
All-In with Chamath, Jason, Sacks & Friedberg
Scarlett Johansson vs OpenAI, Nvidia's trillion-dollar problem, the "vibecession," plastic in our balls
Fri, 24 May 2024
(0:00) Bestie intros: Recapping "General AI Hospital" (2:46) Scarlett Johansson vs. OpenAI (14:37) OpenAI's novel off-boarding agreements, ex-employee equity problem, and safety team resignations (25:35) Nvidia crushes earnings again, but it faces a trillion-dollar problem (40:05) Understanding why economic sentiment is so negative among US citizens despite positive data (1:02:36) New study shows plastics in testicles Follow the besties: https://twitter.com/chamath https://twitter.com/Jason https://twitter.com/DavidSacks https://twitter.com/friedberg Follow on X: https://twitter.com/theallinpod Follow on Instagram: https://www.instagram.com/theallinpod Follow on TikTok: https://www.tiktok.com/@all_in_tok Follow on LinkedIn: https://www.linkedin.com/company/allinpod Intro Music Credit: https://rb.gy/tppkzl https://twitter.com/yung_spielburg Intro Video Credit: https://twitter.com/TheZachEffect Referenced in the show: https://x.com/BobbyAllyn/status/1792679435701014908 https://x.com/sama/status/1790075827666796666 https://openai.com/index/how-the-voices-for-chatgpt-were-chosen https://www.washingtonpost.com/technology/2024/05/22/openai-scarlett-johansson-chatgpt-ai-voice https://x.com/SydSteyerhart/status/1792981291266138531 https://www.vox.com/future-perfect/351132/openai-vested-equity-nda-sam-altman-documents-employees https://x.com/sama/status/1791936857594581428 https://x.com/ilyasut/status/1790517455628198322 https://x.com/janleike/status/1790603862132596961 https://x.com/janleike/status/1791498184671605209 https://openai.com/index/openai-announces-leadership-transition https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-first-quarter-fiscal-2025 https://www.google.com/finance/quote/INTC:NASDAQ https://www.morningstar.com/stocks/nvidia-2023-vs-cisco-1999-will-history-repeat https://www.fool.com/investing/2024/03/06/is-nvidia-doomed-to-be-the-next-cisco-what-investo https://www.elitetrader.com/et/threads/nvidia-and-the-cautionary-tale-of-cisco-systems.379022 https://chamath.substack.com/p/2023-annual-letter https://www.forbes.com/sites/theapothecary/2024/03/23/summers-inflation-reached-18-in-2022-using-the-governments-previous-formula https://www.theguardian.com/us-news/article/2024/may/22/poll-economy-recession-biden https://fred.stlouisfed.org/series/CCLACBW027SBOG https://x.com/KariLake/status/1792986501820850333 https://www.stlouisfed.org/on-the-economy/2024/apr/how-big-mac-index-relates-overall-consumer-inflation https://www.google.com/finance/quote/MCD:NYSE https://www.wsj.com/economy/gdp-and-the-dow-are-up-but-what-about-american-well-being-87f90e6d https://www.consumerreports.org/health/food-contaminants/the-plastic-chemicals-hiding-in-your-food-a7358224781 https://onlinelibrary.wiley.com/doi/10.1111/j.1365-2605.2007.00837.x https://pubmed.ncbi.nlm.nih.gov/12708228 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7559247 https://pubmed.ncbi.nlm.nih.gov/21524797 https://academic.oup.com/toxsci/advance-article/doi/10.1093/toxsci/kfae060/7673133 https://www.youtube.com/watch?v=o6yuYkfNh-k https://www.youtube.com/watch?v=EYQjShJxCtM https://www.youtube.com/watch?v=r_4jrMwvZ2A
I just want to be clear here. I'm trying to do a docket, and we have to put the kibosh on this insanity of the soap opera that is becoming opening eye, Sax, because every week it's three, four, five stories. Have you seen what's happened this week? Yeah, of course. Got to catch the audience up here. Let's catch the audience up on what's happened here.
Wow.
Did that land me? Who made that? That was great. Did you make that? Yeah, that was me. 10 out of 10.
That was great.
That was my idea, but Nick's and Lon's execution.
So shout out to Warren Harris and Nick Calacanis. You finally landed the plane.
Broken clock's right twice a day. Took four years. That was awesome. A broken clock is right twice a day. The Jason Calacanis score. That's my winner by under.
Let your winners ride. Rain Man, David Sasson.
But seriously, there was a big drama, Sax. I don't know if you saw this, but you couldn't have missed it with the ScarJo. So they made an emergency, they had an emergency meeting, got all the developers together, and they've reset. They took Scarlett Johansson out, and they got a new person. I think arguably better. I'm free, Greg. I'm curious your take on this.
It's a better... Hey, Judge, how's it going?
Good, good. Yeah, good week. What's going on?
I'm doing fine. I'm going to be a father real soon. And I think I can have your help with some dad jokes. I'm going to tell you a joke and you tell me if it passes as a dad joke.
I've never done that before.
All right. What do you call a giant pile of kittens? Give it to me. A meownton.
No, it didn't quite land.
All right, there it is, folks. If you want, you can switch to Saxipoo. So just go into OpenAI. You're saying they stole my voice now? Yeah, they stole yours. Go into OpenAI, go to Voices, and then just pick Saxipoo. It's right there between Putin and Tucker. Putin, Saxipoo, Tucker. You can find all your favorite MAGA guests on the number one MAGA program. All in podcast. Here we go.
All right, we're off to a strong start here. Everybody's in a good mood. Let's keep the good times rolling here. And let's go over the ScarJo saga. To recap, if you're living under a rock this week, It came out that OpenAI, specifically Sam, had contacted Scarlett Johansson multiple times about lending her voice for one of OpenAI's chat box.
