AI technology gap is disappearing, the battle over debt and revenue timing is turning into a crisis for Oracle
With the AI technology gap disappearing, the battle has become a battle over debt size and revenue timing. Oracle, an investor in OpenAI's one client, is facing a crisis, and its stock price has fallen. Contrary to its stated policy, OpenAI is securing funding through fraudulent "brokerage fee" advertising. The New York Times reported on the 11th that Oracle's stock price plunge isn't simply due to Oracle's rapidly increasing AI spending. "The question is whether all of these investments, largely for one client, OpenAI, will yield results."
In November, during the AI boom sparked by OpenAI, tech giant Google boasted that its new and improved AI model, Gemini 3, surpassed the technology of the upstart OpenAI, becoming the world's best. However, less than a month later, OpenAI unveiled its own model, GPT-5.2, claiming it was "the best model ever for professional use in the real world."
The New York Times responded, "OpenAI said its technology surpassed several industry-standard benchmarks in computer programming, mathematics, and science." "But for many industry experts, the real story is that the technological gap between OpenAI's so-called baseline AI model and all others has virtually disappeared," the Times concluded, diagnosing the technological gap as narrowing and expanding on various fronts.
The competition from now on is how much funding OpenAI is spending in this intense AI race, and how much it is spending. Oracle, which has been investing heavily in OpenAI, saw its stock price drop 13% on the 11th, while Japanese technology investor SoftBank, which has been betting heavily on AI, fell about 7.6% in Tokyo.
Stock market concerns have surfaced that it will take longer and more money for Oracle's AI bet to pay off.
Oracle expects capital expenditures to reach about $50 billion by May 2026, an increase of $15 billion from its September forecast.
Oracle said it burned through $10 billion in cash in the third quarter, which meant its revenue and profit for the period fell short of analysts' expectations.
The New York Times noted that Oracle already has about $106 billion in debt, which Morgan Stanley analysts predict will grow to $290 billion by 2028. This massive debt burden is significant. “Investment is weighing on Oracle,” he said, adding that “the cost of protecting the company’s debt against default through credit default swaps is now around 1.25%, a roughly threefold increase since September.”
OpenAI, which sparked the AI boom, has been a prominent presence and an industry leader for three years, but now technology analysts and investors are increasingly raising concerns about its multi-trillion-dollar spending plans and whether wealthy competitors like Google are catching up.
Because OpenAI is privately held, investors are focusing on its investors, including its publicly traded partners Oracle and SoftBank, driving down the stock price.
The New York Times reported that “OpenAI accounts for the majority of Oracle’s unrecognized $523 billion in contractual revenue.” Oracle executives noted on the 10th that the company has also partnered with Meta and Nvidia, and that other customers can use the capabilities built for OpenAI. This serves as a reminder that the AI ecosystem is becoming increasingly interconnected, and that disruption to a major hub like OpenAI can have significant consequences.”
Altman, CEO of OpenAI, has repeatedly stated that OpenAI will achieve $20 billion in annual revenue, but OpenAI is still far from profitability.
Conversely, the company has announced that it will invest $1.4 trillion in the computing power needed to build and deploy various AI technologies over the next few years.
When OpenAI launched its ChatGPT chatbot in late 2022, sparking the AI boom, the San Francisco startup had a clear lead over its competitors, a lead it maintained for more than two years.
However, over the past 12 months, companies in the United States and China have built technologies that rival or even surpass OpenAI’s leading models.
“The general shape and form of building foundational models is well understood, and it’s happening in roughly the same way across all major AI research labs,” said Rajan Krishnan, CEO of Vals AI, a company that tracks the performance of cutting-edge AI technologies. "Consumer AI is OpenAI, but if it disappears, the company's value will fall significantly," he told the NYT.
A week after Google's new AI model was released, San Francisco startup Anthropic released a new model, Claude Opus 4.5, that is on par with OpenAI's technology.
New York startup Runway released a model on the 11th that outperformed OpenAI's video generation technology, Sora, according to standard industry benchmarks.
Like its competitors, OpenAI has continued to improve its technology, with its new GPT-5.2 model showing significant improvements in computer code generation and task performance in specific fields such as healthcare and finance.
OpenAI said it will charge customers about 40 percent more for the 5.2 model than its predecessor, the NYT reported. "The model arrived just hours after Disney announced its investment in OpenAI and agreed to license its characters for use in Sora," the source said.
Regarding the technological competition, a source said, "Google unveiled an improved chatbot in mid-November." Altman then sent a "code red" memo to OpenAI employees, urging them to focus on improving ChatGPT and put new projects on the back burner, according to a person familiar with the matter, who spoke to the Times on condition of anonymity.
Altman is now focused on protecting a core part of the company's business.
