Buffett bids farewell, OpenAI disrupts trillion-dollar market value, Google rises powerfully... A review of the top ten global business events in 2025.

date
16:55 28/12/2025
avatar
GMT Eight
In 2025, the global business landscape is being reshuffled by AI: although not listed, OpenAI leverages orders and narratives to fluctuate trillions of market value, becoming a "shadow giant" in the capital market; NVIDIA is crowned with a $5 trillion throne, and Google is challenging AI pricing power; Buffett exits the stage, Musk industrializes space exploration, and Europe urgently hits the brakes on fuel cars. Gambling and reversals are intertwined, old and new orders collide, and this year is reshaping the direction of technology, capital, and the era.
Looking back at the almost ending year of 2025, a series of iconic commercial events have left a clear and profound mark on this year. AI competitions are heating up. The White House ignites the "Stargate", making a bold $500 billion bet on the American AI infrastructure; CoreWeave puts AI computing power on the capital scale; Although OpenAI has not gone public, it leverages orders and narratives to leverage a trillion-dollar market value, becoming the "shadow giant" of the market. NVIDIA Corporation is crowned the world's first $5 trillion company, investing in Intel Corporation and acquiring Groq's inference capabilities, consolidating its computing power throne through alliance and defense. Alphabet Inc. Class C, on the other hand, launches a direct challenge to the pricing power of AI with TPU and Gemini's dual drive. On the capital stage, 95-year-old Buffett writes his "final letter", marking the end of an era. Looking further ahead, Musk puts space travel on an industrial rhythm with 155 launches. Meanwhile, in Europe, Germany hits the brakes, rewriting the 2035 ban on gas-powered vehicles to buy time for traditional industries. As the old order loosens, new forces rise. Alliances, competition, bold bets, and reversals make up the real business backdrop of 2025. At year-end, we review the top ten global commercial events of 2025 to reflect on how this year has reshaped the landscape of technology, capital logic, and the direction of the era. The White House ignites the "Stargate": $500 billion bet on the American AI infrastructure In January 2025, the "Stargate" was unveiled in the White House with much fanfare. OpenAI, SoftBank, and Oracle Corporation announced a four-year $500 billion investment, with an immediate injection of $100 billion to build 20 ultra-large AI data centers across the United States, creating the "largest AI infrastructure project in history." Masayoshi Son became chairman, showing unprecedented momentum. However, two months later, reality began to cool down. In March, construction began at the Abilene base in Texas, planning for 64,000 NVIDIA Corporation GB200 units, with an initial 16,000 units. However, the public differences in the ownership ratio, data sovereignty, and funding pace between SoftBank and OpenAI surfaced cracks in their collaboration. By May, several media outlets reported that the project had "stagnated." The $500 billion was not locked in, and the scheduled summer installation phase was delayed. To secure ChatGPT computing power, OpenAI bypassed SoftBank and independently signed a contract with a third-party data center, forcing the "Alliance Project" to give way to real-world demands. In July, the target further diminished. The company downscaled its 2025 goal to "build a small pilot center in Ohio by the end of the year", a stark contrast to the initial promise of "a billion immediately in place" at the beginning of the year. "Shell-style promotion" became the new tone. On October 30, the project announced the construction of a 1GW park in Saline, Michigan, as a symbolic realization of the "4.5GW total pie", with construction expected to start in early 2026. The scale remains, but significantly lower than the initial blueprint. In November, financing and supply chain completion progressed in sync: Blue Owl provided $3 billion in equity plus $18 billion in consortium loans; in the same month, SoftBank acquired server chip company Ampere for $6.5 billion in cash to "patch" the supply chain. By December, Abilene had become the only substantial progress. Four data centers had been topped out, with electrical and cooling system installations underway, and the plan was to go online with 400,000 GPU units by mid-2026, potentially becoming the "largest known single AI cluster." However, site selections in Arizona, California, Florida, etc., are still not formally contracted. At the same time, risk hedging accelerates. Oracle Corporation has delayed the construction of at least one data center and sold Ampere shares to cash out $2.7 billion, reducing one-sided exposure. SoftBank, facing external financing hurdles, has reduced its holdings of some NVIDIA Corporation shares to raise cash, showing that the $500 billion is far from being met. From an "epic official announcement" to "single-point landing," the "Stargate" has not been aborted but has been significantly scaled back. The