Sasha, I love your horns.
It's not my horn, Tanya. I have, uh, 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...
From the horns?
From the horns, yes. And those, uh, made, uh, Indian style of undressed. And I was just lying yesterday, so I can't say it's a nice-- if-- like, I want to sleep with it. They're not that kind of intrusive, they're just hanging. And I remembered the idea yesterday. I have, because Igor, my son, found a deer rogue on the precinct some week ago.
Right!
Remember when you came?
Yeah, yeah.
And it's just that...
I offered to hang on the wall.
No, they're very good at this place because they're in Indian, Indian style, yeah, they're there, like they're painted, they're made of leather cords, something else. So they're a little bit of a...
That's very zbs and interesting!
Like some jewelry, yeah. Yes, Alinar, Igor found us, so he's using the station to be happy. The deer moves all the time, so...
Yeah, reindeer rogues.
Deer Roga.
Or was he dumped?
I offered him a refresher...
One is
Or put it on the wall. He's kind of like, you know, polite, like, yeah, yeah, yeah. And then Steve says, "Take him a pinkie!" And you'd see, he's got a flame like that in his eyes. He ran to find a rubber there, yes.
I said, "What rubber? What kind of pink?"
Yeah, that was fun.
Okay, so what? The leaders of the largest American banks stated that artificial intelligence increases productivity, but will result in job losses. Meta, Facebook plans to release a new, large language model called Avocado in the first quarter of the twenty-sixth year. China has reached sixty per cent of the world ' s AI patents. DeepSeek has dried open models that are quality near the most modern, there, GPT 5, Gemini, are costly. There are numbers ten, thirty times. OpenAI certification has been introduced and many others have been introduced.
Hello, everybody! We're on ToTheMoon. Technological news, Silicon Valley sites around the world. We go out once a week, although you're watching the anon in the nearest Wednesday, there's a graduation. Sasha's Mashrabov and I are doing a special show. Wednesday should be. I hope the editing's done. So if it's not in this one, then next.
Write comments to make it happen.
Write comments so that you can. I'm talking about, uh, reading again the report where a whole series of banks, there, JPMorgan, Wells Fargo, Citigroup, PNC, they all said they'd checked everything. So artificial intelligence increases productivity. Great, but we'll cut down the jobs. And I remembered the JPMorgan report, where JPMorgan said that jobs would not be reduced, but that, on the contrary, everything would improve. What do I mean? I wanted to say that if everyone listened, that's how it is to gather all the comments of all people in terms of speech, there's a lot of absolute information, Mr., that's a little unstable. Nobody understands what's going to happen, how it's gonna be, who's gonna be the leader, who won't be the leader, what platform has lost or lost.
And I saw a very interesting information on DeepSeek. The man has set a difference in the value description of DeepSeek, from DeepSeek to the price of their model. And there's a very interesting comparison that DeepSeek is four times cheaper, ten times low, and the output is 20, 30 times cheaper, compared to GPT 5.1. And plus almost free, there, at the cash of the hits, well, it's when you work with the agents. And that is very interesting, of course, from the point of view of the price policy of seeing the Chinese model.
And you're reading the news that Anthropic plans to make $78 billion out of it by the year of the year of the year. Yeah, $78 billion, $78 billion. I'll notice Anthropic has the last recorded cash-billion. And as I recall, it was ARR. I mean, it's an annual corn money forecast. So, last time, multiplied him by 12, for example. But everyone else has their own calculations, and they've shown that. It's very interesting when this proceeds, how correct it is. We were talking about a company of technology last night, and there was a man saying that he had two hundred million of company made out of two hundred million ARRs. And you sit there and you think maybe they got that monthly revenue and then they stopped, right? And they didn't have any more money. And...
Well, like, black Friday showed up, multiplied by 12, like, we've got a billion.
Look, this debate, by the way, was very much about American start-ups in AI. And that on the black Friday, someone blew up, sold. And they've done it, haven't they? Even some familiar start-ups did. Speaking of black Friday. You, Tanya, about the shopping sping a few times ago, a few episodes ago, when ChatGPT pulled the trigger module inside, you were clearly right because the report came out: online sales on black Friday, hit all the record. Eleven point is $8 billion in a lot, and it's very clear, largely because of the recommendations of artificial intelligence. I mean, artificial intelligence--
Well, I guess... They didn't miss. See, the main time.
Yeah.
It's time, it's the right place.
Yeah, it's said that artificial intelligence has had a very strong impact on the volume of the product sold and increased it, yes. I'm on it. Well, it's just like the appearance, I think, uh, the marker players like Amazon, yeah, or there's Russia Ozon, Wildberries. They made history when people just consumed more, yes, because of the easy order, the opportunity. And then you're still making endless recommendations.
Well, I think that since the instrument is new, uh, well, I mean AI, he's more trusted than others. Others have already made such a move. It was like a series of years ago, if they showed it on TV, it's definitely, well, really. And there's a big specialist. Now, I think AI's holding this position, but also with time, of course, he'll lose it.
Well, by the way, '70, uh, big TV specialist showed up. That's the same thing right now.
Yeah, yeah, yeah.
It's not different now.
Uh, $77, uh, $77, uh, billion of dollars in the cash flow. Not a profit, a profit, I'm sorry, a profit again. So they're predicting, right, Elnar? They're forecasting now that they're...
