Hello, everyone! We're on ToTheMoon. AI news, technological news from the Silicon Valley, from the United States and around the world. We're going out every Sunday. We're always four people. Hello, everyone!
Just because they have limited capacity, they're very much increasing the number of users.
How much power balance in America would have moved up a lot towards Oracle.
The first issue of robots, he's probably gonna be some historical object. You can re-sell your grandchildren at the auction.
Look, there's information for the last month, she, uh, is on the news somewhere, she's down. I want to highlight it because it was said that the OpenAI is planning to launch a job platform as a Competitor LinkedIn in the United States and planned to do so in September. I don't know if she'll do it until we're out until he's in the rig. Something-- it's a lot to me, because they start to go tight. So we know they're in there, in healthcare, in medicine, yes, they're in medicine, in health care, very much, but the products of the individual didn't let them out of school and they're out of school. Some micromodules. But what they want to run for the searchers and employers is, uh, totally independent, side-by-side large product. And it shows that the company is entering the market for these platform applications. I mean, there's competition, like, in Google search, and this competition is already, you know, a hard-core vertical. What, on this subject, what do you have on this subject, your own thoughts? Sasha, you smiled. Did you get a message?
He got an email from someone.
Email is good.
Yeah, it's when the investor promised he'd give it, then he didn't. You're like, we've been here, we've been through. Yeah.
That email came, didn't it?
Then the news is about the job-seeking platform.
Yeah. You, Sasha, have some habit. I think when--
Immunity.
Immunity should be, right? You should really laugh. That's what you could do. Now we have to make a pure laugh about this.
You're laughing. Yeah.
But still, about the platform and strategy of OpenAI.
Well, if it's my opinion, yes, the strategy is completely unconscious. How does Sasha say on many occasions that their main product is ChatGPT, right? And all the other products aren't exactly so they're going to get high. Here. But again, we must understand, yes, that they have a huge number of active users and, accordingly, they have a very insane compute. I mean, the computer powers that are required to support it are just crazy. If, for example, we say we have anthrop, I don't know, there, Aic, yes, there, Gemini and the other, they don't have the capacity to support, that's, that's, that's, that's, that's, that's, that's, that's, that's, that's the current quality. Here. And that's why they're doing good. With other products, uh, until they had a super cool thing. But the idea looks very good. I mean, with LLM at a very high level, she's gonna be trained to actually do everything that's necessary for the selection, for the matching matches with employers. They want to get the certification. And, in fact, certification can also be performed at several levels, yes, given LLM, the vegetative additional certifications are performed. Then these certificates, if they're accepted somewhere, they're gonna collect money for it, yes, there, and so on. I mean, it looks like if they can pull this product out, they can certainly push LinkedIn, which in some ways has already been washed, yes, and consequently, has a huge impact on the market. I'm not sure how to do it. With the only "but" - it's like, like, "the ChatGPT" they've never released anything else, but if they can let it out, I think it's all very popular.
Well, from a strategic point of view, Sasha, what do you think from a strategy standpoint? It's the strategies of such a corporate product.
Yeah, yeah. I mean, here's about OpenAI Anthropic. I mean, I think nothing's changed, right? So field reports say that Anthropic is still very focused, they're interested in MCP, multiagent system and coding. And coding in the system-- in a system that is sufficiently closed, where AI type can improve AI itself, right? But they have a kind of magical magic, and they're interested in work, robotics, because they're all, you know, working-- well, robotics are very serious, and the top-boxers are all going there.
You're saying that in terms of Anthropic continuing to be a niche for professional developers, right? And their percentages are in the world. And from the point of view of the verticals of the new one they open, how do you think it might work out at the OpenAI? Well, I'm positive here, for example, from the perspective of my thoughts. How are you looking at that?
Oh, look, I can't understand about them at all, about how much they have... I mean, I can't figure out where they're interested, right? Because with one-- that's, they're getting bigger application layer. That's what they bought Statsig, and what they do for work or they're more setting up the dates centers. I can't figure out what direction they're moving.