conclusion is becoming clear: Abilene is poised to become a landmark for AI computing power in the U.S. in mid-2026, but the original concept of "500 billion dollars, 20 parks" has been reshaped by competition, financing, and uncertainty. The real test will come after Abilene starts production, and the operational results and a new round of capital will provide the answers. From "hoarding card business" to public pricing: CoreWeave (CRWV.US) puts AI computing power on the capital scale On March 28, CoreWeave debuted on Nasdaq at $40 per share, with an estimated value of around $23 billion, raising $1.5 billion. The stock opened at $39, and over the next four months, the price surged to a high of $187, with a market capitalization close to $90 billion. This was not just a company going public but the first time that "AI computing power leasing" was publicly priced in the market. CoreWeave's origins were not glamorous. In 2017, it was an Ethereum mine. After the crypto winter, the company transformed 250,000 NVIDIA Corporation GPUs into cloud assets, transitioning into a "pure GPU cloud", avoiding CPUs, not offering general IaaS, only doing one thing: hoarding cards and then renting them out. Its revenue formula is highly simple: priced per card, locked in long term. Major customers sign multi-billion dollar contracts for 3 to 6 years at a time, while small and medium-sized customers pay by the hour, creating high visibility cash flow. In the 11 months leading up to the IPO, CoreWeave signed a $119 billion, 5-year lease with OpenAI; in September, they added another $65 billion, bringing the total to $224 billion. During the same period, Meta locked in contracts worth up to $142 billion, while NVIDIA Corporation promised to take on $63 billion of unsold capacity to backstop demand. These three large orders accumulated to $556 billion, 29 times the revenue of 2024. Valuations also quickly soared: based on the IPO price, CoreWeave's price-to-sales (P/S) ratio was 12 times its 2024 revenue; at its peak, its P/S ratio was close to 47 times, higher than NVIDIA Corporation at the time, earning it the title of "the core of the AI bubble." Behind the controversy, the industry's coordinates have been rewritten. CoreWeave proved that an "AI computing intermediary" can go public independently and gain institutional recognition. Following this, Lambda, Crusoe, and similar companies initiated IPO consultations. Its structure of "long-term lease + GPU collateral financing" also provided the banks with a quantifiable model, driving GPU leasing ABS experiments and lowering the financing threshold of the AI industry chain. More importantly, there is a change in perception. Public market fluctuations have reversed education for customers: computing power is no longer just a hard currency that cannot be acquired, but rather an asset that can be compared in terms of rental vs. sale, flexible transferability. The industry is transitioning from "hoarding card panic" to "on-demand configuration". From mining to the main stage, CoreWeave has added "GPU as an asset, computing power as a service" to its valuation model. $23 billion is just the beginning, what is truly being priced is a new category of capital expenditure. NVIDIA Corporation (NVDA.US) invests $5 billion in its long-time rival Intel Corporation (INTC.US) NVIDIA Corporation and Intel Corporation, long-time rivals, suddenly decide to join forces. On September 18, NVIDIA Corporation announced an investment of $5 billion in Intel Corporation and a partnership to develop chips for PC and data center applications. Under the agreement, NVIDIA Corporation purchased Intel Corporation's common stock at $23.28 per share. Following the announcement of the partnership, the market reacted quickly. Intel Corporation's stock surged by 30% pre-market, while another competitor AMD dipped, dropping by more than 4%. For Intel Corporation, this was a timely boost. In recent years, it had been losing ground in the high-performance chip market and struggled to fund the high investment in advanced processes. The company had received about 10% support from the U.S. government, a $20 billion strategic investment from SoftBank, and accelerated financing through asset sales. NVIDIA Corporation's involvement further stabilized its funding chain. For NVIDIA Corporation, this was a small gamble with a big payoff. The $5 billion investment was not heavy relative to its size, but it opened up the x86 ecosystem and a broader CPU collaboration space. The core of their cooperation was aimed at PC and data center markets. Intel Corporation would integrate NVIDIA Corporation's graphics processing technology into its next-generation PC chips to enhance competitiveness against AMD; in the data center, Intel Corporation would provide general processors for AI clusters built on NVIDIA Corporation's hardware, filling gaps in acceleration chips in general computing. NVIDIA Corporation's CEO, Jensen Huang, stated that this cooperation would lay the foundation for the next generation of computing with the merger of NVIDIA Corporation's AI and accelerated computing