According to research firm Similarweb, ChatGPT is used by more than 800 million people each week, representing a 76% market share, and this "consumer AI" has lost its scarcity value due to technological competition.
The Times reported that Altman urged OpenAI employees in the memo to improve ChatGPT's speed, reduce the number of questions it refuses to answer, and provide more ways for people to personalize the chatbot to their needs. He said these changes should take priority over the company's new efforts in advertising, shopping, and healthcare, and that the company would, at least temporarily, "dedicate some staff from other teams to this multi-week ChatGPT improvement effort."
Portions of Altman's memo were previously reported by The Information and the Wall Street Journal. Currently, ChatGPT is OpenAI's primary source of revenue, with about 6% of its 800 million users paying $20 per month to use a more advanced version of the chatbot.
As Altman noted in the memo, OpenAI also aims to monetize the free version of ChatGPT.
OpenAI plans to prioritize ChatGPT development in the coming weeks, with the company preparing for a larger release early next year after the release of the new GPT-5.2 model this week.
The New York Times reported, "Altman frequently sends 'rally the troops' memos to his company, and his latest message resembled one sent in January after the Chinese startup DeepSeek surprised the world with its chatbot technology." "Following the DeepSeek threat, OpenAI spent several weeks focusing on improving its own chatbot before shifting gears to new endeavors, including monetizing the free version of ChatGPT."
Cash-strapped OpenAI, despite its public service mission, has been advertising ChatGPT to generate illicit revenue. The New York Times reported that “chatbots are not as well-suited to advertising as traditional web pages or searches,” and that “chatbots deliver prose rather than a list of blue links expanded to a few internet addresses from advertisers, and OpenAI has long experimented with various ways to deliver ads unobtrusively.”
The Times continued, “They are also exploring more focused forms of advertising, allowing people to purchase goods and services through chatbots.” “If someone uses the chatbot to buy a ceramic vase from a seller on Etsy, OpenAI takes a cut of the transaction as a commission.”
Over the past 12 months, OpenAI has been using a technique called reinforcement learning to improve the performance of its AI models in specific areas, such as mathematics and computer programming, but this essentially involves learning specific skills through trial and error.
The Times reported that “technologies specifically designed to generate computer code have also become another important source of revenue for OpenAI,” adding that “some software developers and researchers pay $200 a month to use OpenAI’s most advanced coding technology, which is partly due to user companies trying to improve their performance in these areas.” "It's an effort," he said.
OpenAI is currently using similar techniques to train models for use in healthcare and finance, allowing it to sell its products to companies in those markets.
Weilin Qiang, founder and chief technology officer of LMArena, told the New York Times, "Coding is changing the way people in Silicon Valley work, and this will happen in other industries as well."
LMArena is a company that tests the performance of AI systems.
According to Vals AI's latest benchmark tests, OpenAI's models outperform other leading systems in both legal and financial tasks, demonstrating OpenAI's focus on these areas.
The New York Times noted, "OpenAI has worked hard to build a business selling these products to companies by building the enterprise software divisions of tech giants like Google. As a startup, OpenAI is at a disadvantage when competing with Google, which can bundle its cutting-edge AI technology with legacy office apps like Google Docs and Gmail by selling to businesses." The Times also pointed out the limitations of its monetization strategy.
OpenAI is transforming from an AI startup into an IT company competing with Google.
To counter its disadvantage, the young company launched its own web browser specifically designed for use with AI technology this fall, directly challenging Google's Chrome browser.
"This is a multi-front game," Rajan Krishnan, CEO of Vals AI, a company that tracks AI performance, told the New York Times.
OpenAI executives describe it as two companies: one that generates revenue from consumers and the other from businesses, and they argue that AI technology will reshape virtually every business software category.
In contrast, OpenAI's diversification strategy, which involves creating multiple fronts and expanding over time, has led to a burn through of investment funds and, in the long run, a significant increase in debt.
Deputy Prime Minister Baek Kyung-hoon, in his work report to President Lee Jae-myung for next year, stated, "To achieve our goal of becoming one of the top three AI powerhouses, we will secure an independent AI model ranked among the world's top 10 next year and provide it as open source to businesses and academia."
He also announced "administrative assistance" as a key AI project among the top 10 AI projects for the people, including "AI National Tax Information Consultant, Agricultural Product Savings Information, National Heritage Interpreter, and AI Permit Assistant," thus appearing to completely break away from the technological competition among AI companies.
Before his nomination, Deputy Prime Minister Bae stated, "The world is leading the AI industry at the government level to secure AI competitiveness." After his nomination, he spoke of Korea's "A strength," saying, "With a little more improvement in computing infrastructure and data, we can develop world-class AI technology and create a service ecosystem."
See <Bae Kyung-hoon's 'AI China Model,' Blackstone's 'AI Real Estate Investment,' and Alibaba's 'Overinvestment,' June 24, 2025>