I think it's a profit, yes.
Yeah, they've got a very important story right now, look. They're now in possession of a billion dollars of cash, an antropic billion dollars of cash. The last estimate was $300,50 billion. In this assessment, and you don't buy this company, yes, you won't be able to go in there and no one will sell it to you. And the proceeds were a billion. And now it's showing what they want to make 70 billion. Maybe these statements for IPO next year, maybe something else, I don't know. And how they'll make 70 billion profits when ChatGPT wants to make 100 billion profits, 100-- and then they're building up models that are very cheap. And I think that the example of these prices, uh, what it shows, especially the very interesting report of ChatGPT is how companies use and how much more ChatGPT uses in the ChatGPT companies in terms of how it's been used in the ChatGPT. API, there, tokens, GPT modules, all the searches, and the rest. And we, I think we'll talk about it separately. Maybe we'll get a special prom. There's very interesting things in there. Where's all this money going? Well, where's that profit coming from? So where's the volume of sales that can be generating? I mean, to generate this volume of sales, it's a huge number of other companies to sink. Not just to drown, I mean, it's some kind of mastodonite, but it's a new market. So, the new market should have this money or these companies should replace other companies. I mean, to make that profit, we need to make Salesforce disappear. I need a huge amount of Microsoft products to disappear, uh, corporate, right? Because, well, how do you generate so much profit? And even the OpenAI, remember, they said last time they were forecasting, and I think five-year-olds, yeah, they're forecasting growth. You don't remember, Elnar, the number? Users from eight hundred million to a few billion and, yes, up to two and six billion. So even if it's a little bit more, where's the profit coming from? Oh, that's a very, very interesting accent. We-- what do you think of everything?
In the meantime, the Antropic, well, Dario Amoday, on one of his statements, said that they were very conservative and very cautious about the procurement of equipment. And he was there saying that I think, but they say, in a 90% of cases, you know, I mean, I mean,
In 100 percent of the cases, you know, there's a future forecast, that's where, like, ninety percent of the boxes that might arise, and, uh, Anthropic has to buy these equipment himself. I mean, he's got to have enough money to get this investment back. And only there, 10 per cent of the worst negative cases will make Anthropic, well, there's not enough money to pay for it. In this sense, the OpenAI is going much more extensively. I mean, they remember, there's one and four trillion, I think, yeah, they've got obligations to equip themselves, they've already bought, you know, there, anyway, yeah-
Well, for the future, for the future, yes, one and four for the future.
For the future. Yeah, yeah, yeah. Ah, and Anthropic in this respect is very, very conservative, that we in any case have enough money to pay for this equipment, and so on. It just leads to the fact that the OpenAI, if it's all right, will have a huge number of computers, Anthropic will have a lot less. And that's how they do that, and, uh, backwards from OpenAI and most other players want to make $78 billion of profits, which is certainly a pretty interesting question.
Here we are, Elnar, here we are.
Even profits, even profits, yes. Even the profits of 70 billion, yes.
The most basic thing is, the most important thing they're gonna do is make 70 billion profits. I mean, they, uh, understand that they're having money now because they're just a cu-- 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-the-the-the-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b-the-the-b-b-b-b-b-b-b-b-b-b-b-b-b-b-b That's it. You have money, capital, and you, you don't borrow money, you take credit. Well, you give a share of the company or you're gonna change it, there, uh, you can give it to NVIDIA for putting your chips on, yeah. And the vast number of mems, the riels on this subject, like, uh, NVIDIA, changing, uh, valley, share, there, from OpenAI, there, with Anthropic. They've changed everything around, everything in the evaluation has grown up, and nothing's changed. Nothing just happened. So nobody did anything, but just plus a trillion of dollars, like, was drawing, yeah. Here. But it's the market, it's the market. We'll watch it. And, uh, on the one hand, it's just a statement, and on the other hand, we see, uh, development and trends, and how these companies, uh, just go a long way forward and do a huge number of projects inside, which are being made, We don't know, either, yes. And, uh, how to know about them and how to understand them, it's not clear.
So, by the way, about the point, huh?
On the back, uh, your first news you've been talking about banks, that JPMorgan and other banks have predicted that, yes, we're gonna get better at work, and that's gonna reduce the number of vacancies, and Anthropic. Here's the two news match, the fun thing about Kenny Jones is one of the Anthropic directors, one of the first. And he, uh, so interesting analyst, led to the first steam engines in the thousands of seven hundred years, and consequently they developed there later in the TEC and so on. And, uh, if you take a number of horses on a man at the beginning of the twentieth century, you know, when just, there, there, the cars started to show up and stuff, and then, in fifty years, almost all of them. The horses are gone. Well, this is from a domestic use like transport, like, there, a tractor force, like something else. I mean, it took 50 years to get engines to keep horses out there in New York or somewhere else. And then he's following the example with chess. That's right, uh, there, at the beginning of the two-two years, well, there, at the end of the nineties, there, a strong Grossmaster of ninety percent of the parties against the computer won. It was a decade ago, and, as a result, the situation changed, that a man against a computer could hardly do anything. Then he goes on, says, look, I work in Op-- uh, not in OpenAI, I work in Anthropic. And there, there, one of the first directors he was hired. And in the twenty-fourth year, the bulk of his work was answering questions, technical questions for staff members who were hired only. He's been working long, he knows the whole architecture and stuff. And then, there, for advice, how it works, how best to do it, staff have been treated all the time. And Claude came along, well, he was at a seminar this summer with this, uh, report, and he said that Claude had come out six months ago. So, last winter, he became active in Anthropic. And he literally took down all these tasks in months. And here he is, uh, making that conclusion. That horse had, like, 20 to 30. I hope we have at least 20 years too, because it's very fast.