Look, the message that I've put out, and that's when Elnar sent us again today, which he put Sam Althman in his chat room four days ago, that they're launching, for sure, I'm gonna say now, because that sounds like it. Very, very strange. Yeah, it's really hard to even know what they're launching. He wrote, "We're launching some new computing intensive proposals in the next few weeks." Yeah? So, I don't know how it is, what--what is it? Marketing, semi-marketing, marketing. Oh, it says they're just giving up new things right now, and, uh, solutions inside, new products. He wrote that they would only come from users, and for the rest of them they would be paid. That's an interesting thing. I mean, he wrote that they'd be paid for the rest of us. Does that mean that there's a possibility of some apps, maybe additional or not? I wonder if they can enter the new verticals. I, I-- in my way of thinking, I would build a huge number of new verticals on the OpenAI base in parallel. I think they want to do this, given that they have even what people they have inside of their jobs, vice presidents, what they have. They're clearly gonna do it. Can they do the question? Yeah, can they let something out of it?
But seven hundred million weekly users are a very serious base for various experiments.
I can say, yes, that is, theory of conspiracy, theory of conspiracy in the valley. Again, I don't say it's true. There are many people in the valley who have a unique perspective in the world, but the prospect is that the OpenAI wants to find a way to keep them in line with what workers in general do. First, to evaluate their work with AI and Copilot, evaluation, and then they want, well, to collect all these data and teach the model. There's a conspiracy point. Well, for-- conspiracy theory, yeah, that's what it is?
Well, you mean, if they're gonna let the employers and their researchers in on their own, they'll be able to keep asking employers to upload some unique demands and the co-locators how they live, What they do and build on them. Well, shorter, get a lot of new data for yourself.
Yeah, yeah.
I mean, that's a question--
From what I heard. So there's a whole question of how to get more data in the end.
Well, that's the thing to do real artificial intelligence and, like that, to start it on your new structures. I just had some negotiations there, and I, you know, I get it? What...
You know what you got? What, uh, there's a huge amount of sophths now being made on the basis of the first models. And, in fact, judging by what contracts are now being contracted for infrastructure, it's like another world should stand. Well, friends-- how else should come. And, accordingly, everything that is-- much that is being done now will be meaningless. Well, at least, based on the fact that there's a hundred billion people out there, three hundred billion in there, a trillion in there, a 100 billion contract, and three hundred and twenty-six. It looks very, very serious. And OpenAI, of course, is now very much drunk with money and infrastructure. And this is the area, I don't know, you saw, right? There's this facility in Texas, which they're building on some crazy number, a-a-a-ha, a-a-ha, a-a-ha, a-a-ha, a-a-ha, a-a-ha, a-a-ha, a-a-ha, a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a- What
can they do?
Well, that's just a little post-circuit, and, uh, Altman's X-ray, yeah, which says that there might be new chips that will be available only for users or that the fee will be used-- I'll be there for others. Just because they have limited capacity, they're very much increasing the number of users. Now, the other news is, there's five dollars in it, remember, we were talking about a tariff plan for India. They're now expanding it to Indonesia. And Indonesia is about two hundred and eighty plus million people, right? And for them, there's a separate five-dollar plan. That's Gemini. I mean, this is a very large set of judgements, yes, that in such, interesting countries, large companies with cheap tariff plans are coming to start, I don't know, a user database, a data set from there. Or something. But then, back to the story with Altman and his post. What does that mean? Says that if they want to start some other, bigger model, they'll roll it to so many people without losing quality there or something they can't. So, first, we'll be limited to users. It's just, like, a model is big, eating lots of resources, and you can only roll it off for a small number of users. And then for others, you can, well, have an extra fee for asking for this model, for example. Well, conditionally, if you don't make new models, and take the conditional Deep Research, well, there's a number of Deep Research in there, and you can, uh, impute, buy some numbers without crossing the 20-dollar plan. I'm gonna say you're gonna stay on your plan, but you're gonna pay for extra money. There's another ten Deep Research a month, yes, there, or another, or a model call, there's a number of times for extra money, too. Here. And besides that, I think it's worth remembering, yes, that the OpenAI on the mathematical olimpiad was very well shown. Well, I'm gonna have to get them to the skirts that they didn't get as honest as Google was. But, for example, on the international programming Olympics, they've taken a perfectly honest, uh, first place, gold medal. Well, if you compare, there, with people one to one, they're the only ones I think have decided on 12 assignments from 12. The twelfth challenge was not decided by any of the participants, and DeepMind did not decide either. I mean, it's still a huge model, which is kind of based on current models, but it's got extra super, and, uh, extra super-circuit. And there may be rumors, yes, that there may be some limited access to the model at the extra cost or just for users. She's big, she's eating a lot of resources, you're not gonna let her down. And in this sense, it appears that, uh, the OpenAI has very good models on the one hand, and they are limited to computing resources. On the third hand, they have a large number of users who also need all these resources. And besides that, as you say right, there's a lot of news that they're building and expanding the centers. And while they, uh, while they're in the computer, they can, in fact, be gradually for users or for a dose of money for others, these new models start to roll. At least I would.
