E165: Vision Pro: use or lose? Meta vs Snap, SaaS recovery, AI investing, rolling real estate crisis

1. 0:00 – 2:08

   Bestie intros!

   1. JCJason Calacanis0:00

      All right. Friedberg is back. Welcome back to the All-In Podcast, episode 160 something. Your favorite podcast in the world, yada, yada, yada. With me again, the chairman dictator, Chamath Palihapitiya, the rain man. Yeah, definitely. David Sacks is here and back from his time in the Metaverse. We found him somewhere out in space in the solar system, in his Apple goggles, your favorite, sultan of science, David Friedberg is back from the Metaverse.

   2. CPChamath Palihapitiya0:29

      I missed you guys.

   3. JCJason Calacanis0:29

      Welcome home.

   4. CPChamath Palihapitiya0:31

      Thanks for having me.

   5. JCJason Calacanis0:32

      What, what did you discover when you went to your anus in Google Glass? Sorry, Apple Vision Glass.

   6. CPChamath Palihapitiya0:38

      Have you actually used the Apple Vision Pro, J Cal?

   7. JCJason Calacanis0:40

      I ordered them, I ordered them and I walked by the Apple store and I was gonna go in and try them, and there were so many lunatics in there, I was like, "Yeah, I'm not doing it," but I ordered them. You used, you actually used them? What do you think?

   8. CPChamath Palihapitiya0:50

      I ordered one online to be delivered and it was, like, delayed by a month. So I went down to the Apple store and picked one up.

   9. JCJason Calacanis0:57

      Okay.

   10. CPChamath Palihapitiya0:58

       And my kids cannot stop using it.

   11. DSDavid Sacks1:00

       Really? I went down to the Apple store, but got cleaned out by a thief that stole everything, so-

   12. CPChamath Palihapitiya1:05

       (laughs)

   13. JCJason Calacanis1:05

       Imagine the Oakland one. (laughs)

   14. DSDavid Sacks1:09

       Let your winners ride.

   15. JCJason Calacanis1:12

       Rain man, David Sacks.

   16. DSDavid Sacks1:16

       And I said, we open source it to the fans and they've just gone crazy with it.

   17. CPChamath Palihapitiya1:20

       Love you guys.

   18. JCJason Calacanis1:21

       Queen of quinoa. That was crazy.

   19. CPChamath Palihapitiya1:25

       That was crazy.

   20. JCJason Calacanis1:26

       We'll put the video in here. To the idiots who are robbing Apple stores, all the devices get bricked when you steal them and they all have GPS in them. Have you tried it, Chamath?

   21. DFDavid Friedberg1:34

       No, I was too busy working out, making love, and winning.

   22. JCJason Calacanis1:38

       Oh, okay. (laughs)

   23. CPChamath Palihapitiya1:38

       (laughs)

   24. JCJason Calacanis1:38

       Got it. So you were, you were making sweet love.

   25. CPChamath Palihapitiya1:42

       (laughs)

   26. DFDavid Friedberg1:42

       (laughs)

   27. JCJason Calacanis1:42

       You were watching your portfolio go up and, uh, you were just generally winning.

   28. CPChamath Palihapitiya1:46

       (laughs)

   29. JCJason Calacanis1:46

       Got it, got it.

   30. CPChamath Palihapitiya1:46

       (laughs)
2. 2:08 – 20:46

   Apple Vision Pro breakdown

   1. CPChamath Palihapitiya2:08

      computing platform. I remember when the iPad came out and everyone poo-pooed the iPad. I thought it was stupid. I tried to use it. I couldn't get any value out of it. And in 2010 or 2011... When did it come out? 2010, 2011, we started using it with our sales team selling to farmers.

   2. JCJason Calacanis2:22

      Hm.

   3. CPChamath Palihapitiya2:22

      And we gave every sales guy an iPad and they went out in the field with 3G and they were able to close sales in the field meeting with farmers, which had never been done before. You usually had to get a farmer to come into an office-

   4. JCJason Calacanis2:32

      How many iPads did they sell?

   5. CPChamath Palihapitiya2:33

      ... to sell them a product. No. So we, we had like, you know-

   6. JCJason Calacanis2:36

      No iPads. We were selling climate.com software. I get it, yeah.

