There is a sharp pullback in crypto. The reason is a new major language model coming out of China. In other words, an artificial intelligence model, called DeepSeek. DeepSeek users are amazed by this new model. It is currently ranked number one in the Apple app store. Its users say that DeepSeek gives better results than both Open AI's chat gpt and other solutions. Moreover, it is close to free to use, much cheaper. All these have created great admiration for DeepSeek. So why is this bringing down the stock markets? The first fear is whether the US is falling behind China in the artificial intelligence race. Because if a newly released language model can provide more effective results at a lower cost, this could mean the end of the US. This doesn't really make sense. Because newly released language models are usually more efficient, faster, and more successful than their predecessors. There have been many examples of this before, the reason being that they take the outputs of the previous language model as input, to put it roughly, and they do not make the mistakes made during the development of those language models. They are finding ways to do what those language models do more efficiently, more practically and faster.
Also, DeepSeek has been used by amateurs so far, the appreciation comes mostly from them. It has not been subjected to hard benchmarks yet, we will see the results there. But users say that DeepSeek is better, cheaper, in this case, is the US losing to China in artificial intelligence, this is the first concern. As I said, I don't really understand this concern. Because DeepSeek is already open source, developers in the US are now seeing what the Chinese are doing. They will adapt it to themselves and this competition will continue. The real big issue is that it produces it at a really low cost. Because DeepSeek's claim is that they developed this great language model for only $5.5 million and that they did not use high-tech chips such as the latest Nvidia chips, the H100s, while doing this, they achieved this with old generation Nvidia GPUs.
Why is that? Because, as you know, the US is imposing a technological embargo on China. That's why DeepSeekers found our own new solutions, made innovations and thanks to this, there is no need for Nvidia chips, AMD chips, Arm chips. At least that's how the market reads it, they don't actually say it and that's why there is a big sale, especially in micro chip stocks, while they are ready to sell, they also touch crypto a little while they are at it. Everything is being sold in the technology sector. However, the issue of DeepSeek costing 5.5 million dollars is probably a whole story. This is the real breaking point anyway. Because if this work is done at such a low cost without the need for Nvidia's new GPUs, it would be quite reasonable to question American technology companies. But when we look a little behind the story, we see different things.
We know that there is a hedge fund behind DeepSeek. This is actually a fund founded by Liang Wenfeng, who studied at Zhejiang University. It is an expert in shorting stocks and uses quantitative models and mathematical models while doing this. They have been using artificial intelligence for a long time for this. By 2023, the size of the fund managed by Liang will reach $8 billion, and Liang is looking for new hobbies. The story is told like this. This hobby also stands out as artificial intelligence, and he sets himself the goal of developing an artificial general intelligence that thinks like a human. Thereupon, this gentleman goes and brings together around 130 doctoral students from the best universities in China. These 130 doctoral students work on old model Nvidia chips that were previously acquired for about a year, and today's new model emerges.
The first question is, there is a model that 130 doctoral students work on, and there are also Nvidia chips there, even if they are old models, they have a purchase price. There are energy costs. How can something cost 5.5 million dollars with all these? When we look at it worldwide, the average salary of people interested in artificial intelligence at the doctoral level is between $250,000 and $500,000. I don't think it is very different in China either. Because there is serious competition there too. Something that was developed by 130 engineers working for 500,000 dollars and if we do not include electricity, GPU, and all other costs, costs millions. This is of course bullshit, or rather, the company does not say it like that. The company says they trained the latest version of the large language model for 5.5 million dollars. There are previous versions of this, there are trial and error, there are infrastructure investments. First of all, this is all bullshit.
On the other hand, it is a fact that the learning costs of the new technology DeepSeek are cheaper than previous technologies such as Open AI. Because here, for example, it is said that they apply reinforcement models that do not require much human intervention much more strongly, there are certain innovations. I never deny this. But this difference that is told is not possible. Because, look at the story that was sold to the market, Open AI was able to develop its first model for 2 billion dollars, these developed it for 5.5 million dollars, look at the difference. As you know, Open AI is trying to make a chip investment of 500 billion dollars. So this is bullshit, buy it, this is not true.
The second issue is a subject that is seriously talked about in the market and many people stand behind this claim, in fact, DeepSeek also has H100 chips, that is, Nvidia's latest chips. According to one claim, it has around 50,000, and according to another claim, it has around 100,000 of these types of chips. The company may have actually developed this system using Nvidia's latest model chips. How did these chips get there? There is an embargo, but sales to Singapore are very intense. Singapore sales are very serious among Nvidia's sales. The relationship between Singapore and China may not be very far. There are rumors that these chips were transferred from there to here. Of course, these are rumors, we do not know if they are true or false, and DeepSeek cannot say this. Because in this case, the embargo will be broken. It will face new sanctions. But there are such rumors and probably the back of it is true. Because Elon Musk responded to a tweet sent today. It was a tweet about how this company did this with economical, cheap, old chips, and Elon Musk said in his own tweet that this was bullshit. In other words, someone who understands microchips a little bit never believes in all these miracles, in fact, and says that there may be such stories behind them.
