restaurants, he'll do like a 30 second spiel interviewing the owner, he'll grab a bite, then he'll head out. But got to meet him there, Sewer Cafe. Moses Parente: That's nice. Niel Pal: If you guys haven't been in SF, definitely should go. but yeah, got to meet him there. I am interested, there's a guy named Gary Tan. Who's president of Y Combinator? He's teamed up with Daniel Lurie and trying to make, I guess, SF not as much of like a hardcore Democrat place. I know in New York there's like the DSA movement has really kind of picked up. SF politics has always leaned very hardcore left. Whereas now I think maybe since post-COVID, there's been more of a push towards moderate politics, which I think if we look at the numbers, has made the city a lot safer and a lot more economically efficient. Moses Parente: Speaking of you just reminded me about the about something that a lot of people have been fans of like when when we talk about this subject. Prediction markets. Espe Niel Pal: Yes. Moses Parente: especially since I mean I mean we all know that Don Jr.'s on the board. Kind of I think he's on the board of bo of both Calci and Polymarket. Niel Pal: That's sure. Moses Parente: But I know I know that there's something very interesting that I know that you wanted to cover regarding prediction markets. We'd love to talk bring that up if if you're fine with that. Niel Pal: Yes. Yeah, let me share my screen. So this is a tweet. you can see right here, polymarket had the approach where they wanted to be more international. They I know they started off maybe growing in France first, if I'm not mistaken. And they were always gonna be crypto native, that was part of their brand. In terms of getting widespread adoption though, I believe that you really do have to crack the US market, which Calci chose not to do. Calci chose to go and really fight a lot of US regulation. You can see up to this 2020 sorry, what is the year Trump got elected? 2024, yeah. yeah. Moses Parente: Yes, second term. Niel Pal: Up till that 2024 election, they were not able to operate in the United States. And that was really the calling card. See, you know, they grow a little bit, but that compounding effect of being in that presidential election allowed them to get sports betting Moses Parente: Yeah. Niel Pal: legalized, which, you know, since then, you know, the writing's on the wall. They're absolutely crushing it. Moses Parente: So essentially because Calci decided to pursue getting access to the United States market first and fight all those legal battles with the the legislation with the United States, that kind of propelled them in their lead right as of right now to polymarket. Niel Pal: Absolutely, yeah. And you can see in twenty six, you know, they're starting to get some people on. I believe Trump or sorry not Trump. LeBron James. He's a big polymarket Moses Parente: Mm. Niel Pal: advocate. So it's widespread in our sports sports world right now, but like it's a small slither compared to what Cauchy has done. Cauchi in my eyes has done a phenomenal Moses Parente: Yeah. Niel Pal: job in terms of marketing and really penetrating and becoming a part of the US market. Moses Parente: yeah, especially I remember with the the whole Knicks when the Knicks game was going on back in the summer. I'm pretty Niel Pal: No yeah. Moses Parente: sure you you saw it too. They were essentially I don't know if you saw they had this kid who essentially he s he screamed on the Calci micro microphone. my mayor Muslim, my bagel's Niel Pal: Yep, yep. Moses Parente: Jewish, the Pope's on our side, Nixon Niel Pal: My Christian Dior. Moses Parente: five my my cr my Chr my Christians Dior, ni Nixon Nixon five or something like that. And Yo wait Niel Pal: Yeah, yeah. So there's Nixon four and then you have to redo it for the Nixon five. Moses Parente: yeah, my my Christian Zior at Nixon four. Unfortunately, Trump had bad juju and he came and he he he Niel Pal: Absolutely. Moses Parente: he ruined the whole game and he slept through it. But essentially, when yeah, it was our Pope's on our side, Nixon five. That was after they they lost that one game. But hey, go Knicks. And essentially They took advantage of that viral moment and they had the the Calci mic on it and it just like blew up even more. And Niel Pal: Yeah. Moses Parente: I think that with pr like basically like prediction markets are kind of the the whole it's n we can't even call it the future anymore. Prediction markets aren't the future anymore. Prediction markets are the present. That's I think what the direction is heading nowadays. Niel Pal: Yeah. Yeah, no I agree. by the way, Moses Parente: Yeah. Niel Pal: did you watch any of Dream Force at all? This is Salesforce's annual conference. Moses Parente: I saw I saw clips of it on X and LinkedIn, but I haven't seen the whole thing. Niel Pal: Okay. Yeah. So I remember earlier in twenty twenty six, so not long ago, January, there was a thing called the SaaSpocalypse, where all these big enterprise software companies, Salesforce is