   7. CPChamath Palihapitiya2:38

      We had do- we had dozens of these sales guys. We gave them out to our sales agents as well, the independent agents, they started using them and it was like a real game changer in how sales was done in agriculture, and I had never even contemplated that when I first used the iPad. It was like-

   8. JCJason Calacanis2:50

      All right. So let, let's, let's get to brass tacks here. What is the killer app? What do you think in the next five years people are gonna be doing with this thing on a daily basis? Is there a daily use case?

   9. CPChamath Palihapitiya3:00

      I'll say a couple things. One is, like, I feel the same way I did about the iPad, which is I don't know what it is today, but I can tell that there's something there. And I'll give you an example of something I thought about. First of all, the AR is game changing.

   10. JCJason Calacanis3:10

       Okay.

   11. CPChamath Palihapitiya3:10

       If you've used, like, the Meta-

   12. JCJason Calacanis3:11

       Yeah.

   13. CPChamath Palihapitiya3:12

       ... the Oculus Quest, it, like, makes me super dizzy, makes my head hurt, makes my eyes hurt. Like, you're super disoriented.

   14. JCJason Calacanis3:18

       Motion sickness.

   15. CPChamath Palihapitiya3:19

       What Apple solved is that you're, like, still in re- reality-

   16. JCJason Calacanis3:23

       Mm-hmm.

   17. CPChamath Palihapitiya3:23

       ... but then you get to interact with these three-dimensional kind of objects in reality, and it's, like, really well done. It's definitely V1 and there's gonna be incredible changes in the next couple generations, but it gets rid of all that dizziness, disconnected kind of stuff that happens with the, uh, the full VR experience, which I thought was really incredible. Then last week, and I'm sorry I missed this show, we have a facility with my company in North Carolina. We have this giant greenhouse facility, and I was doing meetings with farmers and stuff. I go to the greenhouse facility and there's so much work that the greenhouse techs and lab techs are doing where they're using an iPhone and a barcode scanner and a printer, and they're holding all these pieces of equipment, scanning the QR codes on flowers, taking the pollen out, putting it in the next flower, training each other how to do it. And I was like, I put this Apple Vision Pro on, and I was like, man, all the, the apps and all the tools that we had all these different pieces for that was taking people tons of time, image collection, data collection, could all just be done streamlined-

   18. JCJason Calacanis4:21

       Yeah.

   19. CPChamath Palihapitiya4:21

       ... while you're working. You could have a task list-

   20. JCJason Calacanis4:23

       It's like Minority Report.

   21. CPChamath Palihapitiya4:24

       Yeah.

   22. JCJason Calacanis4:24

       You just go a little into a headset.

   23. CPChamath Palihapitiya4:25

       Yeah, you have a task list on the right, cameras are taking images in the middle, QR codes are automatically scanned, data is being ingested. The task list is kind of, you know, giving folks next steps. They can listen to music while they're working. And I realized for that job, and I met with all the, the team out there and spent time with them, and I actually did the work that they do to get a better sense for the workflow. And I was like, man, literally every aspect of this job will be massively improved and productivity will go up by 10x with these goggles. Will it happen in the next couple weeks or months? I don't know, but my engineering team is looking into it. Can we take it, can we use some software? Can we build some software and can we put this on folks to give them a better work experience, increase our productivity, to do automated data capture? So I don't know exactly where it goes, but I could start to see how this can become a more ubiquitous part of a workforce setting and not just be a video game and movie tool for consumers. So I'm, I'm reasonably optimistic about where this goes. It's definitely V1. I feel like it's the iPad days where no one's really sure where the applications are, but yeah.

   24. JCJason Calacanis5:20

       Yeah, enterprise applications.

   25. CPChamath Palihapitiya5:22

       Enterprise tools, unbelievable.

   26. JCJason Calacanis5:23

       Yeah. Makes total sense. And also training, training, right?

   27. CPChamath Palihapitiya5:25

       Assembly line, workforce-

   28. JCJason Calacanis5:27

       Sure.

   29. CPChamath Palihapitiya5:27

       ... warehouse workers where you're getting real time kind of task updates, data is being ingested, all in real time. And, and by the way, the other thing I'll say is training is incredible. There's spatial video recording on it, so it looks like you're living through the experience that someone else had.