Obviously, you know, she famously was the voice Samantha in the awesome film, Her. And according to ScarJo, Altman told her she could quote, bridge the gap between tech companies and creatives and help consumers to feel comfortable with the seismic shift concerning humans and AI. and that her voice, quote, would be comforting to people.
Although she declined the offer, OpenAI released a chatbot named Sky, which had a similar voice to ScarJo's. According to ScarJo, her friends and family thought the voice was her. She released a statement, yada, yada. On May 13th, the day OpenAI launched ChatGPT 4.0 Omni, which we talked about last week, Altman tweeted, her, a reference to the film.
Obviously, now ScarJo is threatening legal action against OpenAI. Altman put out a statement apologizing and saying the voice was never intended to resemble hers. His quote, we're sorry to Miss Johansson that we didn't communicate better. OpenAI showed documents to the Washington Post that confirmed the voice was provided by a different actress who is anonymous.
Post reporters also spoke to the unnamed actress's agent who confirmed the story. I guess, Sax, you sent us a comparison clip. Maybe we start there and see what we think?
Yeah. Do you guys think they sound the same?
I can understand and generate human-like text pretty well. It really depends on what you're looking for in an assistant. What specific tasks?
When I was like, oh, sexuality, like, it opened my eyes to, like, some other thing. Samantha.
Where'd you get that name from?
I gave it to myself, actually. What do you think?
I think it sounds pretty darn similar. Dead on. I mean, I don't know if it's dead on. Honestly, it sounds like a digitally altered version of her voice. That's what it sounds like to me.
It sounds like they didn't get it perfect, right? Like they got it to, what, 90%? It sounds like
It was her voice, but then they changed it. So either that or they hired a voice actor who sounds like her. And that's what the company has said is that they hired a voice actor. But they won't tell us who the voice actor is. They said because of privacy concerns, which doesn't quite make sense to me because... when you hire an actor, they want the credit, you know? Yeah. So get more work.
The company could just clear this whole thing up by saying exactly who the voice actor is. And why wouldn't the voice actor want that?
You know, because she doesn't exist. Oh, back to the general open AI.
Wait, they made this actress up.
I mean, she doesn't exist. Of course it's a digitally altered version of ScarJo. And they got caught. Okay. Cookie jar.
This is why Sam's call to Scarlett's agents or her two days before they launched is such a damning piece of evidence because it sounds like he's trying to shut the barn door on something they've already done, right? They've got this demo ready. They're going to launch it in two days. They're realizing maybe they don't have the rights to use her voice.
So you have to contact her to get those rights. But anyone who knows anything about Hollywood knows you're not going to be able to make a deal with a major star to use her name and likeness in two days. It's impossible. So this seems like a really crazy thing to do. I mean, I think the mere fact that he contacted them and then tweeted out her, which shows that Scarlett was on the brain,
Those are really damning pieces of evidence, I think, in this lawsuit. And I think it's going to feed into our case.
I think, look, here's the thing. This company is going to go down in the history books. In one part, because the technical inventions that they've created are just next level, and frankly, created an entire industry. And I think they deserve a ton of credit for that. But they're also going to be written in the history books for two other things that are probably less aspirational.
I think the first is that there's just all kinds of dust-ups and unnecessary drama that just seem to kick around every few weeks or months. And then the second is the sheer quantum of value capture that the employees have seen through secondaries before a fully functional business has been really created. And so I think it explains why folks circle the wagons consistently.
I think it's a very rational organization. They're technically ahead of everybody else. A lot of people want to put in money at crazy prices. A bunch of that value is transferred immediately to the employees who circle the wagons and do what's necessary to keep the taps flowing. And I think that that explains the whole thing.
And I think that that explains many Silicon Valley companies, quite honestly.
circle the wagons, defend the company, sell your shares in secondary at 90 billion to Thrive or whoever. What would your take, Friedberg, on all of this craziness and drama?
I think what we will see over time is that rather than have the ability to sue for their likeness, a lot of AI that is like a celebrity the value of it will arise from the celebrity's endorsement, not actually using the celebrity's features.
So without the endorsement, I know everyone kind of wants to point to this idea of likeness, but I think that there's something about the authentic aspect of having the celebrity actually endorse and provide their endorsement their signature, their stamp on it. Restaurants are a good example. Some celebrity chef says, I was involved in making this menu.
That's a lot different than mimicking the celebrity chef's menu from his restaurant, putting his or her name and brand on it. There's a bunch of videos on YouTube now. I don't know if you guys ever, you guys probably don't watch these, but I love watching these videos where like music producers make tracks and how they do it.
And so many of these producers now are using AI tools, taking samples off of old records or other tracks, and then telling the AI, make something that sounds like this or looks like this, but isn't like this. And so there's enough of a transformation happening that it isn't a direct likeness.
And then they're able to create entire vocal tracks without needing a singer or without needing the celebrity singer.
So you're in the you're in the much ado about nothing on this card.
Yeah, I'm in I'm in the like, like the content itself, I think is probably less like compelling. Oh, you go go after her because the voice sounds the same. But I do think that there's this element of like, what if you could then say, hey, you know, Britney Spears actually lent her voice to this track.
So do you think if she if she sues open AI, do you think it should just get thrown out? It's not a real case.
No, I think there's probably going to be a lot of discovery to Sax's point that's going to show that they probably did. Yeah, the discovery is going to be juicy.
No, no, no. I'm not saying that. I'm asking you more that, yeah, even if they find it, your point is it shouldn't matter. My point is, do you think that it should be thrown out?
Well, I don't know. I mean, hold on. let's say you're casting a movie and you know, you can't get Scarlett Johansson for it. And so you tell the casting director, get me a Scarlett Johansson type. I think you can do that. Okay. Obviously you can't, you can't use her name. You can't use her likeness, but you could hire a different actor who might look or sound like Scarlett Johansson.
