[Trump Stokker] Open AI 'all-in' while crying, but 'emergency management' without money

Recently, when Google, a search engine and operating system (OS) dinosaur, shook the AI market with “Gemini 3.0” using tensor processing devices (TPU), OpenAI, the developer of “Chat GPT,” declared emergency management. This is why there are signs of perceptual changes in the AI market, which has led NVIDIA's graphics processing unit (GPU) and OpenAI's chat GPT service for the past 3 years. Other US companies, such as Amazon and Enropic, and Chinese AI companies are also facing intense challenges. In addition to this, Wall Street is also paying attention to the actions of SoftBank Group and Chairman Son Jeong (Japanese name Masayoshi Son), who sold all NVIDIA shares last month for the purpose of betting on open AI. This is due to doubts about whether the future AI industry will result in open AI and the company's large-scale data center. Unlike competitors such as Google, there is no source of cash generation (cashcow), so the “circular transaction” label always follows every investment, which is also considered a burdensome factor in attracting investment in open AI.

━OpenAI, “Code Red” when Google's “Gemini 3.0” pops up... the opposite position after 3 years

According to the Wall Street Journal (WSJ) on the 2nd (local time), Sam Altman (CEO) of Open AI announced that he will distribute memos to employees on the 1st and issue a “red alert (code red)”, and that the launch of other services will be postponed for the time being to focus on improving chat GPT quality. Services that CEO Altman announced he would put off include advertising, health and shopping AI agents, and “Pulse,” a personal assistant service. CEO Altman then announced that he will temporarily induce personnel relocation and have daily meetings with people in charge of improving chat GPT performance.

CEO Oltman took such a nervous attitude because the performance of Gemini 3.0, which was unveiled by Google on the 18th of last month, was unexpectedly excellent. Geminai 3.0 scored better scores than “GPT-5.1,” the latest model of chat GPT, as well as “Glock” of startup xAI led by Tesla CEO Elon Musk in various performance tests such as reasoning ability and coding ability.

Furthermore, unlike existing central processing units (CPUs) and GPUs, Google put even more pressure on open AI by driving Gemini 3.0 with its own TPU designed to process only AI computations at ultra-high speed without performing general-purpose tasks. Currently, Google is the only company in the world that has vertically integrated AI models, semiconductors, and consumer/enterprise platforms all internally. Open AI relies on NVIDIA for semiconductors and other service companies for platforms. After Facebook's parent company Meta considered ways to introduce TPU, the popularity of Open AI and NVIDIA, which had dominated the market, was shaken once and for all. Anthropic, the operator of the AI chatbot “Claude,” which is considered a rival of Open AI, recently signed a cloud usage contract equipped with 1 million Google TPUs.

OpenAI is a company that has never missed the status of “best generative AI” since introducing Chat GPT to the world on November 30, 2022. In December 2022, Google CEO Sundar Pichai was shocked by the rise of Chat GPT in reverse and issued Code Red within the company. CEO Phichai determined that Chat GPT would seriously threaten the search engine market, which was dominated by Google at the time. After 3 years, the two companies' positions were completely reversed. Nick Turley, vice president of Open AI, emphasized on X (formerly Twitter) on the 2nd that “our current focus is to make Chat GPT more effective, continue to grow, and expand access to the world,” and “we will make it feel more intuitive and personalized.”

━Anthropic and Amazon go on an onslaught with DeepSeek... ChatGPT 'shakes'

This is not the only company that fearfully pursues Chat GPT. Anthropic also recently introduced “Claude Opus 4.5,” the latest version of the company's top AI model, “Claude Opus 4.5.” On the 2nd, the British Financial Times (FT) quoted a source and reported that Anthropic has appointed Wilson Sonsini Law Office to prepare for an initial public offering (IPO). This law firm is a company that has had an advisory relationship with Anthropic since 2022. He has also been involved in IPOs for technology companies such as Google, Linkedin, and Rift. Last year, Anthropic also recruited Krishna Rao, Chief Financial Officer (CFO), who played a key role in Airbnb's IPO.