one, ServiceNow is another, it's took these massive tumbles. Since then, they've actually rebounded pretty well if you look at ServiceNow's numbers. I knew ServiceNow's Moses Parente: Mm. Niel Pal: a business was always good. They had partnerships with NVIDIA. So like they would be good. But that Dreamforce announcement is kind of like the we're here sort of moment. The SaaSpocalypse is fake. And enterprise Moses Parente: Mm. Niel Pal: software, you know, is slow moving. It traditionally did it on a per se basis, which doesn't make sense in the world of AI agents. Right? You want to have something headless, which they've now pivoted towards. these companies are gonna do pretty well. and particularly, I think, Salesforce, what's been exciting for me. I had a couple friends at a company called Listen Labs. I don't know if you've heard of them. They did like an AI market researcher. So anytime, say you're like getting off of anthropic. You know, they will actually have a thing at the end like, hey, you know, we would just love to learn more more about your experience here, maybe why you're going off. And it's like an AI interviewer, totally powered by Listen Labs, which recently got bought for two Moses Parente: Yeah. Niel Pal: billion by Salesforce. So Moses Parente: wow. Niel Pal: MA's heating up right now, and I think we're gonna see more more hot Moses Parente: Yeah. Niel Pal: acquisitions happening. We already saw some with NVIDIA, right? They bought Poolside and then recently followed up with Hugging Face. Moses Parente: What did poolside and hugging face offer get if you don't mind me asking? Niel Pal: Yeah, let me search a pool side. Hogging face went for fourteen billion. And I know Kevin Moses Parente: wow. Niel Pal: Durant, he got some headlines for there as well. He got he was one of the first investors in Hugging Face back when like they were like a video game thing. Wow, okay. So, they did not actually do an acquisition. They did a licensing agreement, which is kinda similar to what I mentioned with cognition. with like cognitions or sorry, not cognition. Windsurf, couple podcasts back. They did a deal Moses Parente: Yeah. Niel Pal: with Google and this is a similar thing. they did a seven billion dollar licensing agreement, so there's quite a bit of money going into this space right now. Moses Parente: Mm-hmm. Niel Pal: Yeah. another another funny thing too, I don't know if you got to see this. Did you see the new cognition CRO, Chris Denion? Moses Parente: I saw I saw parts of it on on X but I don't have the full story. Okay. Niel Pal: Okay, let me give it to you. So yeah, I used to work at a company where we're pretty tied sorry, tied heavily into the Snowflake ecosystem. Snowflake has had a pretty awesome kind of go-to-market story. There's a guy named Chris Denyan and another woman named Denise who formed up their sales and marketing compo. So at this point, you know Chris took Snowflake to the largest IPO in history. They're they did relatively well in the public markets. And Since then he retired, he went on some podcasts, he actually teamed up with a guy named Chad Petez to do hiring for a lot of these companies and like serve as an advisor. And one of them was Moses Parente: Yeah. Niel Pal: a company called Factory, which I personally thought was interesting. Traditionally, with these model companies, it was always about the model capabilities and how powerful these models are. Generally, what we're seeing right now is that maybe there's just less of a push and more of a push into like different integrations. And that AI agent taking on autonomous tasks for you. So that was something cognition definitely led the wave in. And Factory Moses Parente: Yeah. Niel Pal: was another up-and-coming player in the space. They had raised quite a bit of money. And in fact, they were probably the heavily advis most heavily advised by Chad Peace's Moses Parente: Hmm. Niel Pal: team. So it was quite the shocker on the internet when Chris Danyon decides to step down from retirement. and being an advisor at RPT to now join the CRO of his number one competitor to that company factory. So there's a there's a bit of controversy. I'll show you the Vinode Kosla tweet. Moses Parente: Yeah. Niel Pal: Did you get to see that? Moses Parente: sh sh show it on the show it on a share screen if you can. Niel Pal: Yes, yes. Gladly. Moses Parente: I also did want to point one thing back when you were talking about the the SAS popocalypse. I think that's a funny name. but I think enterprise software is largely just here to s to stay, largely also on how big of a scale it is. And we all need an enterprise. It's not going away anytime soon in my opinion. Niel Pal: Absolutely, yeah. I mean you think about like enterprise adoption, right? Salesforce is used as a CRM for all these major companies. Anthropic in fact, Anthropic is rent renting out Salesforce's last building. So, you know, it won't go away. It's just from a business standpoint, investors are right to be wary in that it will make your business