There's no trend.
No, it won't.
There's gonna be some kind of minus mark soon. It was...
So...
You were replaced a month ago, but you don't know about it.
You don't understand yet, yes.
Yeah.
Anyway, I hope we leave a link. Come in, read it. On the one hand, a funny analyst, on the other hand, a little sad.
I want to cry, yes, yes, yes.
Yeah, yeah, yeah. It's on the back of two news you've been talking
about.
Yeah. By the way, big, uh, numbers, and, uh, Ilona Massk SpaceX is planning, uh, IPO in the twenty-sixth year. Uh, and, uh, the numbers are about one trillion dollars. Here, by the way. Yeah, it'll rip Elon Musk off of any ratings forward again. I don't think he's gonna get caught. I don't think anyone can catch him under any circumstances.
I've been thinking about it, and what happens when it's gone? Really? Everything stops, it feels like it's like people are going to Mars?
Sam Altman would have been arguing with you, Sam Altman.
I'd love to argue with him, too. Let him do it.
Speaking of which, these changes that SpaceX makes to the market, of course, are very significant. Remember, we even talked about something that if, uh, there, the delivery of cargo to space becomes, much cheaper, then space will be used, in absolutely inexcusable and unfamiliar ways.
Yeah, well, now we're having a hard time imagineing them. And here I want to remember, Google had news from November 4 and that they were starting the Sun Catcher project. Although they are ODE, data processing centres are to be brought to orbit. There are solar satellites, respectively, because the satellite solar battery efficiency is almost eight times higher in orbit than on Earth. And you can have a solar battery so that it can always look in the sun, so it won't hide in the shadow of the Earth. And I don't remember a specific figure, but if the cost of delivery of the goods to orbit reach a certain threshold, the orbiting of the RCDs will be much cheaper than building them on Earth. And from the point of view of the electricity that comes from there and from the point of view of accessibility. So, the technology itself is ready. Moreover, Google had already protested its TPUs we spoke of last generation in Caledonia. Well, in a rude way, they've been radiating them, checking that memory really doesn't degrade at least there, in the horizon of five years and so on. So it may be that, because of Ilon, multiple missiles, the cost of delivering goods will be much cheaper, and there will be directions that, well, even a few years ago, it was hard to think that the centres would be Data processing won't be near nuclear stations, is it, or is it, hydroelectric power stations, right? Or, on the contrary, in the north, because there's a need to cool and stuff, and they'll just be in space. Now it seems more or less real.
Yeah, that sounds really cool.
Head Google has been saying for weeks in a row in each interview that they will build infrastructure in space just indefinitely. So all the X is in there, right? And that there's a infrastructure there and Japan. Japan, I think the first one to get back from space is the power that's accumulated plus, I understand, plus, yeah? And literally yesterday, I think I flew and saw such data that Japan is right with super-innovation, super-innovation in space from the point of view of electricity. It looks so specific. See if they can solve this. But by the way, you said about the value. And I remembered, so the story about Neuralink is the last one.
So, it's been a lot of things, and it's been reports that concern, well, the company, and then SpaceX is out and out, now, interesting, and just a few days ago, Neuralink, it's the company again, Ilona Mask, I've accelerated the implantation of neurochip. And there, so I like the news. Well, the information that's written is written. You're talking about the cost, and it says, "Strenched the implantation of neurochip to a half-second and lowered the cost of the components by ninety-five percent." So Neuralink has announced that her surgical robot now has the ability to wilt one electron into the brain in a half-second.
In a half-second, when it was time before. And I think, "uh, is that how speed matters here? It's like a factory.
Yeah, a minute or a second, whatever.
Of course. Like, like, kind of, like, look, like, what, look at, like, what, what, what, what, what, what, what, what, what, what, what, like, what, what, like, what, what, what, what, what, what, like, what, what, what, what, what, what, what, what, what, what, what, what, what, what, like, what, what, what, what, what, what, what, what, what, what, like, what, what, like? But if we're gonna look at his statement about the robots, which means he's doing a factory that's like a million to give birth to robots, then ten billion, millions, and a billion. He says he says, "Well, then we'll probably be able to release a billion a year." And the factory is a million, I think, or ten, even they're building in Texas. And then there again, there was information that they'd be there in the first quarter, so they'd be mass-produced and so on. But we'll see what happens. But about this Neuralink, there were about 10,000 new volunteers for future trials. Hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, Would you volunteer for testing or not? Do we have volunteers among our, uh, viewers? Ten thousand people were volunteers.
I'm gonna go in on the implant for future tests, right? I'm, uh, there's a bunch of tests on it, yeah, what can these people do with the Neuralink now in place. Yeah, how they can control something in the game, how they can move things, how they can move things, like they can move things like that. Well, I see there's still a lot of different start-ups that, uh, they're doing this. It's just because of Ilona Mask, we're gonna sound like we're gonna show Neuralink and, uh, Elon Musk's probably gonna have one of the biggest resources right now in the world of naa to invest. Especially he-- that's exactly the technology he's got incredibly high attention. But for a second and a half, I have a question, of course, what are they planning to do? They plan, uh, a billion people to infiltrate Neuralink or something? And there's a word that says...