   30. JCJason Calacanis5:41

       Mm-hmm.
3. 20:46 – 32:53

   Meta vs Snap: god-king CEO, dependent on ad revenue, drastically different performance

   1. JCJason Calacanis20:46

      I wanted to talk a little bit about what apparently is gonna be the- the spread trade (laughs) of the last year. Meta is, continued their unbelievable run, and Snap dropped, like, 30%. Here's a chart for y'all of Snap versus Meta. You can take a quick look at it here. And f- just for context, both companies did great during COVID and ZIRP, hit all-time highs in 2021, but they both got crushed due to the ad spend pullback, obviously. But then Meta started to get less focused on their headsets and more focused on AI, started doing the reduction in headcount 22% year over year, from 86,000 to 67,000 in the last quarter for Meta, and their quarterly profits have increased to an all-time high of $14 billion. That's profits, folks, in Q4 for Meta. All-time high for the stock price, $470 a share, $1.2 trillion market cap. Snapped down 60% from its closing price on its IPO day in 2017. Let me just jump to Chamath before I get into more charts and everything. You pointed out, Chamath, and maybe you could explain to the audience just how ridiculous the voting rights were and the massive, uh, dependence that the Snap team and the executives had on stock-based comp. Two issues for you, Chamath.

   2. CPChamath Palihapitiya22:07

      Well, I mean, I think I said it before. I think that case studies have been written about how tilted the governance is in Snap. I think the point is that they basically have infinite to zero voting power over common shareholders, so there's no real feedback loop, and I think that that has probably adversely affected the types of people that traffic in their stock. Now, look, activists and short sellers sometimes have a very bad reputation, but if you steel man their side of it, what they are there to do is to shine a light on inefficiency, and in the short seller case, sometimes impropriety, but it should all lead to companies being better run, right? I think Meta had this example where they had a really big hiccup and everybody, including us, sort of pointed out the levels of spend that they were making really didn't make any sense. I think we had a chart that compared the level of spend of Meta second only to, like, the spaceship program, right? Just, like-

   3. JCJason Calacanis23:15

      Bonkers.

   4. CPChamath Palihapitiya23:16

      ... an enormous amount of money. And look, Mark got the message. He heard it loud and clear. I think he got fed up with whatever was going on there and he fixed it, and it's in the numbers. Now, I don't know Snap because, to be honest with you, I've never taken more than one second to look at that company. And the reason is, there is just zero ability for me to have any useful say, so I've never honestly looked at its performance, I've never studied a single characteristic, I've never trended it, and I think the point is that I am probably where a lot of other reasonably smart folks who could give a reasoned opinion on how to make it better land, and part of the reason is because there is no feedback loop that matters.

   5. JCJason Calacanis24:02

      Yeah.

   6. CPChamath Palihapitiya24:02

      And when you know that, why would you waste your time? At least in the Meta-

   7. JCJason Calacanis24:05

      There are other options, right?

   8. CPChamath Palihapitiya24:06

      There are other options, and- and Meta was another one. You know, you can write a letter, it gets picked up on CNBC and Bloomberg and whatever, and all of a sudden, they kind of pay attention. And I think... And you look at Disney, Nelson Peltz goes and gets Eich Perlmutter shares, buys some more, takes a large position.

   9. JCJason Calacanis24:23

      Thank you, Nelson Peltz. Yeah.

   10. CPChamath Palihapitiya24:24

       We'll see whether that fixes itself. The point is that, in all of these other cases, people are investing the time because...

   11. ... they think that there's even a small shred of a chance that the company listens. But if you literally have no say, you couldn't even do a proxy, you couldn't vote the shares, why would you bother? And I think-

   12. JCJason Calacanis24:44

       Should they-

   13. CPChamath Palihapitiya24:44

       ... that, that's more of an example where maybe there is a... I, so I don't even know why Snap did poorly. And again, I'm not gonna really take the time because it's like, why bother taking the time?

   14. JCJason Calacanis24:54

       Sack, should, should they unwind this, like, no voting common shares, super voting shares nonsense and, and should this go away as a concept in the stock market?