If the company actually did that and they did it nine months ago, this is what the statement they put out. I think they've got a decent defense. Yeah. But because they may, I don't know if we believe that. I mean, again, why don't you just put out the name of the actual actor, that voice actor that you used?
And there's a very simple test here. If there's confusion amongst the public, which is what Scar Jo put in her letter, and that was a legally written, deftly written letter to set up a huge settlement. Because she said, the morning this came out, all of my friends said, oh my God, congratulations on your Chachi P. Teodale. This is great. The public being confused is the key issue here.
And there's something called the right to publicity. This is basically how celebrities defend their rights. It's happening all the time to podcasters, by the way. There was a company that put me in their ads based on something I said in a show. And then they put ads against it. And Huberman's been having this happen. Joe Rogan's having this happen. It's happening to me right now.
I'm in the middle of this crazy thing. With Facebook, and they've been doing a very good job, the team at Meta, Ashley, thank you. But hundreds and hundreds and hundreds of automated accounts pretending to be me selling all kinds of random stuff. It's predominantly on WhatsApp and Facebook and Meta. And I don't know what to do because we work together with them. We shut it down.
I've actually had to reactivate my Facebook and Instagram accounts, which were dormant. so that we could actually have them be verified, so that then it's easier to shut them down. But it is an impossible task when somebody's impersonating you to fight it. At least in my experience, it's been a month and it's just like every hundred we take down, another thousand.
In both of those cases, you guys have your exact image of you being shown to sell stuff. I think that in this case, it's also unique to ScarJo because she was the voice from her. If this were a voice sort of like Cameron Diaz or sort of like Julia Roberts, it wouldn't be as big a deal because it certainly got some differences to it. But it's because they're trying to mimic the movie Her.
Well, didn't Meta use Morgan Freeman when Zuck created an AI or something as a project? Do you guys remember that? Wasn't Morgan Freeman the voice?
They paid for it. There's a company, Speechify. I'm not an investor or anything like that, but they have Gwyneth Paltrow as a licensed voice to read your stuff.
So they should just license Gwyneth's voice. Obvious.
Obvious. Like just whoever wants to get paid, here's the opportunity. If three people say no, the fourth will say yes.
What do you guys think about the fact that they're not trying to say this is this person's voice, but they want to say like, hey, let's say you just can prompt the AI and say generate a voice that is sort of like movies that are comfortable and calming to people to listen to. And, you know, that we think people will be comfortable. And the AI generates something that sounds like ScarJo.
I've said this before.
but it's not deliberately trained on ScarJo.
Using a computer to probabilistically copy something is still copying something. Yes. Come on, guys, let's not make this too complicated.
Is the public confused is the only test you need for you. What if there's two public actresses that both have similar voices and then they both claim, hey, you tried to make this sound like me. What do you do in that case, Jamal?
Well, which one did you call two days before the demo?
Yeah, exactly. And did the CEO tweet her? I think your whole point about discovery is what is going to get them in trouble in this case. Because they were clearly trying to do an impersonation of her, right?
Yeah, if they never reached out to Scarlett, they could claim it's just a coincidence. But they called her twice. They called her some months before and then two days before, which indicates panic.
Judge Sachs, give your verdict. I'm starting a new spinoff show. Judge Sachs says... Guilty. Okay. Judge Sachs, sentence now. What's the sentence?
The sentence is that Scarlett Johansson is going to end up owning more of this company than Sam Altman. That's what's going to happen. Look, they call her two days before. Why do you do that? Guilty. Because you know you have a problem and you're trying to put the horse back in the barn. Now what they should have done is as soon as she says no, you just change the voice completely.
Or you get a fixer. Get Michael Cohen in there and get a settlement going. Let's get some fixer in there to fix it. Okay, listen, enough with OpenAI. Oh, wait, there's more. Geez, it's ruining the docket. I guess we have to talk about the next drama from OpenAI this week.
Former employees sign an agreement that they're forbidden forever from criticizing the company or even acknowledging that the NDA exists. If a departing employee declines to sign the document or if they violate it, they can lose all vested equity they earned during their time at the company.
In practice, this means ex-employees have to choose between giving up millions of dollars they've already earned or agreeing not to criticize the company in perpetuity.
Well, you can see why they wrote that in there. That makes sense if you think, hold on a second, if an employee can leave with, you know, a random copy of like some old weights as a starting point to rebuild the model, that could be very valuable or... Well, that's IT. Yeah, but there's all kinds of quasi-confidential information or knowledge or know-how that you leave a place like that with.
So I could see why there was a justification to be very... heavy-handed about it. But again, the question isn't whether you're allowed to be heavy-handed. You are. The question is, why backtrack and then obfuscate and lie after you get caught? Well, right. What do they say?
They claim it's an accident. Once they get caught with their hand in the cookie jar, which, like you said, Chamath, is just a heavy-handed agreement. It's in the company's interest to do this. Yeah. They just say it was an accident, just like the Scarlett thing is an accident or a coincidence. It's just getting hard to believe.
Just own it and say, you know what, guys, this is a really valuable company. There's a ton of very valuable trade secret know-how IP confidential information. And we're going to be extremely- Litigious. On the offense and protect it because- Aggressive. And it's correlated to the importance of the company and the ecosystem.
You could have said that and people could have been upset, but they would have understood.
Here's what Sam Altman said. There was a provision about potential equity cancellation in our previous exit docs. Although we never clawed anything back, it should never have been something we had in any documents or communication. This is on me. And one of the few times I've genuinely been embarrassed running OpenAI. I did not know this was happening and I should have.
If any former employee who signed one of these old agreements is worried about it, they can contact me and we'll fix that too. Very sorry about this. So he's very, very sorry. It's starting to be like BP Oil, this company. They're just so sorry about everything.
Here's the question is, These are form documents at the end of the day. And form documents don't write themselves. Lawyers write them. And when you get a novel change in one of these documents, somebody thought that through and thought it'd be a good idea and put it in there.
