Buffett Makes Exception Betting on Google! Buys AI Stock at 40x Premium to Challenge Nvidia's Dominance

Buffett once said, “Never invest in a business you can’t understand.” Yet as the “Oracle of Omaha era” draws to a close, Buffett has made a decision that goes against his own rule: buying Google stock, and at a hefty premium of about 40 times free cash flow. This is Buffett’s first purchase of an AI-themed stock—not OpenAI, not Nvidia.

From Red Alert to Google AI’s Comeback

Let’s go back to the end of 2022. At that time, ChatGPT burst onto the scene, prompting Google’s top executives to sound a “red alert.” They held constant meetings and even urgently recalled the two company founders. Back then, Google looked like a sluggish, bureaucracy-ridden dinosaur. It rushed to launch its chatbot Bard, but the demo contained factual errors, causing the company’s stock price to plummet and wiping out tens of billions of dollars in market value in a single day.

Next, it merged its AI teams and released the multi-modal Google AI. But this product, seen as its trump card, only sparked a few hours of buzz in tech circles before being overshadowed by OpenAI’s subsequent launch of the video-generation model Sora, quickly fading from attention. Somewhat embarrassingly, it was Google researchers who published the groundbreaking 2017 academic paper “Attention Is All You Need,” proposing the Transformer model that laid the solid theoretical foundation for this AI revolution.

Competitors mocked Google. OpenAI CEO Sam Altman was unimpressed with Google’s taste: “I can’t help but think about the aesthetic differences between OpenAI and Google.” Google’s former CEO was also dissatisfied with the company’s laziness: “Google has always believed that work-life balance… is more important than winning competition.” This string of predicaments led many to question whether Google had fallen behind in the AI race.

But change finally arrived. In November, Google AI version 3 was released, outperforming its competitors—including OpenAI—on most key benchmarks. One dataset showed that in nearly all tests covering expert knowledge, logical reasoning, mathematics, and image recognition, Google AI version 3 scored significantly higher. Only in the one programming ability test did it fall slightly behind, ranking second.

The Wall Street Journal wrote, “You could call it America’s next-generation top model.” Bloomberg said Google had finally woken up. Elon Musk and Sam Altman both praised it. Some netizens joked that this is Altman’s ideal version of GPT-5. Box CEO Aaron Levie, who got early access to Google AI version 3, said the performance improvement was so incredible they began to doubt their own evaluation methods.

Salesforce’s CEO said he had used ChatGPT for three years, but Google AI version 3 overturned his perceptions in just two hours: “Holy shit… there’s no going back. This is truly a qualitative leap—reasoning, speed, handling of text, images, and video—it’s all sharper and faster. It feels like the world has been upended again.”

Why did Google AI version 3 perform so well? The project lead at Google AI posted, “Simple: improved pre-training and post-training.” Analysts noted that the model still follows the logic of the Scaling Law—by optimizing pre-training (such as using larger datasets, more efficient training methods, more parameters, etc.), the model’s capabilities are enhanced.

TPU Chips Crack Nvidia’s Fortress

A month ago, Nvidia’s market cap surpassed $5 trillion, as the market’s enthusiasm for AI pushed this “AI arms dealer” to new heights. But the TPU chips used by Google AI version 3 have cracked open a breach in Nvidia’s stronghold. The Economist, citing data from investment research firm Bernstein, reported that Nvidia’s GPUs account for more than two-thirds of the total cost of a typical AI server rack; in contrast, Google’s TPU chips cost only 10% to 50% of equivalent-performance Nvidia chips.

These savings add up to a significant amount. Investment bank Jefferies estimates that Google will produce about 3 million of these chips next year, nearly half of Nvidia’s output. Last month, well-known AI startup Anthropic planned to adopt Google’s TPU chips at scale, with the reported deal worth tens of billions of dollars. Reports on November 25 indicated that tech giant Meta is also in talks to adopt TPU chips in its data centers by 2027, in a deal valued at several billion dollars.

The history of TPU dates back more than a decade. At the time, Google was developing an in-house accelerator chip to improve the efficiency of search, maps, and translation. In 2018, it began selling TPUs to cloud computing customers. Since then, TPUs have also been used to support Google’s internal AI development. During the development of models like Google AI, the AI and chip teams worked interactively: the former provided real-world needs and feedback, and the latter customized and optimized TPUs, which in turn boosted AI R&D efficiency.

