SpaceX, AI and Mega-IPOs: What’s Really Happening Behind the Scenes

00:00:00: If Elon really thought that AI was going to make money, why would he take his most valuable resource?

00:00:06: That thinks he has an advantage over everyone and rent it out.

00:00:10: Because a few billion dollars to him is nothing.

00:00:12: if you can make trillions of dollars in revenue The fact even right now spending the time making these kind of deals means that he does not believe that AI will bring him the money to justify his IPO.

00:00:27: Today we asked, will AI labs really make trillions?

00:00:33: And We are talking about SpaceX.

00:00:34: The IPO is it the next super boom?

00:00:38: or Will It be the Next warning sign?

00:00:42: and Does AI know when it Is wrong?

00:00:46: at least Atleast we're Talking About the Pope and the Pope is Warning the World about AI.

00:00:54: Welcome to the Big Bang Tech Report.

00:00:57: Welcome, Alvin van Grelen directly from Seattle here in Berlin.

00:01:02: Yes well yeah it's great to be face-to-face.

00:01:04: this is our first uh in person show so I hope you guys enjoy it!

00:01:09: I hope too.

00:01:10: So Alvin Silicon Valley treats AGI as the next jackpot a super jackpot yes and It's the biggest one history.

00:01:21: And you say It is not true.

00:01:24: What is it?

00:01:26: For you as a Mirage,

00:01:27: why?".

00:01:28: Yeah so I actually just released the new paper about a week ago called The AGI Windfall Mirage and i wanted to help people understand... People who are NOT in the AI industry understand the mindsets of the AI lab leaders And there's this perception out there that whoever gets to AGI first as a Lab will have the technology the global... white collar worker wage pool, which is about twenty to thirty trillion dollars.

00:01:58: So this is why everybody's racing the spend hundreds of billions or a trillion dollars.

00:02:02: two create that advantage because they think that it will create that big jackpot.

00:02:08: and soon we'll have these SpaceX IPO coming this week.

00:02:13: And if you look at the S one filings Which Is The Document That It Goes When It Goes Public the biggest portion of his future revenues, a twenty-two trillion dollars expected from what they call enterprise software.

00:02:31: And Enterprise Software Market today is around five trillion dollars globally which means that there's another seventeen trillion.

00:02:37: and What do you really mean?

00:02:38: Is we will be able to replace enterprise office workers with this software?

00:02:44: I think thats very misguided goal.

00:02:48: But first of all, to understand you are talking about AGI off AGI.

00:02:53: we're not there with AGI.

00:02:55: We were going through this next level right?

00:02:58: Yes that's right.

00:02:59: I don't think we have AGI today but

00:03:01: what is it?

00:03:02: a AGI explainer to our audience.

00:03:04: yeah

00:03:04: so AGI.

00:03:05: There's unfortunately different definitions.

00:03:07: i'll give kind of three different interpretations.

00:03:10: one is technology That will be able to replace existing office workers cognitive workers, whether accountants or secretaries.

00:03:20: They can

00:03:20: plan and execute.

00:03:22: Yeah they can do execution work that's digital work okay?

00:03:27: And then there is the embodied AI.

00:03:29: people think if it's not physical but you cannot touch something and live things have to be physical.

00:03:37: Now, there's a third set of people who says AGI is when they have consciousness in feelings an understanding of itself.

00:03:46: but my personal feeling is that the first definition is already enough.

00:03:51: we don't necessarily need the second or third two for there to be AGI right now.

00:03:56: if you look at humans in general, our biggest advantage over animals and machines is used to be our cognitive capabilities.

00:04:05: And if it can do that then you know in developed countries sixty percent of the workforce white-collar cognitive workers.

00:04:12: So,

00:04:13: AGI is more or less it can work?

00:04:15: It can execute to... Yeah

00:04:19: and has long term memory.

00:04:20: right.

00:04:21: this is very important.

00:04:22: yeah today the memories getting better but it doesn't have really fully mature longer memory.