model not super sustainable in the world of AI agents if you're not gonna go headless. Because agents, right, you don't need multiple seats, which is the traditional way of doing SaaS. Moses Parente: 'Cause also with w with Enterprise, it's I the way how I feel it, it's also like the the quote unquote good old reliable. If it's not broke broken, don't fix it. And I think like once companies get to such a certain s a certain size, they're just like, Okay, for these parts I really don't wanna have to innovate as much anymore. I just wanna have this keep on going. Especially if they're companies not necessarily super tech focused or their main focus is largely something else. Yes. You are a str Niel Pal: Yes. Moses Parente: let's read it. You are a struggling second tier competitor that is more unethical and lying just because you have no decency or sense of proper behavior and shows your desperation. Straight out lying about if Chris being fired, I thought would even would be below even you. Niel Pal: And you know what's funny? Vinod isn't an investor in factory. He's an investor in both. So it shows you like that, you know, successful sibling kind of relationship that they're may have seen right here. Moses Parente: every everyone's always beefing with each other. It i it it's like everyone always says like, they they they want to see you successful but they don't want you to see you mo more successful. That that's how that's always like enemies and rivals are made. Niel Pal: Yeah, yeah. Yeah. Did you by the way, have you been keeping up with the stock market? Moses Parente: I have not as of recently 'cause I got quite busy with some stuff for like work and all that other and other things. But but yes, what other stuff has been happening? Niel Pal: Yeah, I think in particular what was quite interesting, and look, I mean, if you're gonna invest in the semiconductor industry, you have to understand that it's gonna be like a boom and bust thing. You have to, you know, have some I guess balls of steel really, right? 'Cause it's gonna go up and it's gonna Moses Parente: Ha ha ha. Niel Pal: go down. And last week it went down pretty bad. Core Weave, which is a neo cloud provider. I mean look, I Core Weave does not have the best financials. They are running on a lot of borrowed money, but Their numbers came out, was not the sharpest, and it caused just a big drawdown in semiconductors as a whole. You saw the memory space go down, microns, sand disk, Nebus is another one too. And for me, like that was a great buying opportunity because I know that if you look at compute, right, we're extremely compute strained. And we are extremely memory strained as well, right? It's just an overreaction from the market. even with the memory. There was saying that China is releasing like another memory player, which you know maybe causes a drawdown in the markets. Beauty of the Trump administration though. is that they are very pro-US. They're not allowing you to use like Chinese hardware. Maybe Chinese software. You know, we're seeing that with the open weight models. But he's definitely pro-Chinese Moses Parente: Yeah. Niel Pal: hardware in terms of what we're using as their chips. So I'm very bullish still on Micron and I think the stock will continue to do well. Moses Parente: Speaking of that, you just reminded me about again of the anthropic IP IPO that's gonna be happening in November. I saw the SEC financials on Reddit and it the growth has been pretty ste steady until June, 'cause like well, largely like when Claude Cone Claude Code came out, that's what caused a lot of it to skyrocket. But then since June, things have kind of flatlined a little bit. Would love to get your take on the upcoming IPO that's going on. Niel Pal: Yeah, I haven't actually looked at the financials, to be candid with you. But I Moses Parente: Okay, that's fine. Niel Pal: will say that it makes sense, right? I mean, do you remember when Cloud Code was really picking up? I would say that Moses Parente: yeah, like during the summer you just see everyone using it on TikTok and deleting chat G BT and all that. Niel Pal: Hundred percent. Yeah. I mean I remember like even back in November of twenty five. I remember a Cloud Code had maybe just come out at this point. This point I was still looking at Varsell skills. And yeah, tried to figure out how to use a terminal. Was not the not the best of using that. I was a Cloud Cowork kind of guy. But figure that out and just like the amount of stuff you can build was fascinating. Like you were seeing people in New York. The most fun city in the world. Peop So many people were just taking entire weekends off to lock in, just build stuff, right? Token maxing was at an all-time high. Pretty much all these major corporations said, you know, we don't care. We are amazed by the capabilities of these models. We're just gonna keep building, right? So it makes sense. Now we're seeing more Moses Parente: Yeah. Niel Pal: of a cost-cutting culture within orcs, where it's not necessarily cost-cutting, right? I