No wonder Buffett finally bet on Google.

Buffett bought AI-themed stock Google for the first time, investing at a high premium of about 40 times free cash flow. With its Gemini 3 model and self-developed TPU chip, Google has reversed the AI competition and challenged the market position of OpenAI and Nvidia. THIS ARTICLE IS DERIVED FROM AN ARTICLE WRITTEN BY CHANNELWU, AND HAS BEEN COMPILED AND REWRITTEN BY SHEN CHAO. (Synopsis: Buffett waited six years to finally buy Google!) Smashed $4.3 billion, Alphabet became Berkshire's tenth largest holding) (Background supplement: From China's Warren Buffett Duan Yongping, find the key to “holding and getting rich” in the currency circle) “Google holds the whole chain in its own hands. It does not rely on Nvidia and has efficient and low-cost computing power sovereignty.” Warren Buffett once said, “Never invest in a business you can't understand.” However, when the “era of stock gods” was about to end, Buffett made a decision that violated the “house rules”: buy Google stock at a high premium of about 40 times free cash flow. Yes, Buffett bought “AI stocks” for the first time, not OpenAI, not Nvidia. All investors are asking the question: What is Google? Back to the end of 2022. At that time, ChatGPT was born, and Google's senior management sounded a “red alert”, they kept meeting, and even urgently recalled the two founders. But at that time, Google looked like a slow-moving, bureaucratic dinosaur. It hastily launched the chatbot Bard, but made a factual mistake in the demo, and the company's stock price plummeted, wiping hundreds of billions of dollars in market value. Then, it integrated its AI team to launch the multimodal Gemini 1.5. But this product, which was regarded as a handicap, only set off a heated discussion in the technology circle for a few hours, and was taken away by OpenAI's subsequent video generation model Sora, and quickly became uncaring. It is slightly embarrassing that it was the groundbreaking academic paper published by Google researchers in 2017 that laid a solid theoretical foundation for this round of AI revolution. The “Attention Is A You Need” paper proposes a Transformer model opponent that mocks Google. OpenAI's CEO Altman doesn't see Google's taste, “I can't help but think about the aesthetic differences between OpenAI and Google.” Google's former CEO also complained about the company's laziness, “Google has always believed that work-life balance … It's more important than winning the competition.” This series of dilemmas also makes people suspect that Google has fallen behind in the AI competition. But change is finally here. In November, Google unveiled the Gemini 3, which outperformed competitors in most benchmarks, including OpenAI. More importantly, Gemini 3 is completely trained by Google's self-developed TPU chips, which have now been positioned by Google as low-cost alternatives to Nvidia GPUs and officially sold to external customers. Google shows its edge on two fronts, responding positively to OpenAI's software front with the Gemini 3 series; The other is the hardware front that challenges Nvidia's long-standing dominance with TPU chips. Kick OpenAI, punch Nvidia. Altman felt the pressure back last month, saying in an internal letter that Google “could create some temporary economic headwinds for our company.” And this week, after hearing that the big manufacturers bought TPU chips, Nvidia, whose stock price once plummeted 7% intraday, had to personally send a letter to calm the market. Google CEO Sundar Pichai said in a recent podcast that Google employees should catch up on sleep. “From an external point of view, we may seem quiet or backward during that time, but in fact, we are consolidating all the basic building blocks and building on them.” Now the situation has been reversed. “We have now reached an inflection point,” Pichai said. At this time, ChatGPT was released exactly three years ago. In the past three years, AI has kicked off a feast of Silicon Valley capital; Under the feast, bubble worries emerge, has the industry ushered in an inflection point? On November 19, Google released the latest artificial intelligence model, Gemini 3. One test data shows that in the vast majority of tests covering expert knowledge, logical reasoning, mathematics, and image recognition, Gemini 3 scores significantly ahead of the latest models from other companies, including ChatGPT. Only in the only programming proficiency test, its performance was slightly inferior, ranking second. The Wall Street Journal said, “Call it the next generation of top models in the United States.” Bloomberg says Google is finally awake. Musk and Ultraman praised it. Some netizens joked that this is Ultraman's ideal GPT-5. After trying Gemini 3 early, the CEO of cloud-based content management platform Box said the performance improvement was so incredible that they once suspected that their measurement method was wrong. However, repeated tests confirmed that the model won by double digits in all internal evaluations. Salesforce's CEO says he used ChatGPT for three years, but Gemini 3 flipped his perception upside down in just two hours: “Holy shit… Can't go back. This is simply a qualitative leap, reasoning, speed, graphic film processing… It's all sharper and faster. It feels like the world has turned upside down again.” Gemini 3 Why is Gemini 3 so outstanding, and what has Google done? The Gemini project lead posted, “Simple: improved pre-training and post-training.” Some analysts say that the pre-training of the model is still following the logic of Scaling Law - by optimizing the ( of pre-training ) such as larger data, more efficient training methods, more parameters, etc., the model capability is improved. The person who wants to know the secrets of Gemini 3 the most is Ultraman. Last month, before the release of Gemini 3, he was vaccinated in an internal letter sent to OpenAI employees, “Google has done a very good job in every respect”, especially in pre-training, Google's progress may “bring some temporary economic headwinds” for the company, “the outside atmosphere will be more severe in the next period.” While ChatGPT still has a significant advantage over Gemini in terms of user volume, the gap is narrowing. In the past three years, the number of ChatGPT users has grown rapidly. In February, it had 400 million weekly active users, and this month, it jumped to 800 million. Gemini publishes monthly active users, which jumped to 650 million in July from 450 million in July. With the global network search…