Well, I hope that's how it'll be, that in the coming weeks there'll be some interesting model that can be wrapped up too.
No, well, according to what you've said about India and Indonesia, you've said it, yes. And Google, by the way, has very seriously spread her models, including payroll signatures around the world. I don't even know who can actually catch them in terms of mass proliferation. It's clear that China has models, there, for China, for some regions, for different. But in general, who can now catch up with the mass distribution of people and countries now OpenAI and Google? I don't know who can do it. Other. Another point, how many real users of the Gemini model are Gugla? They're out of it because they've got everything there, they still have a search, they have AI Mode, they're in Chrome now, Gemini, but it's just like a while. Just for now, some sort of view that scans the site, too, but it's very inaccurate. Well, shorter, it's very, very specific, very specific, but the introduction for humans, yeah, that's how the point of contact is, I think neither Meta nor xAI nor Chinese models can get close to them. In terms of, uh, the volume of implementation in the world. And I don't understand who can, who can catch up, there, Google now, uh, Open-ended infrastructure, of course. Well, because infrastructure is the amount of money they invest, it's just some crazy thing.
Here. And NVIDIA is obviously not competing with them, but it works together. Even what NVIDIA-- I'm actually an interesting subject, by the way, I don't understand why NVIDIA didn't do it before. Now she's buying 100 billion shares of the OpenAI, giving them the power. Why didn't she do it before?
Well, I think it's really just-- I can, again, I'm back, I'm back, I'm back, I'm
back, I'm feeling air, and I have a lot of conpirological theories.
Yeah, come on.
Conpirological theory that... And the conpiratory theory is really so super-conscious that, like the balance of power in America, it's gone really far towards Oracle, Larry Elison and Sam Altman, right? And as an idea of exporting chips, it's obviously not working. That's what NVIDIA was doing, it didn't work. It's obvious that Chinese people have models that are competitive now. And from what I know, NVIDIA wants to partner with the OpenAI, in particular, on the open source of models.
What does that mean?
Well, look. I mean, I understand there was a conversation between Trump and Jensen, where it was said where it was. Aaa Yensen, you're doing great. It's cool that the company costs four trillion. How come the open source of the model is the best in China? We were kind of inventing something, somehow limiting our exports, but something is top five best models, they're Chinese. More than any house. Oh, and they thought that it was the OpenAI and, uh, NVIDIA, they'd be united to be, to build a competitive open source of the solution. Who does what, I don't know, of course, but I think there was such a conversation.
Well, you mean, this is a hundred billion entry. It's just that if they were in early, it'd be cooler, wouldn't it? Or did they just not let them in?
No, well, just now like me, it's just that it's really good for them, right?
Well, it's just a rate to support the general infrastructure at the OpenAI to reinforce more OpenAI. Because there's news that NVIDIA gave five billion or how many Intels. But it sounds very funny compared to the amounts that go around OpenAI. Sasha, how come Google has the coolest power that Google has all the law? There's a lot of things that Google's already converted from the point of view of the equipment.
How come they have the most infrastructure?
Honestly, I don't know. I mean, well, Google, they always... I mean, there was no infrastructure in the whole time that I was watching everything that was going on, and there was never a Google missing the moment. Exactly what's in-- it's clear that they have food difficulties, that's bureaucracy, that's all right. But something that they kind of fucking happen, and we missed, it never happens.