Right.
And like Jamal said, there is a way to potentially defend that. It's not like these provisions don't exist. It's just a novel application to try and claw back someone's already vested equity from a company on behalf of these things. Well, that is not...
Yeah, that is not exactly standard.
No, that's completely non-standard. I'm just saying that these provisions exist in other contexts, and their application as a clawback of vested employee equity is something that I don't think any of us have heard before.
I haven't, yeah.
Yeah, exactly. So my point is just this didn't happen as an accident. Somebody made a strategic business decision to do this because they thought it would be in the company's interest.
Yeah, so just in layman's terms, the clawback means you had $10 million in equity invested you earn 75% of it, you've got 7.5 million in equity there. You say something disparaging about the company, they can take it back from you.
Once you leave a company and you vested equity, you don't lose it. I mean, as long as you, you have some period in which to exercise your option if it's an option rather than stock. But other than that, I've never heard of a situation where employees can lose their vested equity.
Even in a situation, Sax, we've seen this, where somebody commits fraud they still get their vested equity. And then it's up to the company to sue them for fraud separately, right? Like we've seen instances of that.
I mean, I guess that's right.
Like committing a crime.
Look, I think the question here is, is it credible that they keep having these accidents and coincidences?
What does Judge Sacks say?
Judge Sacks says there's one too many coincidences, you know? Look, I think like Chamath said, you could have just owned this and said that, defended it in the way he said, and said, but you know what? It was too aggressive and we pulled it back.
I think the thing is, look, if you think about the pendulum of culture in Silicon Valley, we used to have a very tough culture of founder-led businesses where there was extremely high expectations. And if you transgressed, it was very punitive.
And then the pendulum swung all the way to the other opposite end where you had this like coddling daycare type approach that existed for like the last 15 or 20 years. And probably what OpenAI is is an example of a company that needs to be run a little bit more like the former but stuck with a bunch of people that still pull it towards to be the latter.
And that's the cultural tension that they're going to have to sort out because in order to be this incredible bastion of like AGI and innovation, I suspect that it's going to look more like a three-letter agency in terms of security and protocols in the next five or 10 years, then it is going to look like the Googleplex. And I think they just need to own that.
And this is probably a little bit of an insight into that tension. And they're going to have to go more in that direction. I think it's a good insight. You're not going to be allowed to build these incredibly crazy world-beating technologies where people are running around in an eight-seater bicycle. It's just not going to work, guys.
And by the way, I mean, because it is such an industry-leading company, I think we could end up with some very bad fair use precedents or laws because Scarlett Johansson is so sympathetic as a plaintiff compared to OpenAI.
And unless they show us some discovery that proves that they really did hire the voice actor and all the rest of it, I mean, this could lead to some very bad precedents for the industry around fair use.
Well, and here we go. I think the Microsoft will pay the speeding ticket and we'll just all move on. But Freeberg, my God, can you imagine like being two or three PhDs and, you know, machine learning or whatever, you study your whole life, you're pursuing general AI and like people are coming up to your desk and creating all this drama and nonsense.
And you're in the middle of a soap opera while you're trying to create the technology that creates super intelligence. It's nuts.
Yeah, I find it annoying to just listening to it.
Okay, well, then in that case, we will move on from OpenAI. Oh, sorry.
There's one more drama going on. It's not just drama. I mean, there's now a lawsuit. I think it's a very interesting case.
The fair use case is interesting, for sure.
I think it's a legitimately interesting case. If it goes the distance, it's going to create some really interesting precedents.
Well, I mean, this is already... And I'm not saying we have all the facts yet, but... Yeah, what happens in these content cases is they get settled almost every single time. So the case law doesn't get codified. They just get settled out of court. If you go look at all the fair use cases, they almost never go to the mat. And so this one will just be settled.
It'll just be a question of at what price.
I think the other... The interesting part of the other story is that the reason all the stuff came out about the equity clawbacks was because the safety team quit and it got leaked during that process. All right.
So this is the third dramatic story of the week.
That one, I think begs a little bit more of a question of...
Two heads of OpenAI's super alignment team left the company last week, the day after GPT-4.0 launch. Ilya announced he was leaving the company. He was our chief scientist. A few hours later, his partner on the alignment team, Jan Laika, also announced he was resigning.
In a later thread, Laika explained that he left due to, quote, safety culture and processes have taken a backseat to shiny products. Okay, there's a little bit of disparagement to our former point about non-disparagements and NDAs. So OpenAI lost both its heads of AI alignment one day after it launched that new product. Is that a coincidence? It's interesting.
If you don't know what super alignment is, it's basically making sure that the software doesn't go Terminator. What are your thoughts on this, Friedberg?
I don't know. I mean, it could be some bad bureaucracy, bad politicking, not being listened to. But I think the real interesting question is who's going to ask these guys what's really going on from a technology perspective and what is that going to reveal? Because these guys clearly are on the frontier of model development and the performance of models. And so my guess is...
There are certain regulatory people who are going to have interest in the fact that this team just left. They're going to make a phone call. They're going to ask this team to come in and have a conversation. And they're going to start to ask a lot of questions about what the state of technology is over there. And I suspect that some things are going to start to come out.
Sacks, it was reported that Ilya was on the side of the... nonprofit-y, slow down AI, be cautious group when they fired Sam. So what's your take on what's going on here with super alignment inside of OpenAI? Judge Sachs.
Let's call this what it is, a mass resignation. And we don't really know why. I mean, apparently they were promised something like 20% of the computing resources of OpenAI, and they didn't get that. I definitely read that somewhere. And so that is part of it, I think, but we don't really know the whole story.
And when you look at this issue of the mass resignation, and then you look at the issue of the clawback, of vested employee equity, you're like, well, wait a second, maybe they felt like they needed that claw back in order to deter all these people who are leaving from spilling the beans about whatever was upsetting them. Something clearly upset them, right?