   15. DFDavid Friedberg25:04

       Well, I mean, Facebook or Meta has a pretty similar concept. I mean, I guess Zuckerberg has 60% voting control, whereas Evan Spiegel has 99%, so Snap is more egregious. The difference is that Zuckerberg is listening and Spiegel is not. The, the reason why Snap is doing poorly is not because its revenue has deteriorated. So I looked up... Let me just put it this way, I asked ChatGPT for their key metrics. (laughs) So assuming GPT is not hallucinating, if you compare 2021 to 2023, their total revenue went up from 4.1 to 4.5 billion, and gross profit went from, call it 2.4 to 2.5 billion. So not a huge increase, but revenue and gross profit were slightly up. But if you look at operating expenses, they went from 3 billion to 4 billion a year, and that is why their operating income or operating loss went from a $700 million loss to $1.4 billion loss in two years.

   16. JCJason Calacanis26:10

       Hm.

   17. DFDavid Friedberg26:10

       So that's the source of the problem, is that they increased their operating expense by a billion dollars a year from 2021 to 2023.

   18. JCJason Calacanis26:21

       Yeah.

   19. DFDavid Friedberg26:21

       It's pretty simple.

   20. JCJason Calacanis26:22

       So do they, they seem like they're the last ones to get the memo.

   21. DFDavid Friedberg26:24

       Yeah, they, they were the last ones to get the memo, and just, just to finish the point, so you saw that a few days ahead of this quarterly announcement where their stock got crushed, they put out a press release saying they're gonna cut their headcount 10%.

   22. JCJason Calacanis26:37

       Ahead of the-

   23. DFDavid Friedberg26:37

       It's too little-

   24. JCJason Calacanis26:38

       Yeah.

   25. DFDavid Friedberg26:38

       It's too little, too late. Yeah, they knew, right? They knew they-

   26. JCJason Calacanis26:41

       They knew. (laughs)

   27. DFDavid Friedberg26:41

       ... had a problem, so they released the, the press release saying, "Oh, we're gonna cut." Well, you should have done what Zuckerberg did. You know, Zuckerberg did a 20% cut last year. He got serious, he got lean and fit. And instead, these guys held out, did nothing, then when they know that the market's gonna crush them-

   28. JCJason Calacanis27:00

       Yeah.

   29. DFDavid Friedberg27:00

       ... they put out this lame announcement, 10%. No, not 10%. Really if you just wanna get back to where you were two years ago in terms of operating expense, you need a 25% reduction.

   30. JCJason Calacanis27:13

       Yeah.
4. 32:53 – 41:06

   Positive signals indicating a big SaaS bounceback

   1. CPChamath Palihapitiya32:53

      point, yeah.

   2. JCJason Calacanis32:54

      Sax, you tweeted that you're seeing a little SaaS bounce back all of a sudden. That's interesting. I am seeing something similar. Last year, last two years, you had a ton of people cutting their SaaS spend, maybe removing the number of SaaS vendors they had, consolidating vendors. Uh, you tweeted many public and private software companies are experiencing accelerating growth after six to seven quarters of deceleration. SaaS recession appears to be over according to the SaaS master, David Sax. You wanna unpack this for us? What are you seeing?

   3. DFDavid Friedberg33:24

      Well, it's still pretty early because not everyone's reported. But if you looked at the big tech cloud performance in Q4, you could see that there's a bounce back in here. This is net new ARR added for AWS, Azure, and Google Cloud. So you see here in Q4-

   4. JCJason Calacanis33:43

      Mm.

   5. DFDavid Friedberg33:43

      ... that there's a huge increase in net new ARR for the- the big cloud computing platforms. And then I think another bellwether is Atlassian. So we're still waiting to hear from HubSpot, Salesforce, Zoom, Adobe, companies like that. They haven't reported yet. But if you look at Atlassian-

   6. JCJason Calacanis33:59

      Atlassian makes Jira, amongst other products. They're based in Australia. Yeah, major-

   7. DFDavid Friedberg34:03

      Yeah, exactly.

   8. JCJason Calacanis34:04

      ... co-collection of SaaS companies, right? It's a collection of SaaS products.

