An AI prodigy and a suspicious death linked to DeepSeek

Let's talk about the Chinese AI company which founded less than two years ago, DeepSeek. It claims that its AI model rivals OpenAI's latest by using lower-end Nvidia chips and costing just 1 20th of OpenAI's investment. But is DeepSeek a true technological breakthrough or just overhyped? Is it a CCP-engineered publicity stunt or does it represent a genuine leap forward in AI? What I plan to do is to get to the bottom of the questions by showing you some of the hidden truth behind the company's success. So first we'll talk about what do the Chinese experts say and then we'll talk about DeepSeek's strange response to its enormous fame and then I'll talk about a genius AI girl and CCP's strategy and then we'll address the suspicious death of a former OpenAI employee while Chinese media was enthusiastic about Monday's Wall Street turmoil, describing DeepSeek's impact as similar to a Pearl Harbor sneak attack. The mysterious Chinese AI company in Hangzhou has refused any media visits. Chinese media 21st Century Business Herald reported said this. Amid this sudden surge of fame, DeepSeek has chosen to go into hibernation, avoiding any form of external communication or interaction. A Chinese investor in the company told the 21st Century Business Herald saying that people are flooding their doorsteps but we couldn't even manage to schedule a meeting with them. According to an official announcement, the company stated this. The company said we're not engaging in external project collaborations for the time being, nor providing private private deployment or related support services. DeepSeek will focus its research efforts to deliver even more powerful models. Please stay tuned. Later on Monday, DeepSeek announced that it was restricting overseas user registrations due to experience, quote, a large-scale malicious attack. I found the company's decisions to go into a dormant state after creating so much publicity odd. And one Chinese observer said the same thing. And he said DeepSeek doesn't even support registration using a U.S. phone, yet it first went viral in the U.S. And he said this playbook has been used before by other Chinese companies. And it goes like this. First, spend big money to hype it up in overseas markets, then block genuinely curious overseas users, make it impossible for outsiders to verify its claims. So this strategy feels like the company is more interested in publicity than real business growth. Now, according to Xinhua News Agency, on the day of the company's RI model launch, its founder, Liang Wenfeng, attended a closed-door meeting hosted by Chinese Premier Li Qiang that was set up for entrepreneurs and technical experts. This meeting was highlighted as a major event of the day and was broadcast on CCTV during primetime news. Liang Wenfeng's attendance at the meeting indicates DeepSeek's success is important to the Chinese government, and it's certainly part of Beijing's strategic goal of overcoming Washington's export controls and achieving self-sufficiency in AI and other industries. Now, in a rare interview that Mr. Liang gave previously, he stated that his company's biggest challenge is not funding, but rather the ban on advanced chips. As early as 2022, the U.S. government banned the export of Amadeus H100 and other advanced AI chips to China. Nevertheless, that didn't stop DeepSeek from obtaining the critical chips. During the World Economic Forum in Davos, the CEO of San Francisco-based Scale AI and the CEO's name is Alexander Wang, he told CNBC that DeepSeek has more advanced Amadeus chips than people think. He estimated that DeepSeek has around 50,000 H100 GPU, but the company can't talk about it as it would violate U.S. export control regulations. 50,000 GPU cost approximately $1.5 billion, and experts say the price would at least double if they're obtained on the black market. So what kind of startup company has $3 billion to burn in one year? And even if money wasn't the issue, 50,000 GPU represents one-tenth of Amadeus' total production of 500,000 units of H100 in 2023. So is it believable that a newly founded Chinese AI startup managed to smuggle in 10% of Amadeus' total production into China under the watchful eyes of the U.S. without the support of the Chinese government? It's not possible. This tells me that the Chinese government has been involved in this from day one. Now DeepSeek calls itself the Pinduoduo of AI, and its American version is T-Move. So the analogy raises larger questions. This isn't just about affordability. It's a deliberate strategy to undercut competitors and flood the market with low-cost alternatives, driving out higher-priced products from companies like OpenAI. Similar to the EVs, solar panels, and the steel industry, I believe DeepSeek is part of the CCP's broader strategy to dominate the global AI market with ultra-low-cost products, leveraging subsidies, government backing, and aggressive pricing to corner the market and out-compete Western rivals. This woman suddenly became famous in the past couple of days. Her name is Luo Fuli. She is hailed by Chinese media as the genius AI girl behind DeepSeek. Here you have a headline saying, this genius AI girl's smile wiped out 4.7 trillion yuan from U.S. dollars, which was the loss of market capitalization. But who is she? She was admitted by Alibaba's Damo Academy to conduct AI research focusing on pre-trained language models. This is all according to official Chinese press. And in 2022, this young lady left Alibaba's Damo Academy and joined DeepSeek's parent company. From there, she moved to DeepSeek as a deep learning researcher contributing to the development of MOE model, DeepSeek V2. After the launch of DeepSeek V2, Ms. Luo shared her thoughts on the model by saying this. She said, in terms of Chinese language proficiency, DeepSeek V2 is truly among the top-tier closed-source models globally. And she said, moreover, with a cost of just 1 yuan per million input tokens, which is only one hundredth the price of GPT-4, it is the ultimate price performance king. Now DeepSeek triggered a price war in China's LLM industry, large language model industry. This is how it earned its nickname, the Timu of AI. Some Chinese questioned the AI genius girl's credentials because she didn't start her training in AI. She didn't went to a top tech university. She went to a teacher's college in Beijing. And many people suspect that it's one of those publicity stunts because the publicity on her doesn't seem normal. And the question is, how could one AI woman beat out 700 engineers at OpenAI? You know, people find it hard to believe. Let's talk about OpenAI's former employee, Sucha Balaji. Let's talk about his suspicious death. Sucha Balaji was a 26-year-old former researcher at OpenAI. He died on November 26 in his San Francisco apartment. Police ruled his death as suicide, 40 minutes after arrival on the scene. However, his family and others raised questions about the circumstances surrounding his death. His mother claimed that her son had been in a good mood or good spirits before his death and had no history of mental health problems. She reported signs of a struggle in his apartment, including blood patterns inconsistent with suicide and a ransacked living space. The family commissioned a second autopsy, which allegedly contradicted the official findings. There are media reports, you can see like Daily Mail had a piece on this, and other prominent figures like Elon Musk and a congressman from California have also publicly called for a deeper investigation. Balaji was a whistleblower who had raised concerns about potential violations of U.S. copyright laws in OpenAI's AI training practices. He resigned from the company in August 24 and was involved in legal cases against the company. So most people who suspect his death link it to him being a whistleblower. That's where most of the suspicions lie. But that's not true according to an independent investigative journalist based in California. There we have George Webb. He has been working with Balaji's family to investigate his death. In a talk Webb gave in Silicon Valley on January 3rd, he linked DeepSeek to Sucha's death or Balaji's death and said that there is a trillion-dollar motive. He said that DeepSeek got the AI training data. In a subsequent post, he also said that Balaji and his friends had traveled to China. And here's his January 1st post where he said all of Balaji's chatGBT4 training work has just been replicated by DeepSeek in China. Now we need to understand Balaji's job at OpenAI a little bit. He was instrumental in gathering and structuring vast amounts of internet data to train GBT4, the language model behind chatGBT. He contributed to WebGBT, a precursor model that integrated web data into AI training processes. His work was pivotal in advancing OpenAI's language models. So George Webb said that Balaji's training data can help any of OpenAI's competitors get ahead. People have called for the FBI to investigate into it, and that's what I will say about this. I believe the hype in DeepSeek is a propaganda campaign orchestrated by the Chinese government.