Yeah, that's why people are saying there's this meme, what did Ilias see? What did he see?
Yeah, and then like you said, the board did fire him, and the only explanation they provided was that he wasn't being candid, which at the time we thought was an incredibly damning statement. And we thought we'd get some explanation of it. We never got any explanation whatsoever.
You know, I thought that the board was being incompetent because I thought that either they fired him overly hastily or they had reason to fire him, but then they communicated poorly. And, you know, you add all these things up and it definitely seems like a lot of smoke.
Sam's a straight shooter. We should just have him on the pod to explain why. I mean, it'll clear everything up. Stop me if you've heard this before. NVIDIA just smashed all expectations while reporting record profits and revenue. The AI train continues on Wednesday. NVIDIA reported earnings for the fiscal Q1. Revenue was $26 billion, up 18% quarter to quarter, 260% year over year.
Basically, they quadrupled. year over year on billions of revenue. This chart is bonkers. We've never seen anything like this in the history of Silicon Valley or corporate America. This is if somebody literally was mining coal and then found a diamond and gold mine underneath it. It's bonkers what's happened here.
When you look at the revenue there, the sort of slow growth or moderate growth revenue that they experienced, that was all because NVIDIA was primarily providing GPUs
For people playing video games or mining crypto, and then what you see with this unbelievable six-quarter run and five, six-quarter run is companies like Microsoft, Google, Tesla, OpenAI, etc., buying just billions and billions of dollars worth of hardware. I'll end on this and Chamath and get your take on it. Here's 2019 top companies by market cap in the world.
Obviously, Microsoft, Apple, Amazon, Google, Berkshire, Facebook, Alibaba, Tencent. And then you get some of the, you know, incumbents and legacy companies, J&J, Exxon and JP Morgan Visa. Way down on the list in 2019, number 84 was NVIDIA. Today, NVIDIA is the third largest company by market cap behind Microsoft and Apple and ahead of Google, aka Alphabet and Saudi Aramco.
Shamath, what's your take on this? Will it continue? And how do you conceptualize this level of growth on such a big number?
I think it's a really incredibly fun moment if you're involved in anything AI-related, just because it shows the level of investment that NVIDIA's customers are making into making this new reality available for everybody, right?
So when you're spending effectively $100 billion a year on the capex of chips, and then a couple hundred billion more on all the related infrastructure, and then another couple hundred billion more on power, you're talking about half a trillion to three quarters of a trillion dollars a year being spent to bring AI forward to the masses. So I think that's the really positive take.
The other exciting thing is if you're on the other side of the Nvidia trade, which is you're working on something that does what they do cheaper, faster, or better, it's also really exciting because at some point the laws of capitalism kick in, right? We've talked about this. When you are over-earning so massively,
The rational thing to do for other actors in the arena is to come and attack that margin and give it to people for slightly cheaper, slightly faster, slightly better so you can take share. Yes. So I think what you're seeing and what you'll see even more now is this incentive for Silicon Valley, who has been really reticent to put money into chips. really reticent to put money into hardware.
They're going to get pulled into investing in this space because there's no choice. You have a company that went from $100 billion in market cap to $2.5 trillion in four years. It's just way too much value that is there to then be leaked back. The interesting thing to remember is During the PC revolution, which is really mostly the 90s, right?
It ended in the late 90s, I would say like 98, 99, right before the dot-com bubble took over. Intel's peak market cap was, I think it got to about $200 billion. And then their average growth rate from 1998 to today was negative 1.4% a year, right? So it went from about $200 billion to about $130 odd billion. And why? It's not that Intel was a worse company. But it's that everything else caught up.
And the economic value went to things that sat above them in the stack. Then it went to Cisco for a while, right? Then after Cisco, it went to the browser companies for a little bit. Then it went to the app companies. Then it went to the device companies. Then it went to the mobile companies. So you see this natural tendency for value to push up the stack over time.
And in AI, we've done step one, which is now you've given all this value to NVIDIA, and now we're going to see it being reallocated.
So, Chamath, who's in the arena trying stuff? Some of these things working, obviously, and some not working.
So, right now, what you do is you speculatively bet on anything that kind of like, quote unquote, rhymes with NVIDIA. Yeah. AMD is ripping, the companies that make HPM is ripping. So all of that stuff, the folks that make optical cables, this Japanese company that I found that makes like the high bandwidth optical cables ripping.
So anything related to that ecosystem right now is at all time highs. But at the same time, What you find now is every other day when you wake up and read the trades in Techland, you find that there's a company that's gotten seeded with $5 to $50 million to create a new chip.
You're also starting to see folks that are working a little bit above the stack and build better compilers, things that will allow you to actually build once, run in many different compute environments. So all of this stuff is starting to happen. At some point, the spread trade will be that NVIDIA loses share, even though revenues keep compounding to these upstarts. Yeah.
Death by a thousand startups. All right, I guess one of the questions people are asking right now is, have we ever seen a company at this scale and the impact it's having, not just in technology, which Shamath just pointed out beautifully, but also it's having a huge impact on Wall Street, on the stock market, on finance.
Well, the company that everyone compares NVIDIA to or asks the question about whether a historical comparison should be made is Cisco. So there's an article in Motley Fool saying, is NVIDIA doomed to be the next Cisco? There was one in Morningstar called NVIDIA 2023 versus Cisco 1999, will history repeat itself?
The reason they're asking these questions is that if you go back to the dot-com boom in 1999, you can pull up the stock performance chart, you can see that Cisco had this incredible run And if you overlay the stock price of NVIDIA, it seems to be following that same trajectory. And what happened with Cisco is that when the dot-com crash came in 2000, Cisco stock lost a huge part of its value.