should probably rephrase that. It's more of a cost awareness. Or people realize that these open weight models can handle a significant portion of workloads. At the same time, you can't avoid these frontier models. These are cutting edge models that are so powerful. Like I don't know if you played around Fable. There's look, I think the moat for a lot of software companies is gone. You could use Fable to do a lot of stuff, right? And so these models are great. These companies are continuing to build and it'll be in part of many different workflows as well. But I do think that their financials are a little bit slowing down right now, which is why I see them pushing hard against these open weight models. Moses Parente: Yeah. I feel also that a lot of the other models have been getting a lot smarter actually. I know Opus has been getting a lot better, especially I mean we saw ChatGPT Niel Pal: Absolutely. Moses Parente: Astral about a month ago. And I think that now every because innovation's just happening at a very rapid pace, everyone else is starting to catch up. Niel Pal: Absolutely. Moses Parente: And that I mean fa I mean Fable is amazing. I I I love it. I played around with it a lot. I have absolutely great things to say about it. But I don't think that it was just going to be in delete like in a lead forever. I remember back when like you just see on TikTok and X and LinkedIn everyone can't stop talking about Claude. But now like everyone else is like peop you see people starting to go back onto Chat GBT, you see people going back onto even like checking out o other models. I know I know a lot of people have been a fan of Gemini. I think Gemini's pretty cool. But I think it's gonna be very interesting to see where all of these stuff go in about a year and see where it takes us. Niel Pal: Absolutely. Yeah, these models are definitely good. I think personal agents have been kind of thing that, you know, people have been talking about more. Alexander Wang came from Scale AI. He's now leading Muse at Meta, which it's kind of been taken off. I've always seen meta as like A player that has a lot of distribution that people are going to check Instagram a lot. I think increasingly you go on your Reels page, more and more people are being creators. So I'm excited to see where they can apply it in that space. Grockba too, Moses Parente: Yeah. Niel Pal: Instinct, I don't know if you tried that out as well. These are good personal agent platforms. They make your life a little bit easier. I don't know like how drastic of an impact it is gonna be, quite frankly, but Moses Parente: Yeah. Niel Pal: it is quite cool to see. Moses Parente: Muse is actually really fun. I've been using Muse recently for the past for the past week and I I think it's pretty fun. but I think there's a lot more that I have to be checking out with Muse. In terms of Grokbot, I haven't used Grok Grokbot yet. I I don't I've only used Grok just mainly a little bit, but I haven't really touched on it too much. But in terms of personalized AI agents, I think that's one thing that a lot of people should be taking more advantage of. Niel Pal: Absolutely. Yeah, I mean at least for myself, right, working in sales. I've seen go to market really go pro AI, whereas I think previously there was maybe like a bit of a pushback. And these models were kind of naive. Like they couldn't take on these workloads. Now they're able to connect to your granola, they can connect to your CRM, they can connect to everything and they really can take on more of these autonomous workloads. Moses Parente: Seeing a lo now this also does bring me back to this we're going back into a little bit of politics again, but in the event the the Democrats take back the House or get at least get in some sort of power, I wonder if they'll be more willing to be able to embrace artificial intelligence and innovation or just rather they're just gonna keep going on the populist messaging that they're they've been going on as of recently, literally because they've been out of power for the past four years at least in the house. Niel Pal: Yeah. I mean TBD, right? I think that eventually people are gonna have to come up to date with just, you know, what's going on. But at the same time, right, I remember having lunch with a colleague. This is probably the only city that actually really loves AI. I don't know if like in across the rest of the country people are this pro AI, right? Keep in mind it's mainly the older population that goes out and votes. So I think that, you know, we'll see more of the Democrat Party eventually adopt and start using AI. I'm sure it's already to some degree part of your workflow at, you know, whatever job you're doing. But it will be kind of a gradual pace. And I do think just, you know, the pushback against these massive companies that are really taking over the world, I think that's just against the Democrat ethos and it will be kind of a a slog in that in that manner. Moses Parente: I mean, everyone I mean, especially people who