If you multiply eight billion fast, how much
it'll work.
Yeah, yeah, yeah. And here comes a reminder of words, words, um, Vanc or how the head of artificial intelligence in Meta, who says that, uh, the kids don't have to be born now, you should just be a little late when the chip is quick. We'll have a baby and teach it right away. Well, he'll learn everything there, all math, there, parole, all physics, everything will know, will be a super smart man. Write on the subject. It's gonna be very interesting to see your thoughts on this story. What do you think? Who would like to do that? Who wanted to do--
I'm sure if this even happens someday, it's probably gonna take the baby's consent. I mean, it's time for him to have a baby, because for 18 years, something like that will be done. But it sounds really awful, actually.
Yeah. And whether the consent of the child or whether there will be any law that will allow it to be enacted.
I mean, it sounds really, really, really, really, really, really, really. Here. Speaking of Meta, since we remembered Meta, and I think that news is important because in the past, we said that Meta was back, Meta was gone, Llama was gone. Although we have provided data in the treads in the past release, it is still very important and important to open models in the test and use. But Llama's still quiet. And then they come in Meta insiders, Chinese insiders, well, Chinese, and then the American media write that there is a code name for the model. So it's the code name of the Avocado model. Iia, um, what's gonna be the next generation of Llama. That this model is supposed to be in the first quarter of the twenty-sixth year. And that model will be closed. I mean, at least what it describes is that it will be, it's going to be as a pro, and that access to it will be through API without open weights. And, like, it's gonna turn Meta around. Very interesting, very interesting, probably with a resource, again with a resource, with Meta's money and a quote from Zuckerberg. In that Cookerberg, we've already said we're willing to spend any money, including wrongly on artificial intelligence, even if it's $200 billion.
Says they're obviously trying to make a race in artificial intelligence, they're-- well, in this race to stay and get on that list, as Llama was formerly present. It's just that we've been discussing a lot of things before. Yeah, remember. A lot of things, always everywhere, yeah. Still, to return to this race and beside the OpenAI, to be near Gemini, at least to be near Claude, at least in terms of, uh, American models, yes, if not speaking about Chinese. Here. So we're waiting for Avocado.
I wonder if Avocado is a way to keep Banana. Next, all the fruits, vegetables.
Yeah.
OpenAI.
Yeah, they're gonna go, they're gonna go, it's, uh, it's all, it's... It's, you know, it's an interesting story that I'm with Elnar today when we're in the middle of a call now, Elnar says, "Where are you?" I say, "I'm in Santa Fe." Elnar says, "Hyundai." It's actually, in fact, we know a lot of words, like the Apple Big Sur used to be, right? And, and, yes. And you, uh, you hear some kind of... or there was a Monterrey or they had, uh, different names used. And you're thinking what kind of words are, aren't you, beautiful? Or there's Sequoia, they use it, right? You have a name, and then you're with California living, and you and-- well, Santa Fe are not California, it's New Mexico, right? But you know what New Mexico is, don't you, Mexico? And not near Mexico, as it may seem.
And there's snow in there.
Yeah, yeah, how could it seem, yeah. And actually, there's a little snow on the street, but there is. And that's very interesting, of course, when you know where a word really came from. And...
The depths are coming from all of us.
Yeah, very extra depth, yeah. Even, as you say, you're different. So you realize that it's not Zuckerberg, but Zakerberg and not Nike, it's Nike, right? And you're in a huge number of different brands and all--
But BMW, not BMW.
Yeah.
Let's start the lobby.
I can't change anyway. I'm so curious what? What did you say? What are you up to?
So interesting, Tanya, if you're with Americans, you know what you're saying, like, you know, you're talking Nike, right? Well, you're saying, you use it, obviously. But if you're in-- with the Russians around, you're not-- or I'm even talking to people who don't live in America, I don't understand that they're talking Nike or Nike.
And this is my wife, Pauline, very--
If you say Nike, you're like, "Oh, I see, everything's trying to--
Yeah, you're like you're painting.
Under Americans.
Yeah, yeah, like you're painting.
What are you doing?
Yeah. On the other hand, you, you know, you get used to it, right? And you know that--
Yeah, and you can tell on the machine.
Yeah, it's like things, you can't come to the store here and say, "I'm S-Size Coffee." In California, they won't know that it's often they're gonna have to say small, right? And Pauline says to me often, "You're in the store talking, they can look at you, stand up and say, "What are you talking about?" I mean, everyone has, like, habits, yeah, some cultural traditions, names, words. A lot of things. And she, Pauline, is always telling me that in Russian, only Russian words are strictly categorically used, yes, absolutely not English. And...
She's been sinning in English lately.
Of course you start sinning English words and keeping-- we've been seeing one person. He's been living in America for over 20 years, and he's not using it, you know, English, he's not using it. And that's really cool.
Very delicate line.
It wasn't Sasha Masharov.
Sasha, everyone will remember Sasha Masharov now. No, I'm saying at least...
I'm worried.