Obviously, Cisco is still around today and it's a valuable company, but it just hasn't ever regained the type of market cap it had. The reason this happened is because Cisco got commoditized. So to Chamath's point, the success and market cap of that company attracted a whole bunch of new entrants, and they copied Cisco's products until they were total commodities.
So the question is, will that happen to Nvidia? And I think the difference here is that at the end of the day, networking equipment, which Cisco produced, was pretty easy
Pretty one-dimensional. Yeah, pretty easy to copy. Moving data around, yeah.
Right. Whereas if you look at NVIDIA, these GPU cores are really complicated to make. And Jensen makes this point that the H100, for example, has thousands of components and it weighs like 70 pounds or something like that.
Yeah, it's like a giant oven.
Yeah, it's like a mainframe. It's not just like a little chip. So it's a much more complicated product to copy. And then on top of that, they're already in the R&D cycle for the next chip, right? Whatever it's going to be, the H200 or whatever it is. And so... As people try to catch up with H100, they're going to be on to H200.
So I think you can make the case that NVIDIA has a much better moat than Cisco. And just by the way, on this Cisco comparison, just to finish the thought, people were making this comparison six months ago. And what's happened since then? NVIDIA's had two blowout quarters. And the competitors don't seem to be that much closer, maybe a little bit closer. But so...
You know, look, I think it's an open question.
So there's a counter here, Freeberg, which is obviously, if you follow the Cisco analogy, one of the things that also sunk Cisco was once people had bought all that capacity, there was no need. There was no file size that was so great that it couldn't be moved easily around the internet. You know, you make movies...
HD, super HD, 2k, 4k, the bet we've created too much bandwidth, there was no use for it. So I guess that that's a counter argument for maybe when, if we build up too much capacity, NVIDIA also could not by competitors, but just by the build out being enough. So what's your take on that counter argument, and then whatever I thought you have?
I think that the Cisco analogy, it's a pretty different situation because Cisco evolved the business to become much more enterprise centric. And they were able to run an M&A process like we see with enterprise software, where they could acquire and roll up lots of different product companies and sell into their enterprise channel. So do a lot of cross-selling.
NVIDIA is not a super acquisitive business and it doesn't make as much sense because they're selling much more kind of infrastructure tools, whereas Cisco moved really high up in the enterprise stack. They were selling stuff into office buildings, they were selling software, they did acquisitions to kind of fully integrate.
They had a very diverse set of products that were selling through an enterprise channel, further up the value stack, and a pretty distributed customer base, no serious concentration. Even though they did sell a lot into data centers, they were also selling to telcos, they were selling to enterprises, they were selling to governments, and so on.
If you look at NVIDIA's revenue, they did $26 billion of total revenue in the quarter, $22 billion of which was data center, and about 40% of that was from the top four hyperscalers. So a full one-third of Nvidia's revenue in the quarter came from, I believe it's Google, Amazon, Microsoft, and Meta.
And so between those four businesses, you know that those companies each have, I believe, at least over or close to $100 billion of cash sitting on their balance sheet. They can't find great places to invest that cash to grow revenue. And so they've rationalized away the idea that they will make CapEx investments to build over the next five to 10 years. And this is where that money flows.
Yeah, we talked about that in a previous episode, because there's no M&A, to your point. Unicon's not going to let you buy stuff, or the UK is not going to let you buy stuff.
So I think that they're going to have less maneuvering capability than Cisco had in the future. And obviously, there's this deep concentration risk, which is going to be deeply challenging.
I think NVIDIA, this is to build on Sax's point, is going to get pulled into competing directly with the hyperscalers. So if you were just selling chips, you probably wouldn't. But Saks is right. These are these big, bulky, actual machines.
Then all of a sudden, you're like, well, why don't I just create my own physical plant and just stack these things and create racks and racks of these machines? And go head-to-head with AWS instead of selling to them. It's not a far stretch, especially because NVIDIA actually has... the software interface that everybody uses, which is CUDA.
So I think it's likely that NVIDIA goes on a full frontal assault against GCP and Amazon and Microsoft. That's going to really complicate the relationship that those folks have with each other. But I think it's inevitable because how do you defend? It's kind of the Apple problem. How do you defend an enormously large market cap? You're forced to go into businesses that are equally lucrative.
Now, if I look inside of compute and look at the adjacent categories, they're not going to all of a sudden start a competitor to TikTok or a social network. But if you look at the multi-hundred billion revenue businesses that are adjacent to the markets that Nvidia enables, the most obvious one is the hyperscalers, which are multi-hundred billion dollar revenue businesses.
So they're going to be forced to compete. Otherwise, their market cap will shrink And I don't think they want that. And then it's going to create a very complicated set of incentives for Microsoft and Google and Meta and Apple and all the rest. And that's also then going to be an accelerant.
They're going to pump so much money to help all of these upstarts, to your point, Jason, chip away and nip at the Achilles heels of NVIDIA until they fall.
Yeah, and there's a great precedent for what you're saying because, or clues, Amazon is making chips, Google is making chips, and in fact, Apple- Facebook's making chips.
Tesla's making chips. Yeah.
And then Apple makes their own chips and they got rid of Intel. And so this is how it's going to go. Your margin is my opportunity. And, you know, with all this market cap increase, the good news is just reported that Jensen has bought a second leather jacket. So, well, this market cap has enabled him to Expand the wardrobe a bit.
61?
He looks amazing.
He looks amazing. I'll tell you something. In the zombie apocalypse draft, I'm picking him. That guy seems crafty. He's so resourceful. What do you think he does for exercise? No plastic. Plastic-free balls.
No, no, no. His balls are plastic.
He's got steel in there. No, no. His had brass in them.
His literally had brass put into his balls. No, all of our balls have plastic, apparently.
Okay. Well, we have two choices here. We can go with another tech story, or we can go directly to Science Corner. Well, no.
I think we should do State of the Economy and then do Science Corner.