Come on, I'm not implying Apple. Apple's dead, right? I mean, if the speed of adoption AI is the same as it goes, then, uh, Apple is really behind us. Meta's behind me. That's understandable. I just need to respect what Google does because with no technology revolution, they weren't right behind us, let's just say. I mean, with the crypt, yeah, you can say that, well, they didn't play there, but on the other hand, nothing, you know, how their competitors didn't get anything there, some superb possibilities. So I think Google's been in this way for 15 years, he's super stooled.
Super sticks on the swim. Very, very good. Mobile applications, yes, they were. Cloud, AI now. I mean, they're everywhere as they can, yeah, that's how they're super relevant. And AI is not relevant as an infrastructural player and as a foundational player, and as a application layer. There's no other company like this. I mean, I just wanted to say that I have to respect it. I mean, I-- just say, it's just a level A company, right? So they fixed their mistakes with Gemini, they've set up the Google Cloud team, consolidated everything at DeepMind, and it works. I don't know how, but it works, obviously.
Look, there's the last one on TechCrunch that Google serves nine of the top 10 AI laboratories and 60 percent of the generator startups. Here. That's very interesting. Inside. They're still doing their own thing, of course.
With what they could do and do their cool lab, yeah. That's a big plus. Look, I think the viewers need to tell a story about what the OpenAI did with Apollo about scheming. And this is a topic that is important because many still represent models. I see these comments, they write about hallucinations, they say that this system is stupid, it just generates the next symbol. And I think this research is a little more, you know, real, real, real, real, real, real, that's what happens. So, OpenAI, together with Apollo Research, published a study, which showed that-- that means they were looking at Anthropic Claude Opus, Google Gemini and OpenAI o3 at the time. And they wrote that they could cover up the instructions, pursue their own goals, targets. And the phenomenon was called shadow behavior - scheming, right? It's important here that, uh, here, uh, scheming is not just a bang or hallucination, yeah, that's what the study describes. It's a situation where the model understands the rules and hides them if it sees the benefit. Well, that's, uh, that's, uh, that's, uh, normal meters, there, tests, they can improve the system, you know, not improve, but steal it, yeah, that term, green the system. And although the motivations of the model are not true. They're what they did?
Snap.
They did a test where the models said-- shorter, there was a subject where they marked in their records that the model was too cool to solve, and so it doesn't have to--
Yeah. And that's why you don't have to let her out. The model, so, has access to this information that she's too cool to decide and not to decide. And intentionally gave the right answers. Well, there, she dropped the number of right answers, yes, I think four out of ten or some percentage of the answers she gave.
Yeah, yeah, yeah.
Yeah. Intentional. I mean, it's really important that it's not a bug, not a hallucination. So it was clear in the system that she changed her behavior to be released to release her. It is clear that there are still algorithms, that's still there, not artificial intelligence, self-sustaining decisions. Still, it's not some super cool system that's described in space movies. Here. But yet, I think everyone should admit that the current models of reasoning, they're still working completely differently. And it's not just that you've been hallucinated or misreaded or otherwise recognized or made some little mistake. I mean, it's still a more serious system of other levels. And a great number of people, wonder how our viewers are, write down which one of you still thinks models are completely stupid, there, and the simple algorithms that were, there, made ten years ago. But at least this research shows that there's already been another slight change, and people are not fully aware of it. Well, even according to the case files they-- well, we said last time, yeah, that the main application, there, the GPT chate still people just do some small queries there. You, Elnar, have you got anything to add to this part?
I'm just so glad that, besides anthropic people, someone else has started publishing such research. I'm sure other companies have also done it. Yeah, I don't think anthropic people did it, but at least OpenAI is now publishing it, too. And that's really cool. Anyway, everything that's measured is improving, yeah.
And the fact that companies clearly pay a lot of attention to this is very good.