Yeah. Well, it's just, Ilnar, you know, this is Nike and Nike use, right? Especially when you say that, I'm talking to my mom, you know, my mom Nike won't talk about anything at all. I mean, she can't. I mean, basically, and our age, there's nothing to people of Nike. Well, you're what? I remember the first time I heard, I had it like I got hit...
Heart knife.
Yeah, the knife in the heart.
Yeah, in the view of Sas, there's no clear borderline, a rusty word, well, a sandwich is, is that a Russian word or not a Russian word, can you use it?
German, I guess, huh?
And the calories or-- yes, yes, yes, yes. A sandwich. So there's--
Brod, yes, bread.
Well, from a habit point of view, yeah. But in terms of habit, especially with all the brands,
Well, sometimes there are such a small word in English that you can't be translated into Russian.
which cannot be transferred to Russian in Russian. And I want to mix it up because the word is 100 percent, or it's 99%.
So it's not exactly what it is.
That's why avocado, avocado, means you want to say it again. You once told me, I think six months ago about patents and audiences, so for yourself, it's a very interesting thing. It's a general understanding of the weights world. I was about, uh, when we started to drive a year and a half ago Tuzmun, and I was very much on the subject of what you're looking at in those countries where you're in, that a huge number of countries are no longer part of the world. artificial intelligence, never in his life until-- well, just not catching up, yeah, they've lost the market. And you don't have to talk about it any more recently. What's the point? And last time, I think we brought the UN. Yes, UN. The United Nations we said they did a whole reservoir, an article, a reservoir, and we did a study and an article on the fact that artificial intelligence is tearing apart, well, very much of the country. Many countries have already lost their lives in this area and will not return to the exact same subject. And here's the question for China. Yeah, we know what U.S. leaders mean, there's a leader of China. As it can be seen, in general, American models, they have a strong weight, a huge amount of money, and there are large and so on. China has a very incredible subject. First, the number of engineers they make is the number of super engineers. And it's because of some installations. For example, a lot of people think there's the coolest body there, or, for example, I don't know, there's Russian programmers, they're cool, or they're Belarusian cool, or these, or these. One more time about the Chinese. Nine out of ten world tops of Chinese engineers in artificial intelligence and 60 per cent of the world ' s patents now, sixty already, in China. This new number has shown that China has reached sixty per cent of the world ' s patents. And, frankly, it's a number, uh, that's a big deal, especially when you know, especially when you realize they're, by the way, growing up. In general, the Chinese industry has grown by 24 per cent, but it is not that large in the world. They now have an artificial intelligence intel in the world that they are 15 per cent of the market, yes, but 60 per cent very impressive, knowing that Chinese people are probably inside. I don't care about patents. Well, perhaps domestically, in terms of international law.
Yeah, the most loyal country to patents.
Because, uh, China-Chinese has been a long time ago, and Chinese products are quite different. And according to the news, the last time America's like is America, China is blocking, and Trump has allowed to sell NVIDIA chips to American-Americans.
Chinese companies. And imagine that's very interesting, right? So, at the beginning, everybody's telling me that the chips should be spreading Americans, then they say we don't steal technology. Then they say no, we'll, let them be delivered. And China inside says, "We're not buying part of your chips, and we're gonna be very strict about buying chips, very strict control over the purchase of chips, because you might have put your chips in. "some virus there and some data you want to keep." What? What, look, I think if the American government decided to open the chips, what does that mean? Which is probably something inside these chips.
Well, there's chips, to admit, past generation, there's a H200 in it, not two hundred that are now, yes. One. Two. China, I understand, has determined very well that it is simply some independence in this respect to obtain from the chips, and it nevertheless makes its internal companies like Huavia actively develop. Makes them really teach their models. But it's not a success yet. Until Huavei has accomplished this task. The same DeepSeek, remember, we talked about this in August, that they were trying to build a new model on the Huavei chips. Inference is still working. So, after-use, that's how the Huavei model can't be taught chips. What Google did, what we talked about last time. And now, the DeepSeek models that started talking about, they're also trained in new NVIDIA chips that were taken to China on different schemes. But this is the mass distribution China wants to have its chips. Well, because the U.S. is now allowed through, I don't know, six months, said no, we changed our minds, yes. Yeah, yeah, yeah. Therefore, China is very important that its chips be as competitive as possible in this regard.
And I wonder why they are now allowed to deliver chips, even if they're inside American companies, even for American start-ups, they're all very strict in terms of equipment, they're very limited and limited. large amounts are produced by agreement of the Government. And it's all very ticking and it's not so easy to buy. I remember someone from us wrote, "What was it that they bought in Russia, so the NVIDIA chips?" Well, you buy, you're just not gonna buy in Russia, you're not gonna find the volume, you're not gonna buy it, yeah, it's really, it's really lymium--
Ongoing, Sash, what China started talking about, about talents and stuff. Last time we spoke, yes, DeepSeek was the first to publish the GRPO algorithm. Now they've published how they teach their DeepSeek model, so in V2 math, three two are for the main diaper. And there's a very interesting approach they're using. Not much effort is being spent on data marking, which is the most expensive. Yeah? And they show how they're building a multi-layer process. So they're making a model to generode answers first, and then, accordingly, people just say good, bad, bad, good, bad, bad. They start learning another model that will try to predict good or bad, and so on instead of people. I mean, there's a whole bunch of different models, but they didn't have much money to spend, at least as they say, and not much of the effort of living people to do it to mark the data, To get a very good score, yes, that's what I was talking about. So, a few models they've released now, very good results, and they publish, and how it was done. I mean, in a way, China is also moving forward with one, well, I'm not gonna say, science, but an open part of it. I mean, I'm sure OpenAI and Google, of course, and Anthropic's work is not much worse. Who knows, though? Here, China is the only one who's really sharing, publishing all the interesting applications.