Core Differences Between TPU and Nvidia GPU

Cost advantage: TPU chips cost only 10% to 50% as much as equivalent-performance Nvidia chips

Specialized vs. General-purpose: TPUs are designed specifically for AI tasks; GPUs are more flexible but more expensive

Energy efficiency: TPUs sacrifice versatility for higher efficiency, delivering greater computing power per unit of energy

Supply scale: Google will produce about 3 million TPUs next year, about half of Nvidia’s output

Google’s TPU is an Application-Specific Integrated Circuit (ASIC)—a “specialist,” designed for specific computational tasks. It sacrifices some flexibility and versatility for higher efficiency. Nvidia’s GPUs, by contrast, are “generalists”—more flexible, programmable, but at the cost of higher expenses.

Buffett’s Bet on the Vertical Integration Model

Google’s AI chips have become one of the few alternatives to Nvidia’s, directly dragging down Nvidia’s share price. Nvidia posted to reassure the market about the panic caused by TPUs, saying it was “happy for Google’s success,” but emphasized that Nvidia is already a generation ahead and its hardware is more general-purpose than TPUs or other task-specific chips.

Google is enjoying a sweet spot: its stock price is rising against the tide of the AI bubble. Buffett’s company bought its shares in Q3, Google AI version 3 received positive feedback, and TPU chips have investors excited. Over the past month, AI concept stocks like Nvidia and Microsoft have dropped over 10%, while Google’s share price has risen about 16%. At present, with a market cap of $3.86 trillion, it ranks third worldwide, behind only Nvidia and Apple.

Analysts call Google’s AI model vertically integrated. As a rare “full-stack” player in tech, Google controls the entire chain: it deploys in-house TPUs on Google Cloud, trains its own large AI models, and these models are seamlessly embedded in core businesses like Search and YouTube. The advantage of this model is obvious: not relying on Nvidia, having efficient, low-cost compute sovereignty.

This is exactly what Buffett values. He has always favored companies with “moats,” and Google’s vertical integration is the widest moat of all. While other companies still rely on Nvidia for compute power and on OpenAI for models, Google already controls the entire chain from chips to models to applications. This autonomy gives Google significant advantages in cost control, technical iteration, and strategic flexibility.

The other model is a looser alliance: tech giants each play their part—Nvidia supplies GPUs, OpenAI, Anthropic, etc. develop AI models, and cloud giants like Microsoft buy GPUs from chipmakers to host these labs’ models. In this network, there are no permanent allies or opponents: they collaborate for mutual benefit when possible, but don’t hold back when competing.

Players have formed a “circular structure,” with capital recirculating among a handful of tech giants. For example, OpenAI spends $300 billion buying compute from Oracle, Oracle spends billions on Nvidia chips to build data centers, and Nvidia in turn invests up to $100 billion in OpenAI—on the condition that it continues to use Nvidia’s chips. Morgan Stanley analysts warn that with little transparency, investors find it hard to assess real risks and returns.

Google’s Cash Flow Advantage and User Base

OpenAI is currently valued at $500 billion, making it the highest-valued startup in the world. It is also one of the fastest-growing companies ever, with revenue soaring from nearly zero in 2022 to an estimated $13 billion this year. But it also expects to burn more than $100 billion over the next few years to achieve artificial general intelligence, plus spend tens of billions more leasing servers. In other words, it still needs to keep raising funds.

Google has an undeniable advantage: a much deeper pocket. Its latest quarterly report showed revenue topping $100 billion for the first time, reaching $102.3 billion, up 16% year-over-year, with profits of $35 billion, up 33%. Free cash flow stands at $73 billion, and capital expenditure around AI will reach $90 billion this year. For now, it doesn’t have to worry about AI eating into its search business—search and ads are still showing double-digit growth. Its cloud business is booming—even OpenAI rents its servers.

Beyond its self-sustaining cash flow, Google holds chips OpenAI can’t match, such as vast ready-made datasets for model training and optimization, and its own computing infrastructure. On November 14, Google announced a $40 billion investment in new data centers. With about a 90% share of the global search market, Google naturally controls the key channel for promoting its AI models, giving it direct access to massive user numbers.

Although ChatGPT still has a significant edge over Google AI version 3 in terms of user numbers, the gap is narrowing. In February, ChatGPT had 400 million weekly active users, which climbed to 800 million this month. Google AI version 3 reports monthly active user data: from 450 million in July to 650 million this month.

Google CEO Sundar Pichai said in a recent podcast that Google employees should get some rest. “From the outside, it may have seemed like we were dormant or falling behind, but in fact, we were laying the groundwork for everything, and now we’re pushing ahead at full speed.” The situation has now reversed. Pichai said, “We have reached a turning point.”