00:04:29: so uh and then I guess one other thing that people say some says i'm not is that it has ability to learn and improve itself.

00:04:36: Okay.

00:04:37: Yeah, and then we see much more on what is very important.

00:04:40: productivity.

00:04:41: because in the beginning said okay?

00:04:43: We have F AI as fun We have it for birthday parties that you haven't a speech or something like that.

00:04:50: You learn a little bit you write articles or something Like that.

00:04:53: And now The next step Is productivity and That's Very Important For The Stock Market In Everything.

00:04:58: Because If You Have Much More Productivity if replace jobs with AI, there will be a boom.

00:05:07: A next-boom?

00:05:08: Yeah so I guess that could be a boom in some areas but i think overall um they're actually could be uh destruction of value in the economy.

00:05:19: yeah you know as i mentioned earlier though white collar workforce wage is around twenty to thirty probably closer than thirty trillion dollars globally.

00:05:28: if your able to replace And somebody is using a twenty dollar-a month subscription to replace a hundred and fifty thousand dollars per year human labor.

00:05:39: That person's not getting the money, there may be savings from but they lose a customer long-term.

00:05:48: If everybody does that and you lose fifty, sixty percent of your customer base the overall market may actually shrink.

00:05:56: It also makes cognitive labor deflate in value in the sense if I used to pay a lawyer five hundred dollars an hour or thousand dollars now to write contract.

00:06:05: Now i can put request into software.

00:06:09: seconds or five minutes.

00:06:10: get a contract back that I paid nothing for.

00:06:13: then the value of that labor goes to almost nothing.

00:06:17: Right, so I don't know if it's a boom in terms of economic boom.

00:06:21: There may be a boom In terms of service like people will be able to get cognitive services that they couldn't you know If you were an Africa.

00:06:27: You probably can find the lawyer now?

00:06:29: You can have you and give your uncertain South America.

00:06:31: Maybe you can't find a doctor of a certain type.

00:06:34: Now you have a ability to do remote diagnosis.

00:06:37: So quality of life can improve.

00:06:40: but i actually feel Like there is potential for the nominal GDP to actually drop.

00:06:48: The problem is of the future.

00:06:50: We do not know exactly what's going on in the future.

00:06:52: It's better with a past because we can explain it in a better way.

00:06:56: Yes, and the future is right now is SpaceX is entropic is Open AI And they will be on the market and there.

00:07:06: There is a lot of money in the market that you warn you want that people can lose a lot of money, and other one will earn a lot more.

00:07:17: Can you explain it?

00:07:18: Why do you want...

00:07:18: Yeah so there's big difference between this set of IPOs coming in over the next few months with Anthropic and OpenAI and SpaceX compared to the past.

00:07:29: One is just scale In the past, you're looking at hundreds of millions of dollars.

00:07:34: I think Amazon went out for about five hundred million and they raised five hundred billion.

00:07:41: Actually no...I think five-hundred million!

00:07:43: And they raise fifty million right?

00:07:44: You have companies like Metta which are raised tens of billions but now a company going up to two trillion raising seventy-five billion.

00:07:58: Seventy-five is more than all IPOs in the last year put together, in terms of new raises.

00:08:05: In fact if you go back to the dot-com bubble people talk about that always like a big numbers.

00:08:11: there was two hundred or so IPOs and together they raised twenty four billion dollars.

00:08:17: well That's one third what SpaceX is going to raise this coming week.

00:08:24: And what's more important is that they now have changed the rules for being included into the index, the Nasdaq-I as well as the Futsi Russo one thousand and those two represent over twenty trillion dollars combined of passive money.

00:08:41: That will be forced to come in and buy proportionally this stock which essentially means that it will transfer the cost of this IPO, or exit of the IPO for all of these billionaire investors to pension funds and retirement funds.

00:09:04: And mutual funds are retail investor but they didn't have a choice.

00:09:10: It used to take a year.

00:09:11: That's the buying program more or less, yeah?

00:09:13: Yeah it used to be a year to get approval to become part of this index now takes five days for The FTSE Russell and then only takes fifteen days four for Nasdaq.