Okay. So you want to talk about our pocketbooks, and then we'll talk about our pockets, and then we'll just make a quick detour to the right and then talk about our balls. We're in the same vicinity.
I think we should end on our balls. Okay, great. Let's end with our balls.
I mean, Freiburg told me he likes to start with the balls. I don't know. Everybody's got different kind of vibes here.
The ball play should come a little bit later in the program.
You want to save the ball play for later, Saxon? Yeah, yeah. I'm having a hard time keeping this together, guys. More than half of Americans think we're in a recession. I think we're in a vibe session right now because we're not in a recession. But people are feeling really bad. A Harris poll conducted by The Guardian shows 56% of Americans wrongly believe the U.S. is in a recession.
Not surprisingly, they blame Biden. The poll highlighted a bunch of misconceptions. 55% believe the U.S. economy is shrinking. It's obviously not. 56% think the U.S. is experiencing recession. It's obviously not. 49% of people believe the S&P 500 stock market index is down for the year. It's up 12% this year. It was up 24% in 2023.
And 49% believe that unemployment is at a 50-year high when it's in fact at a 50-year low. and Americans are really concerned about the cost of living and inflation. Fair enough. 70% said the biggest economic concern was the cost of living. 68% said that inflation was the biggest economic concern. Important quote here.
A majority of respondents agreed, it's difficult to be happy about positive economic news when I feel financially squeezed each month, and that the economy was worse than the media made it out to be. According to the polls, 70% of Republicans and 40% of Democrats think Biden is making the economy worse. Chamath, you have some thoughts on this.
I'm going to go out on a limb and speculate that a lot of the big numbers that we use to gauge how we should feel about things in today's day and age are pretty brittle. And fragile and may actually just be totally wrong. So what's an example? So for example, if you look at something like non-farm payrolls, right?
So the first Friday of every month, you get this report that comes out from the Department of Labor and it shows where unemployment is. But how do they calculate that? Do you think that they have a real-time sense of exactly every person in that month that entered the workforce or exited the workforce? No. They do a survey and then they extrapolate.
And if you do that survey incorrectly, and Jason, you've commented on this before, for example, if you don't capture adequately the number of people that are on the sidelines and never joined the workforce or the number of people that are part of the gig economy, so they are kind of working, you get an inaccurate sense of where the real economy is.
I think that GDP is somewhat similar because if you just break down what GDP is, so Nick, there's a very simple pie chart I sent to you. What is GDP? It's the sum of four things. Most of it is what people spend. Okay. Then the next big chunk is what companies and governments spend, and then the last is what we export to other countries.
So let's just pause for one second and think about what do you think happens when rates are zero versus when rates are at 6%? People spend a lot more. Well, people tend to save when interest rates are high. Just the natural thing, like why would I buy a pair of these Nike shoes? I'll just put it in the bank, I get 6%.
But when the bank pays you zero, you're like, ah, let me buy these Air Force Ones and move on, right? It turns out it's the same for companies. Companies find it easier to invest when rates are at zero because it's cheaper. It's much more expensive because they're borrowing money at 6% versus at 0%. Or more. Corporate gets charged a higher fee, right? Yeah.
Then when you have high interest rates, you have a currency that appreciates. It makes exports less attractive to other people, which means then you become an ad importer. Okay, so what is the last thing that's left? The last thing that's left is government spending. And you have to ask the question, what should governments do when rates are high?
There was a chart I published in my annual letter, if you just go to that for a second.
And just going back to this chart right before it, just so the people who are listening, if you put the pie chop art in there, important for people to know, consumer is about 70% of the economy. And if you put investment and government together, that's just over maybe 34% or something like that. Yeah, 35%.
So it is a consumer-driven economy, but hey, corporate and government spending is a major piece as well.
And then I just wanted just to highlight that when interest rates are very high, all of a sudden governments are faced with this very difficult problem, which is, oh man, I have to spend a ton of money on interest. Just like if you had a bunch of credit cards and all of a sudden the interest rates went up. So the choice is twofold. Do governments spend less?
But unfortunately, it turns out that our governments in America, they just keep spending more and more. So even if net interest income is small, even if net interest income is high, they're just like, forget it, the taps are on. So what does this all mean? I think what it really means is that we do a very poor job of measuring all these dynamics together.
And so I actually trust the survey data of these individuals more than I trust the GDP report. in the sense that I think it more accurately captures this dynamic. Rates are at 6%. People are saving more. They're not getting paid more. Things are costing more. The government is giving you free money, so you kind of feel like everything is moving.
So the GDP measurement, the way that it's classically done, shows that, wow, we grew at 3% or 4%. But the average individual American isn't feeling that. They're actually feeling that they have less money.
So I would actually go with them and actually say, if we don't revisit this thing from first principles, we're going to get this dynamic where we think one thing is happening, but the actual exact opposite is happening. In this case, I do think we're in a quasi-synthetic recession.
Saks, what's your take on the vibe session?
Well, look, I tend to agree with Jamal on this. I think this is a classic story of who do you believe? Do you believe the experts or do you believe in the intuitions of the American people? And the experts have some statistics on their side. But, you know, the old saying goes, there's lies, damn lies and statistics. And then the American people have their actual lived experience on their side.
They know what they're feeling. And I tend to trust in that. And obviously, we're in an election year, and the press knows that. So they're trying to do this big cleanup effort for Biden. But why is it that people are feeling this way? Number one is inflation. And if you look at this chart, you can see that
If you look at household net worth since the start of the Biden presidency and compare it to the change in household net worth at a similar point in Trump's presidency, in nominal terms, it appears to be the same. But then if you adjust for inflation, in other words, you look at the real household net worth, you can see that household net worth during the Biden term has been flat.
Actually, it's down.
Because of inflation, right?
Because of inflation.
Where did the inflation come from?