are, I guess, on the more progressive leading side, they like to say that, we hate billionaires unless it's someone like George Soros or Michael Bloomberg who's pumping money into their elections. Michael Bloomberg just announced that I mean, you you can see it on the news that's that he just said that he's going my bad. Michael Bloomberg just announced that he's going to be donating eighty million dollars to Democrats to make sure that they take the House. And Niel Pal: Mm. Moses Parente: I think that is a little bit ironic when I guess people on the progressive side they would say, we hate all billionaires unless they're funding our elections. And I think once like we see Democrats in power, we will see a more gradual acceptance towards artificial intelligence. Because AI, in my opinion, is here to stay. It's not gonna be going away any any time soon. I know a lot of people have been very bullish in terms of ant like deregulating things on the Republican side for the the purpose of either winning the AI war or making sure that they just build and innovate as much stuff as possible with as with as little roadblocks in the way and we'll be seeing a more gradual acceptance towards artificial intelligence by both parties. And by the world too. Cause I know everyone was Niel Pal: Yeah. Yeah. I agree, right. Moses Parente: everyone was largely afraid of the internet. When I mean, I'm I'm only twenty six, but so obvious obviously like the internet's old old older than I am, but as we've seen technology advance so much and it became a part of our day to day lives, we eventually just said like, okay, w this is pretty much normal. Like using the internet's gonna be very normal. And before we had like flip phones. Now we have Now we have iPhones and smartphones and it's all being integrated in our lives on how we live on a day to day basis. And with AI in the mix, everyone's gonna be using it. Maybe is it to the point that they're gonna be using personalized agents to help them in their day to day? I don't know. I think th I remember you were talk we were talking about in a podcast episode a while back that in America we don't actually use AI agents in our day to day day lives, and that's something that we could be we should be doing. But In a in a way, it's gonna be mo much more normalized and kind of integrated into society where laws will be a lot more lenient on it. Niel Pal: Absolutely. I roughly agree with that. It'll take time, right? You know, there was like you said, right, a fear with the internet. It'll become more and more involved in our workfl workflows, but I guess considering that we saw big tech and their rise, you know, over the course of our lives I think there will be a push from the Democrat Party, which I don't disagree with, and that holding big tech accountable, making sure that it continues to help sustain the people that are helping provide this training data, at least while I'm in San Francisco, I've been kind of amazed that buying training data is just so powerful here. Like these companies like After Query, Marcore, which is right next door to us, these companies are blowing up. They cannot get enough data. You know, these labs are so hungry for it, so People who are providing a resource here, I think, should honestly be recognized as an extra thing in society. And frankly, like I don't know, maybe like this is we're gonna see like a new job category emerge of just helping labs get training data right now. Which I don't know how yeah I was gonna Moses Parente: Spe no, sorry. Niel Pal: say like I don't know how like good I feel about that. Probably not the most enriching work. I've some work or projects. But it is such a growing space and will only continue to grow further. Moses Parente: The New York's city tech scene has also been growing a lot too. as you saw recently the Databricks' Dana Bricks research Niel Pal: Yep. Moses Parente: and development space is actually gonna be expanded in Manhattan. I'm actually looking pr I'm actually looking forward to that. I think people are sleeping a little bit on the New York City tech scene and that people should keep their eyes peeled out. Niel Pal: Absolutely. I actually am friends with Tasso who runs the D Databricks New York City office. So Moses Parente: nice. Niel Pal: yeah, yeah. They're they're a great company. Databricks is just world class. New York City, again, right, like these are this is the most fun city in the world. Google, if you go try to get into their New York City office, it's competitive. 