   9. DFDavid Friedberg34:08

      Yeah, so net new ARR would be the amount of growth in that quarter. And this is on a year-over-year basis. So you can kinda see Q4 of '21 was the absolute peak and then plummeted, and then it actually went negative for about a year.

   10. JCJason Calacanis34:26

       That's- that's tough to be in a company with net new ARR going negative, yeah.

   11. DFDavid Friedberg34:30

       Yeah, that doesn't mean, by the way, the company's shrinking. It just means that the amount of net new ARR, which is the amount of growth, is actually smaller than that same quarter a year before.

   12. JCJason Calacanis34:43

       Yeah.

   13. DFDavid Friedberg34:44

       And then in Q4, you could see there's some acceleration here, that they're starting to add more, they added more net new ARR, I guess 33% more in Q4 than they did over the previous year.

   14. JCJason Calacanis34:55

       And part of that, Sax, is because the comps are lower, and they kinda bottomed out, yeah.

   15. DFDavid Friedberg35:00

       They bottomed out and now they're re-accelerating.

   16. JCJason Calacanis35:02

       Yeah, that's great.

   17. DFDavid Friedberg35:02

       So we're, you know, we're starting to see this in some of my board meetings as well, where in 2022, everybody was missing their numbers and reforecasting down, and then they would miss the reforecast.

   18. JCJason Calacanis35:13

       Yeah.

   19. DFDavid Friedberg35:14

       So by 2023, the forecasts were very, very conservative. And I'd say, and now I'm seeing companies beat their- the sort of the lower forecast in Q4. This wasn't happening earlier in the year, but finally, I think people are starting to beat their, sort of their lower forecast for Q4.

   20. CPChamath Palihapitiya35:33

       That's the question that I was curious about. What do you- what do you actually think is happening? Is that we've re-baselined these businesses, so now what would have looked like just a massive miss over the last two years now looks like a beat because we've just completely reset expectations? Is it that? Or is it that the economy is actually expanding and we can count on some reasonable growth rates? Is it a combo of the two? What do you think it actually is?

   21. DFDavid Friedberg36:01

       Yeah, I mean, it's definitely a new baseline in the sense that, I mean, if you go back to 2020 or 2021, we considered good growth to be, you know, 2 to 3X year over year. And now if it's going from 60 to 80% growth year over year, you're happy. So there's definitely been a- a lowering of expectations. That being said, you still seeing these numbers, there has been a bottoming out, and we're starting to now grow from this new baseline. So for example, I think with Atlassian here, we are seeing an increase in spend basically and- and growth, right? So the way a recession is typically defined is, uh, two quarters of negative growth, right? We had six to seven quarters...... of decelerating or negative growth.

   22. JCJason Calacanis36:48

       In SaaS, in tech, yeah.

   23. DFDavid Friedberg36:49

       In SaaS-

   24. JCJason Calacanis36:50

       In SaaS.

   25. DFDavid Friedberg36:50

       ... which is why I called it the SaaS Depression or, or B, the... Yeah, it was actually kind of a depression. You're right. But now we're seeing quarter over quarter growth. So growth is re-accelerating. Growth is higher than it was. So is it gonna get to where it was? That probably will take some time, but it feels like the problems in the ecosystem worked themselves out and now we're back to growth again.

   26. JCJason Calacanis37:12

       Yeah, I can add psychologically, uh, because I'm on a couple of SaaS boards as well, and psychologically it felt like, you tell me if I'm right, Saas, Saks, if you saw the same thing, there were two years of calling up customers and they were like, "We're, we're consolidating vendors. And by the way, we did a RIF, and so we need 20% less seats. So we're gonna have 20% less SaaS companies that we're buying from, and we're gonna have 20% less seats." So you, you start putting that all together. Man, everybody was just in psychological triage mode. "We cannot spend money. I don't want to lose my job." So you're, if you're a procurement person, you're the CTO, you don't want to lose your job. You don't want to have more cuts. So you're like, "Well, I can cut some software costs. Do I get points for that?" And the points you would score for the last two years was cutting costs. With the market ripping and y- you now got a really, you know, efficient company, you're like, "Hey, can we spend a little bit on SaaS to make the remaining employees even more, you know, productive?" Okay, maybe that's a reasonable discussion. And then people are playing ball in terms of negotiating prices. So that's the other thing I see is, like, people are like, "Well, we'll take your software, but here's what we want to pay." And then they're coming to the board and saying, "Can we do this deal?" Would have been a million dollar deal, but it's a $200,000. Like, yeah, take the money, take the money. Let's, let's bear hug that customer.