00:09:23: And this is how nasdaq was able to convince SpaceX To IPO on NASDAQ versus On you know other other stock markets in New York Stock Exchange right.

00:09:35: so This I think very dangerous thing.

00:09:38: because because the reason we used to have... a period of time is what they call seasoning or price discovery, right?

00:09:47: Because you understand after about a year.

00:09:49: Is this company worth it?

00:09:50: Are the legitimate?

00:09:51: are trying to cheat you?

00:09:52: but if you only have fifteen days...

00:09:55: Okay!

00:09:55: You've got

00:09:56: no chance.

00:09:57: ...you haven't even seen an annual report, quarterly reports and all these mutual funds will have to spend billions and billions....

00:10:07: And you have a story machine on top Elon Musk's Story Machine big one who's always very positive.

00:10:14: He

00:10:15: is the best marketers in world, that promotional leaders and I have a lot of respect for him as person but i had to say this fast track, essentially fast entry rule that is being applied.

00:10:28: It's a very dangerous game to play and now it has opened.

00:10:32: Anthropic and Open AI will use the same rule to do this thing for the next group of retail

00:10:39: investors.".

00:10:39: So your prediction was first stock is going well?

00:10:45: A few days or weeks.

00:10:46: then when the first numbers come out with quarterly reports.

00:10:52: Well, my

00:10:53: quarterly reports will be another ninety days.

00:10:56: But then at sixty-days the lockup ends.

00:11:00: so which means all investors cannot start selling their stock.

00:11:04: and as soon as the lock up ends because indexes are still forced to buy proportionately whenever more stocks available they have to buy a proportional amount.

00:11:13: So you essentially do direct transfer of money from average investors into venture capitalists.

00:11:24: But right now everybody's using more and more AI, so I think normally you would say if everyone is using AI that they have much profit and will get rich especially this AI models?

00:11:38: If the only provider of AI in the world yes absolutely!

00:11:42: That could be a case because there are monopoly.

00:11:45: The problem as an industry has about ten or fifteen labs all within percent of each other, and half those labs are in China.

00:11:55: Those labs have open source.

00:11:58: They essentially free to download.

00:12:00: then you can run it on your own computer or server if use their service do.

00:12:05: inference is maybe one-tenth at the cost per token as the frontier US model.

00:12:11: so that open router report that just came out is showing that thirty to forty percent of traffic, uh...that are being routed.

00:12:20: it's going through the Chinese models.

00:12:22: One year ago It was one per cent.

00:12:24: Yeah, so it's essentially gone from one percent to forty percent.

00:12:27: And if you look I would think that within the next six months they'll probably go through sixty or seventy percent.

00:12:31: because If You have a choice To pay you know one cent for something versus One dollar For Something and It's Essentially One Or Two Percent Difference in Capability Most People Will Choose to Pay The One Cent.

00:12:43: yeah That's For Sure!

00:12:46: Paradox.

00:12:47: No, no so jevins paradox is something that essentially the idea Is that the cheaper?

00:12:51: Something gets to more people you use it.

00:12:53: yeah and That does seem to be true So far.

00:12:56: but what people don't realize is if your losing money on every token Right?

00:13:03: And more people use it.

00:13:04: You actually lose more money, so OpenAI Anthropic and SpaceX or XAI right now they are losing respectively three billion, three-and a half million in four point two billion dollars per quarter.

00:13:18: Their traffic is going up, but they're losing more and more money.

00:13:23: And this does not even include the true cost because doing all these capital expenditure that they are amortizing over four or five years so real spending their putting in a much bigger number than that.

00:13:37: So for you there's no IPO already?

00:13:41: Well, it's not even me.

00:13:42: If you look at Sarah Fryer who is the CFO for OpenAI she publicly said that financially they are not in that position, but Sam Altman pushed her to go IPO.

00:13:59: And then she was interviewed and said my job is create the maximum optionality for a company.

00:14:04: so I'll do what i am supposed to do.