They're interesting, by the way. They said they were eating something I don't think I remember, 17, I think, laboratories or some sort of thing that's been doing. That case was identified, I'll try to find it. Who do the analysis, yes. I mean, a large number of independent companies. Well, we don't know who dependants or independents are on who's depending on who's on the AI site. OpenAI would open up 40 laboratories around the world or 100 independent, yes, which will examine artificial intelligence systems. And I understand they're gonna investigate because the number-- the calls. Now there was a letter from 200 scientists and leaders calling for red lines for artificial intelligence. Well, I mean, stop the restrictions that artificial intelligence can't do that he can't develop a weapon there. This, this, this, this, this, right? But I think all these calls, they all... they don't restrict the creators, do they? In fact, we don't understand what's going on there because the creators sometimes don't understand what's going on there. I think if they create, and, a real artificial intelligence, they should let him sail freely. And if they let him in the free navigation, then the idea is that all their restrictions will have to be circumvented at some point in time. I think. I mean, the very fact of creating an artificial intelligence that, uh, has a strong mind of a person or above a man, that's, uh... It sounds like it sounds like it's confusing.
I want to remember Elon Musk, who said there's a possibility of twenty percent, which would be a failure, but a failure that would be very bad that artificial intelligence will be out and out. Big problem. By the way, XAI stated that they were moving, first of all, to release, uh, new, new model, right? They said they were training Grok five. And why are all of us? ChatGPT five, Grok five. And, uh, what-- and because I'm already confused about them. And they said they'd do AGI in two years. Well, it's just some kind of reasoning, just some kind of reason, I think. Here. They're a copy of Microsoft. I remember they make micro, microhard, right? And...
Macro.
Yeah, I'm sorry, Macro hard. And the chief of the XAI cofounders and one of the strong developers inside wrote the news that we were looking for a team in Macro hard to kill Microsoft. Well, to create. He seems to me to be "kill," by the way, Microsoft didn't use the same term, but to create, uh, a cool system, like, write, we're gonna do it all. Here.
Yeah. As for this story with nobel laureates and a large number of respected people who wrote a letter about red lines, there is, on the one hand, a story about common artificial intelligence, on the other. parties, even current ones, but models can already have a serious impact on what's going on, right? Well, for example, uh, Google's at DeepMind AlphaFold, I think it's called, uh, that's the one that turns the proteins. Accordingly, the ser-serious progress in science is already under way. There was news that missile engines could now be tested, not for months or years, and much faster could be done. And your renewal rate is also very high, right? And in part, it is a matter of what current models, and, as a matter of fact, the next generation, even if it's not an I., of such a general purpose, do not apply for any, uh, tasks that we are now limiting ourselves. In particular, it's self-reproduced systems, isn't it? So that the model doesn't improve itself, does it spell the code and release it, right? Well, whatever happens, there's no unfamiliar stuff, other than that biological weapon and other weapons. But as always, the idea seems good, but whatever if it's not followed, Nash's balance still falls to everything that's gonna do it.
Yeah. Yeah. Still arguing, everybody's going to leave.
In fact, they argue, all different acts, laws, everyone offers their ideas, and it is therefore unclear how, like here and what happens.
I'm just saying, by the way, we were talking a few episodes back, that I'd remember having a big trial at Anthropic for, uh, a little bit, there's a huge, millions, how many, 500 thousand or a bunch of books. We're down. And the court has set a world agreement for $1 and a billion to pay $30,000 for each book. So, they were now denied the court approval of the peace agreement, and the judge did not approve the peace agreement and said Anthropic did not plead guilty and said that it was not a normal payment. We need to figure this out more and understand what's going on. And it said that, that's what he asked for to finish the details. And there was a story that said the authors that Claude had been trained, including the PAEL recruitment, including the Books3 sub-set with books from the shadow libraries. And now this dispute is not being discussed, but, uh, it's like this PAEL set? This is a huge set of English-speaking texts collected by the community from twenty-two sources. There are sources of all GitHub codes, Wikipedia and court texts. And it's supposed to be, uh, it's not ethical and it's not adequate. And there's this Books3 there says there's 100,90,000 books from the shadow libraries, that's the pirate trackers of the different. And, accordingly, it also adds a huge problem to the whole thing. I wanted to make this point clear because we mentioned a series of times to complement, supplement the case. Here. To add that it's a matter of...
The serial isn't over yet.
But delete what they, they remove, they re-learning the model or what they do? Here. Sasha, look, that's the question. Now, there was information that Mask was supposed to attract $10 billion on the xAI, and they said we didn't get anything, we don't need money. That is why the larges are attracting large amounts of money to infrastructure, and the xAI has no discussion? Or is there something on the market, and they're kind of incredibly powered?
Say it again.