   27. DFDavid Friedberg38:25

       The market is generally an escalator on the way up and an elevator on the way down. So the recovery is going to take a long time, but at least we've bottomed out and we're in recovery as opposed to continuing declines.

   28. JCJason Calacanis38:37

       Yeah.

   29. DFDavid Friedberg38:38

       By the same token, if you're a startup and you're not seeing improvement in your Q4 sales, then you no longer have a macro excuse for why you're not doing well.

   30. JCJason Calacanis38:48

       Interesting. And then Freeberg, you added, you know, y- you were like, "I'll, I'll make my own software." You said, uh, you know, some SaaS war- software is too expensive. I'll put a developer on it. And so how's that working out for you? Are you still in that mindset of like, yeah, maybe we should build our own software?
5. 41:06 – 1:12:16

   VCs are split into three camps on how to approach AI investing

   1. JCJason Calacanis41:06

      All right, let's talk a little bit about VCs and how they're investing in AI. There, there seems to be three camps shaping up here, Chamath. You know, one group is like, "Uh, the incumbents are going to win." You know, Microsoft, Google, Amazon, everybody, they're going to win the day, so they're going to wait and see. Then there's another group who's sitting it out because they're like, "Hey, open source is going to win." Meta's committed to open source and collaborative platforms. I've been playing with, uh, Hugging Face with Sandeep, as well as you, Chamath, and it's pretty amazing what's happening over there. And then a bunch are obviously placing bets right now. The valuations are absurd. Founders Fund and Andreessen Horowitz, two notable firms, are approaching it differently. Founders Fund bought into OpenAI at a $29 billion valuation. But aside from that investment, they're generally avoiding the AI deals. On the other hand, Andreessen is betting heavily. Character.AI, Replit, ElevenLabs. Mistral, you're also in Replit, Saks. So what do you think? Is open source going to win the day? You've been picks and shovels the whole way. You've been talking about compression. Maybe this isn't actually a good market. What, what's your thinking as a capital allocator, Chamath?

   2. CPChamath Palihapitiya42:15

      I think foundational models will have no economic value. I think that they will be an incredibly powerful part of the substrate, and they will be broadly available and entirely free.

   3. JCJason Calacanis42:27

      Wow.

   4. CPChamath Palihapitiya42:28

      So if you think about that, any closed model, especially a closed model that operates on o- on the open internet, is not very valuable. And any open source model that operates on the o- that trains on the open internet will make that so. So in that world, things like Mistral and LLaMA...... will essentially decay the market to zero.

   5. Hmm.

   6. So if you, if you're looking at any economic value that has been captured up until today, if it has been captured by having a proprietary closed model trained on open data, that economic value will go away. And I think Google and Microsoft and Facebook and Amazon and a-all these startups have a deep economic incentive actually to make that so. So now you can evaluate what that means. So if you get an open model from Hugging Face that's just kick ass, where do you spend money? Well, you're gonna have to spend money to actually train it, to fine-tune it, maybe to have some pretty zippy inference, and all of that means that there's a new kind of substrate that has to be built, which is all around the way that the tokens-per-second are provisioned to the apps that sit on top of the model. What that means is you need to go back to 2006 and '07 and say, "Okay, when we first created the cloud, who made money?" And fast-forward 18 years later, it's the same people that are still making money. So the people that made money in 2006 and '07 were Amazon, principally, because of EC2 and S3. The perfect analogy of EC2 and S3 in 2024 is the tokens-per-second provider. Now, there you have to double-click and say, "Okay, well, what does a tokens-per-second provider need to do to make a lot of money?" And I think the ultimate answer is you need your own proprietary hardware. So who is in a position to do that? Amazon has announced that they have an inference and training solution. For training, Cerebras has announced a pretty compelling solution. Google obviously has TPU. Then there's a handful of startups, including one that I helped get off the ground in 2016 that I funded called Grok. All of those companies are in a position to build a tokens-per-second service. Then you have companies like Together.ai, which basically just go and take venture money and wrap NVIDIA GPUs. And you can debate what the advantage will be there. One could say, "Well, it's not really a huge advantage over time." So my refined thoughts today are sort of what my initial guess was when we started talking about AI a year ago, which is the picks-and-shovels providers can make a ton of money and the people that own proprietary data can make a ton of money, but I think open source models will basically crush the value of models to zero, economically. Even though the utility will go to infinity, the economic value will go to zero.