00:14:07: But she publicly says She does not believe That The Company Is Ready To Go Public.

00:14:13: And if you put this machines, these engines into a body.

00:14:17: You have embodied AI and This is the next big market isn't it?

00:14:22: Yeah

00:14:23: I think your robotics in the integration of robotics with AI Is definitely a big market.

00:14:28: um i don't know If It's going to come as quickly As some people predict because you Know elan has been talking about it for about probably three four years and yeah He's saying that they're gonna you know A few years sell a billion Robots a year.

00:14:40: Um optimistic, but you know he's been known to be optimistic about certain things.

00:14:48: But it is long run.

00:14:49: I believe that his right and i don't think its going as quickly.

00:14:52: coming back from China You just came Back From China Yes!

00:14:56: That was very impressive

00:14:58: And visited some robot company.

00:15:01: so maybe you can tell me what your thinking.

00:15:04: I was really, really impressed because i've seen formerly times about used to box a little bit some exercise doing some exercises and then i see some robots boxing.

00:15:16: And the astonishing thing is three years four years ago you pushed them.

00:15:21: they fall down couldn't get up in a second, more or less than seconds.

00:15:27: they are getting up and fight again.

00:15:29: And I was really astonished.

00:15:32: what's going on then?

00:15:33: i've met your friend there.

00:15:37: He is the CEO of Unitree?

00:15:38: Yes!

00:15:39: They're already building thousands-of-thousands robots... ...and mass production because they have so much competitors in other regions Galileo and other things, I have never thought that they are already producing it.

00:15:55: And then i found somebody... ...and they're producing hands!

00:15:59: And he told me.. ..that they've gotten a case study from Europe or something like that.

00:16:06: They had to touch the product with their hand.

00:16:09: That's fake or real?

00:16:12: Normally my hand is best but you can feel it But its not like this.

00:16:18: wonderful, great job.

00:16:21: Yeah yeah I mean there's over a hundred robotics companies human rights robotics company in China and some ways i feel like The problem with China is because it's so big, whenever there's a hot topic everybody rush in to do it.

00:16:35: So I think another three or four years are probably the five left right?

00:16:41: A lot of these companies will not survive but because they're so many competition creates innovation and speeds up innovations.

00:16:49: this actually why America may be as far ahead that they might have been particularly on physical side.

00:16:57: even today.

00:16:58: American robots, you know, eighty or ninety percent of them depend on Chinese parts and Chinese actuators to operate.

00:17:05: Right?

00:17:05: So if we look at the Unitree R-One robot is only five or six thousand dollars And it's maybe four feet tall.

00:17:15: It's a little shorter robot but for some use cases that already pretty good.

00:17:21: the US robotics, or the Boston Robotics models.

00:17:28: Those are...

00:17:29: Korean from Hyundai right

00:17:30: now?

00:17:30: Well it's actually a Korean company, Hyundai bought American Company but they cost you know hundred to two-hundred thousand dollars.

00:17:38: and so such big difference.

00:17:41: multiple X. I went into their factory few months ago and its impressive!

00:17:46: And one prediction touched me a lot for this CEO that was in three years.

00:17:51: In three years, you can buy such a humanoid robot and tell him what to do.

00:17:57: And then eighty percent he will do it not one hundred.

00:17:59: He won't fix my bike or something like that because its very difficult but eighty percent of the other things I give in an order and he'll do it.

00:18:08: Yeah!

00:18:08: Well...I believe thats possible.

00:18:11: The good thing about robots is once they learn something then they can just download it and have a new skill.

00:18:19: This is very different than humans, every one has to be trained

00:18:22: individually.".

00:18:23: And you know that people are scared when I'm coming back from China.

00:18:31: what did you see there?

00:18:35: People say we need to stop this or do something... You call it the three point five rule.

00:18:44: Yeah, so there was research that came out in twenty thirteen from Erica Chenoweth who is a researcher about social sciences and she went back to look at the last hundred-and-twenty years.

00:18:56: There were over three hundred mass protests And whenever it was more than three and half percent of the population participating activates activated.