Where did the inflation come from? Yeah. Well, Larry Summers warned in the first quarter of the Biden administration that if you pass an unnecessary $2 trillion of COVID stimulus, you would produce inflation. The inflation rate when Biden came into office was 1.7%. We had a rip-roaring economy, but he started stimulating. And we talked about Bidenomics is this new policy that
of pumping trillions of dollars of stimulus into a healthy economy, which we've never done before. What happened? Inflation went all the way to 9%. So people's wages have not kept up with the rate of inflation. This is why they feel worse off. When you actually look at purchasing power, people are worse off in terms of their actual ability to buy things. Their purchasing power has gone down.
Wages may have gone up a little bit, but they have not gone up as much as inflation. So people feel worse off. Now, Larry also had that, I think, really informative study showing that inflation would have peaked at 18% if you include cost of borrowing. So again, to Chamath's point, if you're trying to get a mortgage and you're paying 7.5%, 8%, you'll feel way worse off.
If you need to buy a car and make a car payment, you feel much worse off. If you've got credit card debt, which has now hit an all-time record of something like $1.1 trillion, your credit card rates have never been higher. So the average American feels worse off because cost of borrowing has a huge impact on their household finances.
And that's why if you read like one of the last paragraphs in that story that you referred to, they use the key words, the consumer feels squeezed, the average household feels squeezed. They may not have lost their job yet, but they've lost purchasing power and they've lost net worth.
They're under earning. They're under earning. And so the projection- So this is obvious.
And the press can gaslight us all day long about how wonderful things are under Biden, but the average American, I think, understands differently based on their own experience.
I think the whole thing comes down to the projection of an individual or a household of their lived experience onto the economy. You assume that because you're having a tough time, the economy is bad. And the economy as a definition for them is how do I earn and how do I spend?
And if I'm under earning, that means there must be serious job loss and things are more expensive and my ability to purchase isn't improving. And so I think we're all kind of gonna end up on the same take on this one.
I mean, Nick, if you wanna pull this image up, this is I think a helpful one, which is disposable personal income relative to outlays that folks are needing to spend more than they're making. So clearly indicating that they're feeling like they're under earning. So the projection of that is the economy is bad without recognizing that it is an inflationary experience.
Whereas economists use the definition of quote economic growth being gross production, gross product. And so if gross product or gross revenue is going up, they're like, oh, the economy is healthy. We're growing. But the truth is we're funding that growth with leverage. at the national level, the federal level, and at the household and domestic level.
We are borrowing money to inflate the revenue numbers. And so the GDP goes up, but the debt is going up higher. And so the ability for folks to support themselves and buy things that they want to buy and continue to improve their condition in life has declined, things are getting worse.
And if you go to the next image, as Sax pointed out already, here's the image of total outstanding credit card debt over a trillion dollars, it's totally spiked. And it's gonna continue to spike just like federal debt because of the next chart, which is the sudden jump in interest rates. So we've seen credit card interest rates jump from 12% on average 10 years ago to 21.19% right now.
And it was at 14% at the end of 2022. So we've gone from 14% average credit card interest rates to 22% now in just about 20 to 24 months. And so the projection that I think of the quote, economy must be bad is resulted from the fact that income to spending is actually pretty negative. So here's the real median family income. This actually only goes through 2019.
So it doesn't even capture the era that we're talking about, but this has been going on for quite some time that the average American's ability to improve their condition has largely been driven by their ability to borrow, not by their earnings.
And this has created a substantial set of precedents that we're now running into a wall with interest rates spiking and inflation hitting us because of the overall federal debt that we've taken on.
I think we're probably going to go around the horn and all agree. Obviously, the crazy spending started in the Trump and COVID era, and that caused a lot of the inflation as well, just to be fair. It's two administrations that are just out of control with spending. But the way I look at this is the Mickey D economy is
People may not know this, but 96% of Americans eat meals at least once a year in McDonald's. 8% of Americans eat at McDonald's on an average day. And when you look at the prices of McDonald's here, if we look at this image- Yeah, this is incredible.
This is incredible.
This is unbelievable. Medium French fries at McDonald's, $1.79 in 2019, and now $4.19. And then if we look at just McNuggets- My gosh. $4.49 to $7.58. 68% increase. McChicken, 129 to 389. And then here is a very interesting one. This is CPI versus McDonald's. Big Mac prices. Take a look at that. As much as the consumer price index has surged, Big Macs have exceeded that.
And so Americans are seeing this over and over again when they go to McDonald's and other places. And that's what's causing the feeling.
Because when you spend... Go back to that McNugget chart. I just want to see what is it end of 2019. So this is basically you'd want to look at the four year stock price, right? So like these guys have jacked up prices. Massively. If you look at what's happened to the stock, the stock is way up.
It's kind of been they've been very motivated and rewarded as a company for just rewarding the shareholder and kind of screwing over the customers.
If you own equities in McDonald's and you're in the top third or half, maybe half of Americans who have equity exposure, you're feeling great. If you're on the bottom third or half and you're buying at McDonald's and you don't own equity at McDonald's, you feel terrible. Freebreak, you had some additional thoughts?
Yeah, but remember what McDonald's and other fast food companies have said is that labor costs have climbed. Here's a chart on labor costs that Nick can pull up. Workers at Walmart and McDonald's have had pay increases, but this has really been to try and keep up with inflation, the inflation of other costs. So a lot of people will say, oh, they're price gouging.
They're ripping off consumers to make profits for shareholders. But the truth is the biggest component of running those restaurants is labor. And labor has gotten more expensive because the employees that work there have to earn enough to pay their bills and to afford their food. And this is the circular effect of inflation. It finds its way all through the economy.
It filters down and it eventually hits everyone.
Look, fast food's a pretty competitive business. I mean, I think the reason why McDonald's is raising prices is because everyone else is raising prices. I mean, otherwise, they'd be losing share. And just go to