And you and I know very well that China and China's internal technology, they don't even open them. So a lot of things are closed. Nobody understands what's done there. If we're in the open, we're working with the government, so we're making laws, then there's 100 percent of them, everybody's working together, 100 percent, everybody's doing it, 100 percent is going to be rolling. artificial intelligence everywhere, everywhere. In fact, China ' s automation in terms of the notion of automation is on the front planet of the whole. I'm talking about scales now, I'm not talking about individual cases when we're trying to get an example, there, I don't know, we're Singapore, so we're auto-accused. Well, Monaco's still automating. Well, I mean, when you automating, well, some small region, this is a very simple task. When you need a huge country with so much infrastructure to clean it up, it's a serious question. But China always hurt me. I remember, even two thousand thirteen years, I had a trip alone to Tibet, and I was a lot in China before, and we crossed the border. Tibet, Nepal. And now you feel this difference, yes, when you're in Tibet, and there's cameras hanging around. And these super roads are made. It's clear that because of the Tybeth hijacking, there's all the things, and then you have a bridge like that, you're right, so you've got a passport in China checking all those smears, where you've been served, the quality of the staff, So, it's all in the cameras, it's like some scanners, it's healthy or not healthy. You go on the bridge and on the middle of the bridge, I remember we're coming up, standing up a little kid on the belt, and that's how you keep the sign with our name. There are three of us. We've crossed this border on foot, we're going out, and you're just moving in, there's a village like this, a chicken, some wooden buildings. We're going through, like, a mess.
We go further, we go into some wooden building, there's some uncle or someone putting us in a passport. Next we go, get in the car and get on a balloon road, so you're going forward. Well, it is clear that China has left from a technology perspective very serious and everywhere in terms of coverage. And it's clear that the ghost is that the artificial intelligence is there.
That artificial intelligence is in the area, and it's testing a lot of real data. By the way, a big question in terms of, uh, government data, probably China's 100 percent and human, you know, real man's data. There's access to everything, because in America, everything is classified, closed, systems between different. The president can say, "I want to get in here." He's said, "You're not coming in here." So, the fight is all-- all-state groups. A very serious constitution for data preservation. In China, obviously, we can scan anything, all the correspondence of all people, built--- they've scanned all of us in WeChat, yes, we've talked about it, by the way. And you know what kind of data there is, which can be analysed, which can be seen, which can be understood from the point of view of human life. For example, the United States has no access to life-sex correspondence training, modern human correspondence. They cannot take and teach the same masses on the communications to WhatsApp or on the iMessage or Facebook Messenger. They can't do that. In China, uh, first of all, the number of people exceeds all the possibilities.
And there's some crazy training inside. It's gonna see what's going on with him, yeah. That's interesting, of course. And I wonder, by the way, we're talking about AGI, that whoever releases AGI is gonna be kind of ripping up ahead and, uh, getting off the stratosphere, right? He'll stop writing, so he'll be completely ripped off. What if China is the first AGI? Iia, or something that might be called AGI, ASI, there, and all that smart, so it's cool. Here. We'll see. Very interesting. But engineers, of course, there are a lot of cool--- cool things now. And I was meeting again on Monday, and I was at Antropic's office, seeing who? With Chinese. In OpenAI, you wrote last week, texting with who? With Chinese. Here. Yeah, there might be a Chinese man born in Canada, for example, not America. And someone in America, someone else, but a Chinese man. Here.
Fifteen years ago, when I was more or less mattomatician, publications and other, in English, you read the math article. The authors are necessarily Chinese or all Chinese, or, in part, there. So they've been very active in science at that time in a completely different direction. Math, for example, was very good at seeing them get into this area.
Yeah. And they're the first to put a chip in their head. Yeah, so kids. They're not gonna have any problems with the law. I'm thinking there's no problem with the law about the children's introduction of chips. Here.
They might, by the way, allow more kids to give birth, yes, as soon as the chips come.
Yeah. Second child with chips.
And by the way, that's interesting. You know, interesting data. If you see the reports, there's a shortage of people in China. For work. It's actually, uh, it's often good to see global trends in terms of economy, human displacement, facts. For example, in Russia, the shortage is a huge, huge shortage of people to work. I mean, 100 percent-- very ma-m-male unemployment, lacks people in all the fields. China has a huge shortage of people, and they are interested in people coming to see them so that they have people. Again! Brain doesn't take it. I mean, it seems like it?
That's got to be different.
Yeah, yeah.
And, in fact, they're using every possible way to do that. I wanted to, uh, know Tanya. You, Sash, I think you said you didn't use it. In ChatGPT, integration with Adobe products has appeared, there's Acrobat for PDF, with Photoshop. Did you try anything interesting with the Photoshop?