   7. Did any of you guys see Chamath's interview with Jonathan Ross?

   8. JCJason Calacanis45:38

      Nope, not yet.

   9. CPChamath Palihapitiya45:38

      He, you put it out, right, Chamath? You made it public?

   10. You know, I did it just for my subscribers, but Jonathan is the founder and CEO of Grok, the company that I just mentioned. And the quick version of that story is I was li- I would, I would pore over the Google earnings results in the mid-teens of 2000 'cause I was pretty actively investing in a bunch of different public equities, and Sundar said in a press release, he mentioned that they had rolled their own silicon for machine learning called TPU. And I was like, "What is going on that Google thinks that they can actually roll their own silicon? What must they know that the rest of us don't know?" And so it took me about six or nine months, but through Sunny, I got introduced to Jonathan, and then we were able to get Jonathan to leave Google and he started, and he- Jonathan was the founder of TPU at Google, and then he started Grok, which I was able to lead that funding round in, in 2016, so eight years ago. Anyways, I did a, a Spaces with Jonathan talking about the entire AI landscape and AI acceleration to my subscribers, but it was so good. I gotta say, he is, he was so impressive that we kind of, like, figured out a way to just play the Space and tape it and then we published it to everybody. So it's on, it's on my Twitter for anybody that wants to listen to it.

   11. Oh, okay, great.

   12. It is-

   13. Awesome.

   14. ... amazing. He is really impressive.

   15. I was sitting on the 17 going to Santa Cruz, not moving for hour and a half, and I listened to it, so that I- kept me alive.

   16. (laughs)

   17. But, uh, I thought it was really great.

   18. What, what did you think? He's great, no?

   19. He's great.

   20. JCJason Calacanis47:18

       Yeah.

   21. CPChamath Palihapitiya47:18

       He had some great insights, and I think he's very compelling in arguing why some of the big cloud providers today that are offering infrastructure for AI model training and inference are gonna be challenged if someone can build full stack and be su- and do it successfully. So it was a really good interview. I actually think it's really worth listening to. But I enjoyed it. Yeah, thanks for putting it out there. I was, like, literally just sitting at, I was sitting in the car browsing Twitter and I saw your thing and I clicked on it and then I just ended up listening- (laughs)

   22. Well, yeah. It was a little-

   23. ... to the whole thing. (laughs)

   24. It's a, it's a little hard actually when you do a Space for your subs, you can't actually just flip a switch and then release it to all of your followers. So we actually had to, like, literally play it and then just capture the audio out and then republish it. But anyways, despite that inconvenience, if anybody's interested in learning about AI hardware, he is very compelling and he's very educational.

   25. JCJason Calacanis48:16

       So, Sax, your thoughts on just how you're approaching investing in AI, if you're specifically investing in the underpinnings of AI, picks and shovels, yada yada, or if you're just looking on the application level and it's, you know, you know, that kind of approach.

   26. DFDavid Friedberg48:30

       Well, we, we divided it, the space, into three categories. Uh, one is the, the models themselves, the foundation models, which can be either open source or closed source. There's infrastructure, so like Chamath was saying, it could be, like, model training, it could be vector databases, tools that developers use to create the AI stack, typically inside their enterprise. And then the third would be applications, which can be things like copilots or it could be a pre...... AI app that's using AI to kinda turbocharge its capabilities.

   27. JCJason Calacanis49:05

       Yeah.

   28. DFDavid Friedberg49:05

       Most SaaS would be in the application bucket, and so that's principally where we're focused, although we do look at infrastructure plays and models. However, I do think there is an argument for... I mean, really, with the question of commoditization, well, like all the model companies just get totally commoditized. Really, we're talking about OpenAI, right, because they're the leader. So the question is can they maintain their lead? I do think there is an argument that OpenAI will stay in the lead and, and actually do quite well, and I think there's a few points there. One is that if you're a consumer, you just wanna use the best GPT.