A third important issue is that the aim of companies in America is not to develop large language models. Large language models are actually an intermediate stage, and many technology experts were already saying that these large language models would no longer have a great competitive advantage after a while. The real aim is to go to artificial general intelligence. In other words, reaching human-like minds, reaching artificial intelligence. In this context, if you believe that this can be done without powerful chips, if you believe that this is just a matter of large language model optimization, I think you should sell your shares. But that is not the concern of the Americans. A powerful operating system is also needed to reach artificial general intelligence. This reality is quite clear in my mind. Of course, you will do your own analysis and remember that I am not an artificial intelligence expert after all. I am an investor, I am just looking at it from that perspective. But my mind works in this direction.
Another issue is that we know that the Chinese state is behind everything that China develops. The Chinese Communist Party is definitely behind this DeepSeek project. In this case, all the work you do on DeepSeek actually becomes data for the Chinese Communist Party. As you know, this is an open source program, and since it is open source, developers are also developing new applications on it. These applications are actually being developed to serve the Chinese state. You can do that if you want. Because you can say that when I work with Open AI, I am actually working for the US. It makes sense, but the problem is that US companies will never use this application. Because they will not want their own data to fall into the hands of China.
The main big market in artificial intelligence will be the corporate market, this is clear. America's Hyper Scalers, that is, the big seven, big banks, other big technology companies, the main customer base is there and they have zero chance of using DeepSeek. I think they will continue to use solutions like Open AI, even if they are a little more costly and less efficient. So, to cut a long story short, I may have confused you a little but the point is that DeepSeek looks like a high-quality new big language model. Many of its solutions may be better than the current ones. In a week or two, the Americans will imitate those solutions, it seems quite possible that there will be a commodification in that area and the cost and price advantages will disappear. For example, I had already sold my Microsoft shares in this context. Because I thought Open AI was open to this type of competition. But when it comes to infrastructure, that is, when it comes to the microchip world, I think the situation is completely different. I think that all these new artificial intelligence models, especially when going towards artificial general intelligence, require microchip investments and there is another interesting issue here.
If the efficiency of a resource increases, the demand for that resource actually increases. If DeepSeek has discoveries that will increase the efficiency of microchips and these will continue to come in the coming days, this will increase the demand for microchips. Because it becomes possible to produce new solutions from microchips more economically. There are many examples of this in history. In this context, I think the demand for microchips will increase even if the whole story behind DeepSeek is true. Plus, if DeepSeek really creates such a big competitive concern for America, Americans will of course strive for more efficient software. So obviously, this cannot be done by just investing money in microchips, efficiency must also be increased. But on the other hand, I think they will toughen their competition in this area even more in order not to be defeated by China.
So when we look at it today, technology stocks may take a beating for a while. Because the market loves these kinds of rumors, but we are basically seeing artificial intelligence moving towards a new stage. We are seeing the competition between the US and China moving to a new dimension. At such a point, I predict that the US will increase its investments in artificial intelligence data centers, not decrease them, and there are so many gaps in the story told, especially on the cost side, that I frankly do not buy them. The Chinese state has made many announcements about technology in the past. Then, as you know, they turned out to be empty. For example, some may remember that they were going to place a new moon in the sky, an artificial moon. I can list dozens of such projects.
This does not mean the following. The Chinese are working very hard. The Chinese are also developing good technology. China currently has good universities. They are in serious competition with America. I have no objection to these, but sooner or later this business comes down to microchips. There is still no competitor to Nvidia's microchips in microchips. The Chinese still cannot produce ASML machines, which are the machines that produce those chips. Everything else seems temporary to me. Of course, I could be wrong, what I am telling is by no means an investment recommendation. But I wanted to share my market opinion with everyone.
I do not currently have a change of opinion on the US stock exchanges. There are many balance sheets coming this week. As those balance sheets come, the CEOs' statements will come into play. We will see our way more easily within the framework of those statements. For example, Tesla's balance sheet is coming this week. Right now, I see that Tesla is also falling in the early market. However, Tesla has nothing to do with artificial intelligence large language models. As you know, Tesla is about real world artificial intelligence, in other words, it has nothing to do with what DeepSeek does in areas such as image interpretation, but they are selling that too. They are selling Bitcoin, they are selling everything. We are in that market right now.
The information, comments and recommendations contained herein are not within the scope of investment consultancy. Investment consultancy services are provided within the framework of the investment consultancy agreement to be signed between brokerage firms, portfolio management companies, banks that do not accept deposits and customers. The comments in this article are only my personal comments and these comments may not be appropriate for your financial situation and risk return. For this reason, investments should not be made based on the information and comments in my articles.