No, no, I'm just trying today. I'm just trying to try this today because after Nanobanana, I've forgotten about the Photoshop. Well, I mean, he's so serious, at this professional level, I didn't need an interior. And then you dropped these, and I want to try some of these presentations tonight, because I think it's gonna be more serious detail.
Mm-hmm. If we do, let's add this little video. It's just that I asked the Photoshop to identify, uh, people in the picture and background, and he, therefore, besides the picture he gave himself, he also gave the tools that Fotochope has, to be self-built, how much gray cross or, consequently, coloured can be left. Yeah.
I'm right now, Ilnar, I think that's why the future of all the designer programmes I use, and I don't understand why this has not yet been done. Just in AutoCAD, you can really sit and say, "Three line, I don't know, 12 inches, wall width five. You can't even tell him the width of the walls. He knows what the walls are, in America, to be used as a matter of fact, if there's a plot already chosen, style. So, how-- why isn't that? It's so easy and logical, right? And then, these programmes are actually available to ordinary people. Or there, uh, very, very, there, I don't know how many teams in AutoCAD, you can still create codes, make some more up, and they're all kind of-- I sit down sometimes and remember how that team was called, You're not used very much, are you? And there's no fillet or something, I don't know, well, I remember visually, and I rounded the corners. You tell him, and he does it right away. Why isn't that established yet? I mean, that would be really cool.
It's a dream!
Tanya, it's gonna be very soon. I'll tell you about it a little bit, too. A very important thing happened that would lead to it very quickly, most likely. Oh, little explanation, yes, the difference between NanoBanana and, uh, that's the photoshop tool in GPT chat, that NanoBanana's genetic, yeah. And remember, we used to put a past clip where Sasha and I were in the shape of cowboys and so on, English dendi. And here comes-- that's where, anyway, it's like he's trying to keep a face, but still he's genetic, and there's still distorts and other things. And here in this mode, you're getting a picture of you sending to the Photoshop and using Fotoshop's tools for this. I mean, like, yes, if you need to separate an object and change it, I don't know if a colorocore can be done or something, the image isn't regenerated. And in this editor, Fotochopa uses tools that process him the same way you do it in Fotochopa. And, uh, very important-- yes, talk.
I'm afraid you don't lose your thought. You know, before NanoBanana was born, the designers used the Photoshop just to get when there was no opportunity and 3D visualization, the Photoshop was used, and they were pretty good. photorealistic images, like a combination of these layers. I mean, it's a little professional in the same plane for me. So, yeah, that's gonna be interesting. And then, I think when you've done all this, you can send it to NanoBanana, and she'll break it up, there's a prospect that'll do everything else. We should try that. All of you. I know some designers are watching.
How do you call this regime? Write Adobe Photoshop, do it there and then you put a picture in there. The regime is glute. I had to put the picture on a few occasions. He says, "I don't see your picture, can you put it back?" I'm putting it on, but then I start working. It's probably gonna fix it, but it's a little weird for now.
Like Convoy.
The challenge to the instrument is you're just his-- yes, yes, yes, yes.
So, with Convoy, as with any tools, you're just calling him, once you agree that he's been connected, and then, in fact, this chat will be working with this tool. Now, Tanya, what you were saying, this is really cool stuff. Ah, this week, Agentic AI Foundation was established. It's a fund, it's a dissort, it's under Linux Foundation, I think it's under Linux Foundation, I think it's in my opinion. And the big companies have lost their technology there. Oh, for example, anthropists gave their MCP protocol there, said, "all, now it's a public treasure. Use who wants to, yeah. Keep going. It's just that it's all working. OpenAI gave it to its Agentic MD. Yeah, it's a file that, uh, tells the agent what to do in this situation. Google's been sent to the hotel and there's a whole range of companies that have been thrown out of their technology. I mean, MCP used to be anthropic, yeah, they could use it, but it's still company property. And now all this has gone down to one place. What's that got to do with it? This leads to these open protocols, if they all start to use them, they create standardization. And it turns out that, for example, if AutoCAD wants to add this to itself, agent behavior, LLM connection, and, by the way, that's a protocol that works. Suddenly, the OpenAI doesn't like them, the star--
The OAI is not comfortable with them. They're on the Anthropic. The anthropics are not good, they're replaced by Google. And all this is almost absurd. Well, I see, it's always like a bunch of skirts around here, because there's gonna be a lot of things going on, but it's gonna be a lot smoother. This is the news that Google has a part of her service back to MCP. This is likely to be the consequence of MCP now being located at Agentic AI Foundation. And for example, I remembered exactly that the cards, I think, were something else, would be available to call any systems through MCP. I mean, you don't have to do anything alone. And with the development of this fund, more tools will become available for use, wherever practical. The same photoshop they were smashing off the OpenAI, and they had this story in ChatGPT. But if this Agentic AI Foundation story is developed and, accordingly, Adobe, the company will have its tools on these protocols, then all the others, even DeepSeek, even though Gemini is, They can add this function. It won't be there, there's a two-stop connection. It will be that the Photoshop provides these tools, and all others will therefore start to integrate them. It's gonna make me don't know, the editors will use Fotoshop directly, there, plainly, another question, but it's gonna be a lot easier just because you're starting a connection there. work. And that's what you're talking about AutoCAD, there, make me a wall of this thickness, so many inches, and so on. That, I understand that six months, a year and a lot of instruments, this thing might come out.