According to FT, Anthropic has not yet selected a listing organizer, but recently prepared an internal checklist and is discussing potential IPOs with major investment banks (IB). Depending on the situation, it may be possible to enter the stock market earlier than Open AI. The corporate values assessed by Open AI and Anthropic themselves are known to be at the level of 500 billion dollars (about 730 trillion won) and 300 billion dollars (about 440 trillion won), respectively.

According to the Chinese government agency Xinhua News Agency and Hong Kong's South China Morning Post (SCMP) on the 2nd, DeepSeek, a Chinese generative AI startup, also officially launched the latest model “DeepSeek V3.2” and the high-computing-specific model “DeepSeek V3.2-Speciale” on the 1st of this month. DeepSeek revealed that DeepSeek V3.2 showed performance comparable to Open AI's “GPT-5” through enhanced learning and expansion of computational capabilities. They also claimed that DeepSeek V3.2-Speciale surpassed GPT-5 and had a similar level of reasoning ability to Google's “Gemini 3.0 Pro.” It was acknowledged that compared to Gemini 3.0 Pro, it is noticeably inferior in terms of token (basic unit of word or sentence) efficiency.

Amazon also unveiled its own AI chip, “Nova 2,” a new version of its AI model “Nova 2,” along with its AI chip “Traineum 3,” and a “Nova Forge” service that supports individual companies to create their own AI models at the annual cloud computing event held in Las Vegas, USA on the 2nd. Amazon Web Services (AWS) introduced that Traineum 3 has more than 4 times better computing performance than its predecessor, “Traineum 2,” and energy consumption has been reduced by about 40%. He also explained that AI model training and operation costs can be reduced by up to 50% compared to using NVIDIA's GPU. AWS announced that it has already begun developing the sequel “Traineum 4,” which has more than 3 times the performance of Traineum 3. Matt Garman, AWS Chief Executive Officer (CEO), emphasized that “Traineum 3 is the industry's most cost-effective in large-scale AI training and inference.”

━NVIDIA “The investment of 100 billion dollars cannot be confirmed”... Son Jeong's “all-in” strategy is also uneasy

As the popularity of open AI is flourishing, NVIDIA and other companies that wanted to create an ecosystem centered around this company are also showing slightly different attitudes. This is because Wall Street's doubts about profitability are growing since it is likely that open AI will only see a deficit until at least 2030. Since CEO Altman himself launched a “bubble theory” in August, every company announcing the news of investing in open AI has repeatedly plummeted due to suspicions of circular transactions. OpenAI seems to have become the main company that summoned the nightmare of the “.com bubble (Internet industry bubble)” in the mid to late 1990s. A circular transaction is a contract consisting of a specific company funding open AI, and open AI purchases the company's products and services with the profits obtained there.

According to Reuters on the 2nd, NVIDIA CFO Colette Kress said at the “UBS Global Technology and AI Conference” held in Arizona, USA, that they have not yet signed a 100 billion dollar (approximately 147 trillion won) data center investment agreement promoted by OpenAI. Earlier, on September 22 of this year, NVIDIA announced that it would invest up to 100 billion dollars to build a 10 gigawatt (GW) data center in collaboration with Open AI, and it received intense suspicions about circular transactions. CFO Cress added that the details of the contract with OpenAI were not included in the published pre-order figures, and that contracts related to Anthropic could also be added to the numbers. NVIDIA's stock price crossed the $180 mark for the first and last time this month due to this news.

Jensen Hwang, CEO of NVIDIA, is also responding sensitively to the recent suspicion of a circular transaction with Open AI. CEO Huang is yen Vidia held an online press conference on the 1st to announce that it has decided to acquire about 2.6% of the semiconductor design software (SW) company Synopsis for a total of 2 billion dollars (approximately 2.94 trillion won), and highlighted that “this investment was not linked to a semiconductor purchase agreement.”