   29. JCJason Calacanis49:45

       You wanna use Google.

   30. DFDavid Friedberg49:46

       Exactly.
6. 1:12:16 – 1:16:30

   Rolling real estate crisis continues

   1. CPChamath Palihapitiya1:12:16

   2. JCJason Calacanis1:12:16

      So, there's been a lot of discussion in real estate. You, you shared a video with us. Why don't you kick it off for us here, Friedberg, what's going on in commercial real estate. And Sax, you've got holdings in a lot of those as well, so let's kick up the commercial real estate challenges of the moment.

   3. CPChamath Palihapitiya1:12:28

      Well, I mean, I think we're teeing off of Barry's comments at this event last week. He and I met backstage because I spoke right before him and then he gave this talk which is available on YouTube where he talked about the state of the commercial real estate market, and particularly he talked about the office market. Just to take a step back to talk about the scale of commercial real estate as an asset class in the US, Nick, if you'll pull up this chart, the total estimated market value of commercial real estate in the US across different categories is about $20 trillion with about $3 trillion being in the office market, which is specifically what he was talking about, and he was saying that in the US we're seeing people not coming back to work and all these offices are empty and we've talked a lot about these offices being written down, so how significant of a problem is this? So-... $20 trillion asset class. Obviously, the multifamily market is probably not as bad as office and retail, which are the most heavily affected, each of which are about $3 trillion apiece. The rest of these categories seem relatively unscathed in comparison. Industrial, hospitality, healthcare, you know, those, those real estate sectors are probably pretty strong. Data centers obviously growing like crazy. Self-storage is a great market. If you pull up the next image. So, it turns out that of the $20 trillion of market value, there's about $6 trillion of debt. So you can kind of think about that 20 trillion being 6 trillion owned by the debt holders and 14 trillion by the equity holders. And the debt is owned roughly 50% by banks and thrifts, and this was this concern that we've been talking about with higher rates, is the debt on office actually gonna be able to pay the debt on retail going to be able to pay when half of that debt is held by banks and thrifts, that as we've talked about, have such a close ratio to deposits that you could actually see many banks become technically insolvent if the debt starts to default? Barry's point that he made was if you look at the office market, which, you know, is marked on everyone's books as $3 trillion of market value, he thinks it's probably worth closer to 1.8 trillion. So there's $1.2 trillion of loss in the office category. And if you assume 40% of that 3 trillion is held as debt, you're talking about $1.2 trillion of office debt. A reduction from 3 trillion to 1.8 trillion means that the equity value has gone down from 1.8 trillion to 600 billion. So they've lost... Equity holders in office real estate have probably lost two thirds of their value, two thirds of their investment. And who owns all of that? Most of that, 60 plus percent, call it two thirds of that, is likely owned by private equity funds and other institutions where the end beneficiary is actually pension funds and retirement funds. And so if two thirds of the value has to be written off in these books, and it hasn't happened yet, what's gonna happen to all these retirement funds? And this is where going back to my speculation a couple months ago (laughs) kind of gets revisited. If you're actually talking about a two third write down on the value in these funds, most of that being pension funds, you're not gonna see governments let that happen. You're gonna see the federal government-

   4. JCJason Calacanis1:15:47

      Yeah, see Alan Belan, right? Yeah.

   5. CPChamath Palihapitiya1:15:47

      It's gon- there's gonna be some action at some point, and it's unlikely the office market is gonna ma- suddenly rebound overnight. If this stays the way it is, who's gonna fill that hole for retirees and pensioners? Because we're not gonna let that all get written down. Someone is gonna step in and say, "We've got to do something about this." And there's gonna need to be some sort of structured solution to support retirees and pensioners because that's ultimately who ends up holding the bag in this massive write down. He didn't go all the way there in his statements. He was talking more about his estimate of 3 trillion to 1.8 trillion, and then I tried to connect the dots and what that actually means. And ultimately, there's gonna be some pain felt by retirement funds that's gonna need to be dealt with somehow. So, uh, Sax, I don't know if that r- if that sits right with you?

Episode duration: 1:28:20