00:19:08: There was essentially a change in the regime or it's positive decision and these are all nonviolent protests.

00:19:15: In fact, the nonviolent protest was overall twice as effective as the violent protests.

00:19:21: so I'm not advocating for violence but if people have concerns about speed of progress on this technology they need to voice themselves.

00:19:34: there is an opportunity.

00:19:37: I don't think it's about technology is bad.

00:19:40: That's not really what i'm saying, but how are humans applying this technology?

00:19:46: How are they doing in a way that preserves the dignity of people and gives value to an average person versus giving the value to individual owners or DAI.

00:20:06: into a global public good versus turning it in to a weapon or turning

00:20:14: though there's some protests, some protest in the streets and universities.

00:20:19: And that is one who was very famous.

00:20:21: it is The Pope.

00:20:24: Yeah!

00:20:24: He says AI must be disarmed.

00:20:29: What does this mean?

00:20:31: So I actually read his encyclical.

00:20:33: It's a very long I think its forty five thousand words or something like And I actually was impressed for somebody coming from the clerical space, to write about technology and so eloquently.

00:20:53: The main meaning is that this technology is useful, but we have to use it in the benefit of humans.

00:21:01: And it's not a replacement for humans and will be an independent being.

00:21:08: so this why its little bit controversial?

00:21:11: because at same day he brought in Anthropic also speak about the same event.

00:21:18: perspective on AI is that AI will become an independent conscious being and they are birthing the next species of living animal or a living being.

00:21:29: So, so they actually have... On that area, They had very different view but it came together about safe usage for positive value to humanity.

00:21:40: So we have protests, people who are very scared about it.

00:21:44: and here in Germany.

00:21:45: We have a new book from Katarina Zweig And she says does AI know that it knows nothing?

00:21:54: And so addressed to the people said okay AI is not... It's basic!

00:22:00: Nothing you need to be scared of.

00:22:02: What do think about her title?

00:22:05: Yeah I mean I respect her as an author, and she has a computer science background.

00:22:13: But she comes from an algorithmic area.

00:22:16: that's more the symbolic side of computer science which is saying I make the rules, I execute the rules right?

00:22:23: So it was algorithm-based AI that she's familiar with.

00:22:26: but what happened over last ten years?

00:22:28: really more to connectionist theory of AI essentially training in having learn versus telling it specific rules.

00:22:38: And if you're, its a rule-based system I agree with her.

00:22:41: that is executing based on the rule doesn't understand but they say learning systems.

00:22:46: to be honest nobody right now knows whether or not there's some level of an underlying understanding or appreciation thought pattern underneath fit and I think we should leave the potential for that to be true.

00:23:04: And, for her to be so concrete about her views that it is not possible and cannot do something...I would be a little concerned

00:23:20: Yeah.

00:23:20: And maybe the systems, they're all becoming better and better every day, every month, every year.

00:23:25: so... I mean we talked earlier about this when you said yourself even when use it today You can see that It gives you a perception That seems to know what is right or wrong.

00:23:35: We'll tell you if your are right or not Right?

00:23:39: So i feel same thing.

00:23:41: In fact When i work with I am constantly amazed in terms of, when I give it five or six different reports to say find the commonalities between this.

00:23:53: Where are these reports missing things?

00:23:55: And for it to be able pick out specific points

00:24:03: I've done some research about Gavin Newsom, the governor of California.

00:24:09: And i remember that he has a new law and no company in california can lay off people because of AI.

00:24:17: but it was wrong!

00:24:18: The system told me this is not only an idea... ...it's something he brought-in.. ..and its like a radar system which is echoing what's going on with the economy.

00:24:30: so the system told You are wrong, you have to look in another way.

00:24:36: And that's really interesting.

00:24:38: I've never had this experience before.

00:24:40: Yeah and then we talked about earlier.

00:24:42: but how do use it is also important.

00:24:45: if you create a system be my system make me feel good Then they will always kind of agree with you.

00:24:51: But when start putting the custom instruction tell me truth.