stack with Intel Corporation's CPU and vast x86 ecosystem. Intel Corporation's CEO, Pat Gelsinger, emphasized that x86 is still the foundation of modern computing, and the company will continue to drive innovation. Both parties stated that the partnership would not change their independent strategies, and NVIDIA Corporation would continue to use Arm technology to design its own processors. This shift from a competitor to a partner is more of an inevitable outcome after years of rivalry. The $5 billion investment is not just a capital transaction, but rather a significant signal in the restructuring of the industry: in the new era of AI-dominated computing, closed defenses are becoming unsustainable, and alliances among strong players are forming the new survival rule. The roller coaster ride of the shadow giant: OpenAI stirs the AI capital narrative in 2025 In 2025, OpenAI did not go public, but it became the strongest "emotional engine" in the capital market. Funding, orders, and executive statements consistently amplified the entire AI industry chain, unfolding round after round of roller coaster market trends. From the beginning of the year until April, the story revolved around infrastructure ignition. OpenAI, in partnership with SoftBank and Oracle Corporation, launched the $500 billion "Stargate" AI computing alliance, with an endorsement from Trump in the White House. Oracle Corporation saw a single-day surge, SoftBank also skyrocketed, and the market quickly cast OpenAI as the "AI infrastructure pioneer," leading to a reevaluation of computing power, servers, and chip stocks. From May to August, cracks in the halo began to show. GPT-5, released in August, performed poorly, overshadowed by Alphabet Inc. Class C's Gemini 3.0 and xAI Grok in multiple metrics. At the same time, the mismatch between an "annual income of around $200 billion but a valuation exceeding $500 billion" was frequently mentioned, and the "AI revolution" began to be replaced by the "AI bubble," causing a noticeable cooling of concept stock trends. In September, OpenAI announced that Luxshare Precision Industry would be manufacturing its first consumer-grade AI hardware, leading to a trading limit increase for Luxshare; in the same month, Oracle Corporation's remaining performance obligations rose to $455 billion, its stock price hitting a new high, pulling emotions back to a high level. In October, financial issues stirred up market sentiment. SoftBank recorded a $14.6 billion unrealized gain from holding OpenAI shares, achieving record quarterly net profits, and a stock price surge. However, OpenAI's acquisition of financial AI application Roi and its deep ties with AMD led outsiders to start calculating its cash burn and financing intensity. The CDS market saw default swap trades of several tech giants for the first time, bringing debt risks to the forefront. On November 7, emotions exploded. With a CFO's statement hinting at a need for federal government financing for chip funding, interpreted as "even the strongest AI companies lack money." That night, the market saw a total evaporation of about $500 billion from the six major U.S. tech giants and computing suppliers. CEO Altman clarified overnight that "government guarantees are not needed", but this only resulted in a limited rebound, causing the market to suffer from a "trust deficit". By December, the competitive landscape had reversed. Alphabet Inc. Class C's Gemini continued to gain positive reputation, with the Alphabet ecosystem doubling within the year; in contrast, the "OpenAI series," including Oracle Corporation and SoftBank, dropped nearly 40% from their highs, with a market capitalization shrinkage of over a hundred billion dollars, transforming the "OpenAI concept" from a premium label to a risk exposure. Looking back over the year, OpenAI's "stock price effect" was pushed to a boiling point by the grand infrastructure and hardware narratives, then continued to be stirred by model competition, debt worries, and statement errors. The capital logic has shifted from "changing the world" to "realizing cash flows." The ability to propel revenue of about $200 billion toward sustainable growth will determine whether the AI sector in 2026 will mature or continue in a cycle of overestimated values and high volatility. NVIDIA Corporation becomes the world's first "$50 trillion company" On October 30, NVIDIA Corporation's stock price rose by about 3% to $207.16, bringing its market capitalization to $50.3 trillion, becoming the first company in the world to surpass this milestone. This figure surpasses the total market capitalization of AMD, Arm, ASML, Broadcom Inc., Intel Corporation, Lam Research Corporation, Qualcomm, and Taiwan Semiconductor Manufacturing Co., Ltd., and is greater than the combined market capitalization of major stock indices in utilities, industries, and essential consumer goods in the S&P 500. Over the past six months, NVIDIA Corporation's stock price has accumulated approximately 90% gains. The current scale even surpasses the market capitalization of major stock indexes in Germany, France, and Italy