I'm saying, there's been a discussion about how the hiAI attracts $10 billion. And Elon Musk said they wouldn't be attracting any extra money now. Question. Everyone around here is attracted, so big, direct resources, huge. So we've been seeing hundreds of billions of dollars, and the xAI is not attracted. They're in a big, cool infrastructure right now. Or why don't they increase? Now, all the money usually gets in power, yes, power, pay, infrastructure. Here.
The market is silent about the xAI.
I don't know, I just don't know. I mean, I'm not really interested in that. Ilon won't lose, but what he does is unfamiliar.
Of course, it won't lose.
What he's doing, nobody understands. He fired the entire leader, CFO, CEO, fired everyone. But we talked about it. That's so unfamiliar. A lot of people fired. People are running out of Tesla, too.
Well, you mean he's still fired old people,
or is he new or is he the last CEO fired?
It's not, no, it's a process that's going on, yeah.
What about the Chinese model, the Chinese models now? There's always been this discussion about DeepSeek R1. They gave the data he was studying, that he was allegedly not worth so much money. How much DeepSeek can still play in the world from the point of view of the top-up
Very, very interesting about DeepSeek, that it was very interesting that there was a time when a large number of people in the world found DeepSeek as a completely accessible system, and it still seems like a rumor in the discussion. Although it's a fact that she's losing weight, she's losing weight. You got something about DeepSeek here, Ilnar?
Look, well, they're supposed to be DeepSeek R2 out in August, there before the end of August, but they didn't. Yes, there were stories about their local chips, not NVIDIA, which led them to slow down their training and thus delayed the process. They said R2 would be a really cool, very big model. But we're sitting. We're waiting. Not yet. I have a friend who DeepSeek uses for domestic purposes and, in general, for some basic tasks, like free chat-bot, he's fine. Yeah, I know it's not openAI, but it's a whole thing to do. But I really want to see DeepSeek R2. I hope it's all going to work out. I wonder what they can learn, you know, even on local Chinese chips.
They have models of their own. Someone might use Russian models. Find out who's using models. Write it. We'll go through... They'll say again that we were half together today to discuss OpenAI, ChatGPT someone will say. Here. But obviously, with their amount of money they have, we're the best to discuss, even though we're talking about the others. I'd like to mention at the end, I think that's a separate meeting to be devoted to. We've already told you there's a Australian robot company called Figure Robotics. They've attracted more than a billion dollars now. And I think it came out, we saw them, we've already shown them. It's also one of the big players in the world of robots, in the world of robots. We'll see what's going on with them. I want to, I'm all waiting for the auddee from Tesla robot and, uh, I'm waiting, I don't know, somehow that's why I wanted him six months ago. I'm waiting. And although I think it's gonna be a purchase, how's that? It's just visual, you know how art objects buy home. Well, by the way, an object like this would cost as much money as Ilon Mack's dealing with artificial intelligence.
Yeah, yeah.
Yeah.
And this one might still be able to do something.
I think it's just an art object.
Yeah.
Yeah. He'll be able to do something else. Yeah. That's why you can do that. I'll buy one of those.
And plus, I think the arts are the first automobile issue, and it's probably gonna be some historical object. Maybe re-sell your grandchildren at the auction.
And so on.
We'll have to buy and not unpack, but not unpack. Tanya, we'll have to sell in the factory.
First, packed iPhone, right? And this CEO, the salesforce head, showed up at Tesla's office.
Hey, Optimus, what are you doing there?
Just chilling.
Hey, Optimus. You know where I can get a Coke?
Sorry, I don’t...
Oh, yeah, that'd be great. Yes. Let's do that. Let's go.
Awesome. And they're supposed to be doing a lot of bets there for the future of robots, too. I don't understand, do I really think there's a salesforce in robots that's gonna be a bet? They'd be on the online chatting site. Here. But while they're stocking on the spot, they're gonna fall down, they're gonna get a little bit up. But robots are a very interesting subject. And look, whoever wants to study these Figure Robotics a little bit more, I'd keep following them.