00:24:56: Tell me multiple views on any argument It actually can give your right.

00:25:01: end wrong and point out issues with what you're saying.

00:25:04: And I think most people need to change the default settings on their AI because a default setting is to be helpful, and to be honest ,and to be friendly .And sometimes being friendly it's doing too friendly.

00:25:17: Because they want to make you feel good right?

00:25:20: Nobody likes to be told that they are wrong but i think we need tell the AI its okay.

00:25:26: So we talked a lot.

00:25:27: We talk about SpaceX, the Pope and people who are against it in favor of it.

00:25:33: And what is our main message for this show?

00:25:35: You

00:25:37: know I think one thing that i would like to kind of lead people with Is That People Think Where In The Beginning Of A Massively Growing Economy I guess new revenue sources.

00:25:54: Actually think that the SpaceX IPO and The Coming Anthropic and an open egg IPO is the beginning of the popping up the AI bubble, And now it's maybe a little bit controversial or counterintuitive.

00:26:08: but when a company goes public what happens?

00:26:11: Is at their books become publicly available and viewable.

00:26:16: And a lot of the ways that revenues are being accounted for, and costs have never been seen by the public.

00:26:24: I think very soon they will be.

00:26:27: The economics in this industry right now is not sustainable.

00:26:32: But we have had already this discussion, I think months ago that we have a bubble and AI bubble.

00:26:38: And everybody said Jensen Wang no there's long way to go.

00:26:44: until now everything is fine.

00:26:46: if you see NVIDIA in all these big companies the revenues are growing more than expectations.

00:26:54: but Next year, next month there will be a bubble.

00:26:58: That would be well.

00:26:59: now I think There is the bubble today but the popping of the bubble we'll take A few months.

00:27:03: i would say by The end Of the Year it Will Be very clear.

00:27:06: and here's an example you look at xai or spacex.

00:27:10: They used to be in AI company.

00:27:12: Now what they've become then last Few weeks Is now they are a hyperscaler?

00:27:16: They're just renting out their data centers.

00:27:18: now yeah To google and and to anthropic And Their Collecting a Billion Dollars each per Month from them.

00:27:25: so essentially All they did was, they built this data center and then the renting out.

00:27:30: If Elon really thought that AI was going to make money why would he take his most valuable resource?

00:27:36: That thinks he has an advantage over everyone and rent it out because a few billion dollars to him is nothing.

00:27:42: if you can make trillions of dollars in revenue The fact even right now spending time making these kind deals means that he does not believe that AI will bring them the money to justify his IPO.

00:27:55: There are interesting times, very interesting times and so we will catch up with everything.

00:28:00: We'll ask a lot of questions.

00:28:03: you give us wonderful or interesting answers And our show is over!

00:28:10: We need to celebrate this first time being face-to-face on the show but we would be back again here in September.

00:28:17: Yeah...we'd be at Big Bang AI Festival and have just met some people from government.

00:28:25: what's going on in the world and we have wonderful discussions here right now.

00:28:29: And with our people who are watching us, please send us some emails ask any question because it is getting more interesting especially for working as a journalist to see We are in the most interesting times ever.

00:28:47: Yeah,

00:28:47: we're at a very interesting time and also it's important that we have to take action so as not just let our future play out by itself right?

00:28:55: So you know...we talked earlier about the three point five percent rule.

00:28:59: It is time for people to exercise.

00:29:02: at three points five percent.

00:29:04: We get the future that we deserve.

00:29:05: To

00:29:06: stand up, to say something and well...to be clear!

00:29:09: To know what's going on and not guided by fear Not guided by anything else only by information And to be aware of it because we have very positive times to a very positive developments.

00:29:24: And well, it's dual use.

00:29:26: there are negative developments too?

00:29:29: Yeah and I think the government they're actually starting to wake up and want do the right thing but don't know what to do.

00:29:35: so we have give them guidance.

00:29:38: Thank you for being here.

00:29:40: this was The Big Bang Tech Report.

00:29:42: We see each other in few days.