combined. Three years ago, when ChatGPT had not yet been introduced, NVIDIA Corporation's market capitalization was only about $400 billion. Subsequently, the demand for GPUs for training and running large models surged, with market capitalization exceeding $1 trillion within a few months, accelerating: in February 2024 to $2 trillion, in June $3 trillion, in July of 2025 to $4 trillion, and in October surging past $5 trillion. Demand is the core driving force behind this leap. NVIDIA Corporation disclosed that its Blackwell chip, released last year, has already shipped 6 million units, with another 14 million units on order. Jensen Huang forecasted at the GTC conference that total sales over the next five quarters would reach $500 billion. However, with the rapid rise comes increasing doubts. Some investors are comparing the current AI stock trends to the dot-com bubble: companies are investing billions of dollars, increasing debt burdens, while real income remains limited. The valuation is also under pressure, with NVIDIA Corporation's stock price about 33 times next year's expected earnings, higher than the S&P 500's average of around 24 times. From $400 billion to $50 trillion, NVIDIA Corporation has achieved a "10x stock" leap in three years. It has not only reshaped its destiny but also defined the computing power coordinates in the AI era. However, as it soars past trillion-dollar heights, the growth story must continue to materialize in order to justify this shiny crown. Buffett's "final letter": I'm "just lucky", but "Father Time" is catching up, I will "keep quiet" On November 10, 2025, 95-year-old Buffett wrote what he called his final annual shareholder letter of his life. The phrase "I'm going quiet" signals his retirement as CEO and withdrawal from daily management at the end of the year. This legendary investor, who has led Berkshire for almost sixty years, is officially reaching his final curtain call. The baton is passed to the new CEO, Greg Abel. Buffett hails him for his excellent management, diligence, and honest communication, and jokes that he hopes he "keeps at it." From then on, the annual letters will be written by someone else, but Buffett promises to continue writing to shareholders every Thanksgiving, maintaining an emotional bond with Berkshire. The most moving part of this "farewell letter" is Buffett's reflection on his life. At 95, he is grateful for his "bursting luck": born in America in 1930, having enjoyed health and the dividends of his time. Almost dying of appendicitis in 1938, he even asked a nun for fingerprints in the hospital, imagining himself solving an FBI case. Reflecting on his partnership with Charlie Munger and many other friends, he jokingly remarks that "maybe there is magic in the water of Omaha." Now, with a slower pace and struggling to read, he continues to work every week and occasionally strikes gold with new ideas. In his moment of goodbye, he once again highlights his business wisdom, specifically targeting corporate greed. Buffett criticizes the disclosure of executive compensation that triggers a competition of "who's wealthier": often, what bothers very wealthy CEOs is that someone else is wealthier. He advises Berkshire to avoid leaders who aspire to retire at 65, seek to be "outstandingly wealthy", or attempt to establish a "dynasty". His long-termism also contrasts with the current speculative trend. Starting from Berkshire facing financial difficulties in 1962, he developed it into a business empire spanning insurance, manufacturing, utilities, railways, and brands like Dairy Queen and Fruit of the Loom, emphasizing the importance of avoiding paths that may turn a company into a "mendicant". One letter, wrapping up a whole life. Saying farewell to power, holding onto principles, giving back to society. For the market, this is a historical transition for Berkshire; for investors, this is a final demonstration of data-driven and rational investing. 155 launches later, Musk sends space travel into the "industrial era" In 2025, SpaceX redefined the meaning of "space-scale" with an almost feverish launch rhythm. By November 22, Falcon 9 had completed 150 orbital missions, along with 5 Starship test flights, totaling 155 launches for the year, with an average of one rocket launch every 2 to 3 days. Just one site at Cape Canaveral accounted for 94 launches, about 40% of the global total. The aerospace industry is presenting the rhythm of an industrial assembly line for the first time. The cornerstone of Musk's strategy was the concept of "reusability." Booster B1067 went on 31 successful flights, setting a record at Shanghai New World; as more than 20 boosters in the active fleet exceeded 10 flights, a certification goal of 40 flights per booster was being pursued. With 532 first-stage recoveries and 454 re-flights during the year, the success rate exceeded 97%. The fairing and control surfaces were simultaneously reused, reducing the marginal cost of a single launch to the range of $150,000 to $200,000, transforming rockets from "expensive equipment" into "assets with a quick turnover." On the other end, the