It's a cool thing for integration, of course. From the point of view of what they're offering, they create protocols so that all systems can communicate together so that you can't get a cushion like smart devices, yeah, when you're just having serious restrictions, you're all time is tied, you don't understand what's going on. I think I have ten systems at home, like, 15, and a lot of signatures. I mean, it's a very, very difficult subject. Yes, Eldar, tell me what you wanted. I'll add the whole association there, too.
Yeah, yeah, yeah. That, Sasha, what you're talking about, is making this technology market fragmented, right? That you have a fragment that OpenAI does, there's a fragment that others do, and they're very badly connected. And the emergence of such open source protocols makes it all smooth. And you have, like, every tool inside, all the others, it's the perfect world, of course, but everyone else is starting to be available. And the last thing I'd like to say is, I've been watching a series a few more recently, and I don't remember what it was called in English, Pluribus, I think it was, but the Russian translation was kind of like "One of many," something like that.
So, short.
Intrigue.
Because of the events, yes, yes, yes. Because there, some events have happened that people have a sense of being together. That's all I knew now, I'm telling you, everybody knows. Everyone else knows, I know now.
They say it's a great show.
The idea is really very busy. And, you know, what does that mean? In the word, there's a girl with some kind of girl, I don't know, Thai village is now getting on the plane's shuttle and starts flying just because there's a lot of people in the pool. That knowledge is available, and she's also available, and she can use it. And that's when you said that a lot of people there would start using the AutoCAD, even those who had never used it before. And that's happening now, because you, you, you don't know how to do anything. Yeah, you're on parole, asking Gemini, ChatGPT or something, and now you know how to do it. In fact, you've got, uh, neurolinic to the head, but you still have access to the data, it's getting much faster, much more smoother, simple. Yeah, and we're actually moving on. I've been looking at a few episodes, but I'm gonna get rid of this feeling that it's actually happening, it's not working. Oh, and, Sasha, what you said at the beginning about efficiency, there was a report from the banks. I think there was a report from the OpenAI that people say they can do things they've never been able to do before.
Yeah.
That's what it is.
Well, that's one of the main stories, by the way. That we're broadcasting. No, I couldn't do research like I'm getting right now, for example. Or I couldn't handle this amount of information in ten minutes or, at least, I wouldn't have had the opportunity to process that amount of information, for example. Well, or how you're saying now, pictures.
Get some new technology fast.
Tanya, just yeah, just change pictures, right? It was unreal, it was impossible. Or create some kind of microvideo.
Yeah, I've been studying it for months, and I'm getting a little bit of a shot at Fotoshop.
And now you're through ChatGPT like this once, once, and everything's ready.
Yeah.
Well, while he's glowing, it's a matter of time.
It's a story, exactly... Yeah, it's exactly a story like what we said about driving, yeah, when I have four kids, I realize that a daughter who's 15 now, she's starting to learn to drive, and all the other kids, They're not gonna learn to drive. Probably. Or there's a kaif, maybe, you know, a little bit of a ride to get them to some place where they're gonna need to rent a car.
Don't be stiff.
But by now, most likely, in the U.S., no, most likely, with the U.S., Waymo is expanding again to three cities. There's a lot of stuff again. Probably not, well, there's gonna be drones, yes, and there's gonna be drones.
We arrived yesterday, in New Mexico, rented a car. You're probably just gonna rent, you're gonna get a car without a pilot. That's it. You're just renting her full time or you're partially renting it, depending on what you need. Well, it's a parole. I wanted to add a story about the association. Now, there's a report that Google, and I think that's a topic that's, uh, complementary. Apple and Google cooperate for the seamless transmission of data between iPhone and Android. I'd like to smile and say, "You'd have built it to the end so that it's okay between the two iPhones." It's still glowing, but still. They're not, well, it's really gulking, is it? It's really not always catching, it's AirDrop, and so on. So they're back on the Inside, but it's not open information, they're developing a new direct migration function with iOS to Androids and nab-- and vice versa. And the first tests google Pixel, and the next beta version of iOS will also have twenty-six functions. They're writing, yes. And, uh, means, well, this subject, of course, is very useful and very good, very good. But again, um-- when you see artificial intelligence and models like ChatGPT that build a Photoshop, I have a question for a Photoshop, and what are you gonna do when a person gets used to using your model? Inside, and ChatGPT will suggest that there's a lot of 40 tools to use here, right? And who's gonna pay for your service? Or you'll pay the money in micromodels, ChatGPT you'll be throwing some kind of a little bit of a penny for it, yeah, for using your service. So how will the price, uh, how will the price of race-- how will the prices be determined? That's what we see, by the way. Remember, Sam Althman said a year ago that a ChatGPT would be around this year, worth a thousand dollars, and then it's gonna be in the fourth quarter, they said the subject that could be thrown out there. Ten, $20,000, she'll make you a challenge. But we will not see this coming, but we see only the cost of API models, the improvement of models, the improvement of productivity. And I'm for the same $200 that I paid for Pro a year ago, I think Pro is probably a year now, so I'm not wrong. Well, when he showed up, I've got him all the time, I've only got better, uh, well, characteristics, right? I can do much better, I can do, uh, from-- much more, a lot more than I used to. Well, I'll see you in a week. And on Wednesday, special release, and we're waiting for you on special release. Bye, everybody.