'Cause everyone else wants to live in New York. Moses Parente: Yeah. Niel Pal: So to see them continue to double down on the tech, I'd say San Francisco is still probably in like Silicon Valley as a whole. probably still the best place to be if you want to get into tech. You want to be ahead of these trends. But New York City is a second place option in my eyes. And it's just a ton of fun. There's a lot of convergence of different ideas. And I think right now like the main barrier to building is really just your creativity and your ability to figure out how to creatively apply these tools. I spoke to some of the engineers on our team. No one has handwritten any code. Like in the past six months. They even their PRs, they're getting their agents to review that. And you know, like we're we're a legitimate company, Moses Parente: my gosh. Niel Pal: right? Like, you know, this is industry. So I think that's kinda where we're going and I'm excited to see what the future holds Moses Parente: Yeah. Niel Pal: of just, you know, different perspectives and you know, different ideas, how we can get that into software that works for us. Moses Parente: Hmm. Hopefully we can get your friend on the podcast here. Niel Pal: Yeah. Moses Parente: I think that would be pretty cool if we could both Niel Pal: Yeah. Moses Parente: interview him. I think that that would make for a great conversation. But that that's f that's obviously for another time. I think like that's pretty much the same thing for largely a lot of engineers. We haven't really been having to handwrite code in like forever. If if if you're a bad coder, now is the best time to be a ba basically an an an an an engineer in in all in all honesty. Because Niel Pal: Yeah. Moses Parente: when you're it the ability your abilities as an engineer in my opinion have largely been increased by tenfold because of these AI tools. And Niel Pal: Absolutely. Moses Parente: it's largely in my opinion going to result in a lot more innovation and you're going people are going to see like, because these AI tools are basically just making engineers output like tenfold or a hundredfold, we're going to be building things at a much more faster pace than ever before. Niel Pal: Yeah, no, I agree, man. And it is cool. Like I think, you know, one of these episodes we were talking about education. I believe building Moses Parente: Yes. Niel Pal: should be part of the core curriculum, right? one of these engineers absolutely, Moses Parente: Build baby build. Niel Pal: yeah. One of these engineers I was talking about, I was talking with, he was saying like I honestly would probably try to study business and have like a real lean into entrepreneurship. I personally disagree, I think computer science is probably still the best maybe thing to do if you're looking to become a software entrepreneur. But I do believe that we all will take on more of a product manager approach of the different agents we run. And at least within my company, evals have been huge for us. So one piece of advice I would have to say an undergrad who's looking to get into industry, learn how to do proper evals. It's not really particularly hard, and it's something that, you know, it's more of a product manager skill, but it'll be increasingly more important as continue to build out and these models increase their capabilities. Moses Parente: Anyone who's an undergrad, make sure to to write that down. Niel Pal: Yeah. Moses Parente: I I mean I I de I definitely like where your f your friend is actually where your friend is going. I think that he does have a a good point in what he's making. But I do have to say I do one hundred percent agree with you with in terms of that having a CS foundation is probably the best way because in my opinion, besides the study and grit that you you get from learning all those f from like learning in like your classes. I think the ability to be able to solve problems comes a long way. Even while you can have AI and all your AI agents basically just write your code and do all your PRs for you. I think that there is still a good skill to have to learn how to learn and to how to solve problems. Because in business when you're going to be a have you're basically just going to be solving problems all the time. Now, are they going to be like problems that you see in the textbook or lead code style or something like that? Not necessarily. But it gives you an analytical mind and it teaches you how to think. And I Niel Pal: Mm-hmm. Moses Parente: think that's something that can carry you a long way. And you can see it in many of the CEOs that are who are leading Fortune five hundred companies right now. Many of them don't even have an MBA. Now I'm not saying that having Niel Pal: Yeah. Moses Parente: an MBA is useless. I think it's great to make connections and meet new people, but I think but as you can see, a lot of these people largely have STEM degrees. The I th I believe the CEO of Microsoft he has a chemical engineering degree. And Niel Pal: Yeah. Moses Parente: like just just to name one example. And you see all these people who are basically becoming business lead leaders with an engineering background or some sort of STEM background. I think we're starting to see more of that translate over and I think it's not at the point where it's the the future anymore, it's the present. And we're Niel Pal: Absolutely, yeah. Moses Parente: that might be dictating away on how CEOs and business leaders in the future should be taking their education. I still think having a Niel Pal: For sure, right? Yeah. Moses