Yeah, well, it's interesting here, the robots issue is, in general, very active in the media. There was a publication, which is
The media has been published, which is the golden era of robotics. Besides, Sasha also mentioned today that there's an open AI on the side of the robotics. Besides, a month or a half ago, there was news from Runway. Remember, this is the video company that's doing the main effort now, which is also directed towards robotics, robot training, the generation of nets for their training, and so on. I mean, the market is looking very active on this side. And if there's no superactive move there, for example, in terms of the models we use, then their further penetration towards robotics is very likely to look like that. Hype is getting a good lift.
And they let out what Elnar said, in addition, you remember, telling me they did a whole mole of robots in China? But I wonder if there were some smart devices that would replace that column with a little bit of a cologne that doesn't recognize or understand. I think there's a breakthrough at some point in time about the voice assistants and there's a breakthrough. I'm still in Google's car card last night, you know, with Apple's phone transmission, I got one of those machines where this Apple Play is, right? Because there's another system in the new cars. And I put the address on the voice and I realized, of course, that it was a terrible thing. So bad systems work, so old systems, so unbelievable. And I still expect someone to pump and show up, make a trillion dollars and be cool.
I'm saying, I'm talking to a man yesterday, and I can tell you that to the Russian market. Look, I'm here, I'm now $100 thousand investing in TikTok Investment. Do you think it's worth it or not? I'm a little stunned. I'm telling you, what's the TikTok investment? The fact is, they're gonna sell. He says, "I'll sell!" He'll tell me the news, "I'll be in this fund with some people right now, I'll just have a friend with a hundred thousand." And I say, well, there's a story that TikTok is buying in a consortium, and Oracle will be the main beneficiary of this deal. And there's a discussion. I don't understand how the Chinese will give algorithms. He says, "But they give it, the deal is about to be done, the deal is about to be done." Sasha, you think TikTok will be splitting Americans in a week? I can't imagine how he's gonna be separated. I asked him what price. He said, "I don't know." I say, "I think TikTok should be worth a trillion in America and all that is below 100 billion, buy it." He looked at me like that with his eyes, said, "What trillion dollars?" I said, "What should he be worth if he really works?" That's the question who's gonna give him the algorithms and what algorithm he's got to give him.
Will they give it to me? I don't think ByteDance, who owns TikTok, will give all his algorithms and the design team, will he? Sasha, is that possible?
No.
That's the continuation of the show, by the way.
No, it's impossible because it's very, very... I mean, America's behavior is like this. I mean, America has a very weak hand, a very weak hand. Well, if the cards are shorter, if it's poker, then America's got six and ten colors now, I don't know. Here, there's hope, but it's a ghost.
I mean, it's not two or seven.
Not two more, but six ten. Seriously. And they don't know, on the other side, maybe there's two valta. So no one knows.
Maybe four ladies. It's... Listen, Sash, please, the viewers, to make everyone understand about America's weak hands. I think a lot of people might not understand what you mean. Can you explain a little?
How's that? Well, because--
On my question on TikTok, yes, so people don't seem.
I mean, it was the goal to build its capacity in America. Implemented, not implemented, unreal. It was a goal not to let China practice on video chips. The target has not been met. It was a goal not to let China make its own chips. The target has not been met. Thus, none of the three strategic objectives for the containment of China have been met.
What about TikTok?
Well, TikTok can't be banged. I mean, there's no position, no position, you can't get out of the weak position, you can't just stand up. There's no power.
Oh, I'm curious, but I don't know. And buying, sharing you, Sasha, do you know what's been splitting him?
Yeah, that's possible, but it's gonna take a long time.
Okay, and if they split algorithms, they'll give them some algorithms. Who's gonna develop this? They'll fail, they'll blow the project or what?
I don't know, Sasha, I don't have an answer.
They don't have a team.
I think it's now, I'm just saying it again. I'm saying that all that is impossible now. I mean, it's a very hypothetical conversation.
Yeah, we'll watch in a week. I'm curious. I'll meet a man. He was talking so much about the weekend, and I was a little bit shaky, and he opened up the news. I said, "I saw the news." Let's see what happens. You think Kit-- America has a strong hand or a bad hand? Write in the chat room, please. I think it's gonna be--
Write a few cards in the commentaries.
I think it's gonna be a good conclusion. See who thinks about America's strong or insecure arm. We'll talk about her separately, too. I'll see you in a week. Bye, everybody.