As the situation looked like this, Wall Street's eyes turned to Chairman Son, who gave everything to Open AI. According to Japan's Nihon Keizai Shimbun (Nikkei), Chairman Son criticized “people who ask if AI is a bubble are not smart enough (not smart enough)” during the “Future Investment Initiative” of the international finance conference held in Tokyo on the 1st, led by Saudi Arabia. Chairman Son said, “AI will earn 10% of the world's gross domestic product (GDP) after 10 years, and the amount will be 20 trillion dollars (about 2 trillion won) per year,” and emphasized that “if you invest 10 trillion dollars over 10 years, you can recover it in just half a year.” At the same time, regarding the sale of all NVIDIA shares held by SoftBank Group in October of this year, he said, “I cried and decided to eat mustard to invest in open AI, etc.,” and “Actually, I didn't want to sell even one week.” SoftBank sold all 32.1 million shares of NVIDIA worth over 5.8 billion dollars in October to focus on the Open AI “Stargate” project, which invests 500 billion dollars (about 730 trillion won) over 5 years to build AI data centers across the US.

━The 'circular transaction' bubble theory if you just invest... pay attention to whether the '.com bubble nightmares' will go away

Despite Chairman Son's large-scale investment, Wall Street's doubts about Open AI are still a financial issue. This is because it is not comparable to competitors such as Google in terms of its ability to generate cash, and the IPO process is poor. Shim Jia is now at a point where suspicion of circular transactions is attached every time Open AI receives an investment or is executed. Even though other types of investment are not easy from the standpoint of a company that requires large-scale investments and cannot make money.

Even when OpenAI announced that it would invest in shares in Thrive Holdings on the 1st of this month, Wall Street questioned the circular transaction structure as if they had been waiting. OpenAI discussed the size and amount of shares in Thrive Holdings, but Bloomberg Communications and CNBC pointed out that “OpenAI has added another circular transaction with this investment.” Earlier, OpenAI also caused controversy over circular transactions when investing equity in partner companies such as AMD and Coreweave.

Thrive Holdings is a company founded in April of this year by Thrive Capital, which was founded by CEO Josie Kushner in 2010 to establish or acquire AI-related companies. Thrive Capital is a venture investment company famous for its strategy of investing intensively on a large scale in a small number of companies and holding shares for many years. According to Bloomberg News Agency, Thrive Capital initially invested in open AI with a corporate value of approximately 27 billion dollars (approximately 39.7 trillion won) in 2023, and then led the financing process.

After the assessment that Google actually overtook Chat GPT, OpenAI also revealed that it had decided to acquire Neptune AI, a Polish startup that inspects AI model training courses on the 3rd. Neptune AI is a company that makes software that helps detect problems that occur while supervising and analyzing AI model training processes. It is solved as a paving stone for speeding up model training and correcting problems in advance using Neptune AI's software. OpenAI did not disclose specific transaction terms, such as the acquisition price, on this day. Jakub Pachoki, chief scientist at Open AI, said, “We plan to iterate several times to deeply integrate Neptune AI's tools into our AI model training,” and hoped “we will be able to watch how the model learns through this.”

On Wall Street, although Open AI is being threatened by competing companies, it seems that it is still not at a level where it will be driven to the edge of a cliff in a short period of time. However, it seems clear that the dominant position of chat GPT will actually end after 3 years. It is said that for the time being, there is a high possibility that American companies as well as Chinese companies will compete fiercely for AI model leadership. Furthermore, Chairman Son is a person who had the experience of being the richest person in the world during the dot-com bubble and lost 99% of all wealth due to the collapse of the bubble. Chairman Son later stepped up together by investing in telecommunications and Internet companies and became involved in the myth of the success of Alibaba in China and Coupang in Korea (headquartered in the US). Whether Chairman Son's open AI betting will be another success story based on past failures is expected to be a topic of conversation in the future. Based on his close relationship with President Donald Trump, Chairman Son also planned a golf event attended by 70 CEOs of major global companies at Marlago Resort in Palm Beach, Florida, USA from October 17 to 19. This was an event to raise political funds for President Trump on the occasion of the 90th birthday of Gary Player, a legendary golfer from South Africa.

※“ Trump Stocker (Stocker)” is a corner that conveys field stories and analysis of current issues related to the US market, business, policy, politics, and diplomacy that may help with investment during the time of President Donald Trump. Subscribe to receive great news about the US.