Parente: I still think that an MBA is very much valuable in terms of I mean I I I mean I don't have an MBA, but in terms of g making connections, meeting new people and overall developing more industry knowledge from the connections that you make and the classes that you take. it's still there's still a very good asset to have of how to solve problems and how to think critically. Niel Pal: Definitely. Ray Dolio too. he actually mentioned this in his book, right? Moses Parente: Yeah. Niel Pal: You want to create a culture of systems thinking, but have the flexibility and I guess social wherewithal to understand how these can change. So I think that's like a good framework for people to have, right? Be able to think in systems, figure out how you can scale them and allow them to compound, but you do need to have these social skills and understanding of human intuition, right? Which will allow you to design careful systems. Moses Parente: Kind of like game theory in a way. Ca like understand what people want and what the the incentives that they have in order to gain from doing XYZ. I think being able to understand people is largely a very underrated skill that people kind of overlook. A lot of people think Niel Pal: Yeah. Moses Parente: like, my just my technical skills are just gonna get me far into life or but I think being able to understand people fundamentally will carry you far. Niel Pal: Definitely. yeah, generally like Moses Parente: I did see that picture I I did see that picture of you you meeting him back in New York. I thought that was pretty cool. Niel Pal: Yeah yeah yeah yeah yeah it was quite cool to see. I was a fan of his book, you know, even back a few years back, so it was quite cool to get to meet him in person. yeah, this is something John Collison, who I actually got to see at a browser-based conference as well. John Collison is the founder of Stripe. He's very bullish on double majors. So while you're in university, you have the chance to, you know, maybe toy around with different entrepreneurial ideas. I recommend people to explore, you know, thinking about say would, you know, like thinking about the human experience and understanding how you can scale that, right? Ultimately like if you look at the CEO's job, I'm amazed by how much sales there is there, right? And especially at the enterprise level, a lot of this sales stuff is gonna be human relationships. It's fairly cerebral. That requires a level of extra human touch that I don't think you necessarily get in a CS education. So you mentioned, right, Microsoft CEO. I agree, right, systems thinking is absolutely important. But honestly, that extra personal touch you get in MBA, I know he went to a booth MBA. That I do think is underrated and more people should consider. Moses Parente: I I really do like that that way of thinking. Having kind of like a both like a systems level way of thinking, but then there's also that human interaction with it. And you can either get and I mean there's a lot of that there's a lot of majors that people can probably take on that depending on like what college you go to, but I think kind of like pairing that with a minor or if your credits allow it and you want to and you might maybe take some extra classes or you may you want like a second degree or whatever. Have like a stem pi a stem based major pi a place with a humanities major, like CSN psychology, just to name an example. I think that Niel Pal: Yeah. Moses Parente: is very valuable and something that can take you a long way. Niel Pal: Absolutely. Moses Parente: Especially with like as you can as you see with let's say a a lot of people, because you a lot of people who do take these majors, they're able to learn something that's completely different than their C S degree. And it gives them like a ba basically a different layer of thinking. And and th STEM major doesn't necessarily have to be C S. I think a a lot of STEM majors largely kind of come into the form of problem solving. And when Niel Pal: Yeah. Moses Parente: you have a way of thinking that's both problem solving but also thinking of ways how humans interact and what their motives are, their incentives, why they do X, Y, Z. They in g in in a way game theory in in in a sense, it really teaches you a lot. Niel Pal: Definitely. And I mean kind of an unrelated thing right here, but I was gonna bring up the fact that Sorry. we're gonna have to edit this out. No, they're they're Moses Parente: No, no that's all good. Niel Pal: cleaning the office behind me. I just lost my train of thought. Moses Parente: shoot, no, it's all it's all fine, it's all fine. Niel Pal: Okay. what is I sing? Yeah, dude, I don't know. I'm sorry. we'll edit this out. Moses Parente: No, this it's all good. Well, it's all good, it's all good. But again, it's all good. Well, thank you, Neil, for always coming on the podcast. Niel Pal: Of course. Moses Parente: You are we're always happy to have you on and have a great day, guys. Thank you again. Niel Pal: Yeah. All right. See you guys.