Google is emerging as the focal point of contention among top investors in AI assets.

On one hand, Buffett is stepping in personally, increasing his purchases as the price dips.

Berkshire Hathaway has significantly boosted its stake in Google, investing tens of billions of dollars over two consecutive quarters. By the end of June, Google had climbed to become Berkshire's third-largest holding, trailing only Apple and American Express, and even surpassing Coca-Cola, a position it has held for over three decades.

On the other hand, diametrically opposite actions are being taken.

A cohort of influential investors, including Duan Yongping and Bill Ackman, are substantially reducing their stakes or exiting entirely. In the second quarter, H&H International Investment, under Duan Yongping, slashed its Class C Google shares by 46.88%.

More intriguingly, in August, he remarked, "It seems there's a cultural issue at Google; they struggle to retain top talent."

Bill Ackman was even more resolute. At the end of last year, Pershing Square's Google holdings were valued at approximately $2.1 billion. Six months later, not a single share remained—a complete sell-off.

This raises an intriguing question. All these investors are looking at the same Google, so why are they reaching such vastly different conclusions?

Today, Silicon-Based Gentleman will delve into this disagreement among top investors regarding Google.

/ 01 /

Buffett Increases Stakes, Duan Yongping Reduces

Investing in Google marks one of the most significant decisions Buffett has made in recent years, if not the most crucial.

By the end of June, Berkshire Hathaway's combined holdings of Google A and Google C shares had reached a market value of $37.764 billion, constituting 12.62% of its entire U.S. stock portfolio. Google had become Berkshire's third-largest holding, trailing only Apple and American Express, and even surpassing Coca-Cola, which it has held for over three decades.

What's even more noteworthy is the pace at which these stakes have increased.

By the end of 2025, Berkshire Hathaway held approximately 17.8 million Google shares, valued at about $5.6 billion. By the end of the first quarter of 2026, the holdings had surged to nearly 58 million shares, with a market value close to $17 billion.

In the second quarter, Berkshire Hathaway continued its aggressive buying spree.

On June 1, Google issued $10 billion worth of shares exclusively to Berkshire Hathaway, including $5 billion in Class A shares at $351.81 each and $5 billion in Class C shares at $348.20 each—both prices were more than 6% below the day's closing price.

In addition to participating in the private placement, Berkshire Hathaway also continued buying on the secondary market.

In the second quarter, Berkshire Hathaway's Google A holdings increased by a net 24.54 million shares, and its Class C holdings increased by approximately 23.6 million shares. After accounting for the approximately 14.2 million A shares and 14.4 million C shares allocated in the private placement, an additional approximately 10 million A shares and 9 million C shares were purchased on the secondary market, valued at approximately $7 billion at the period-end prices.

In essence, Berkshire Hathaway has spent over $10 billion to increase its Google stake for two consecutive quarters. This pace is uncommon in Berkshire's history. The last time this occurred was with Apple. More importantly, this Google investment was initiated by Mr. Buffett himself.

Another major investor in Google is Li Lu.

By the end of the second quarter this year, Himalaya Capital held a total of 2.5433 million Google A shares and 2.4513 million Google C shares, with a combined market value of approximately $1.775 billion, accounting for 47.94% of its entire U.S. stock 13F portfolio.

In other words, nearly half of Li Lu's publicly disclosed U.S. stock assets are bet on Google.

However, unlike Berkshire Hathaway's rapid stake increases in recent quarters, Li Lu's bet on Google came much earlier. He established a position in Google C as early as the second quarter of 2020, when the stock price was around $60; he significantly increased his stake twice during the market downturn in 2022.

Unlike Buffett and Li Lu's firm optimism, some top investors have begun to "abandon" Google.

Duan Yongping also significantly reduced his Google stake in the second quarter.

According to the 13F filing by H&H International Investment, by the end of the first quarter, it held 3.706 million Google Class C shares (GOOG), with an end-of-period market value of approximately $1.063 billion, accounting for 5.31% of its entire U.S. stock holdings. At that time, Google was already H&H's sixth-largest holding.

By the end of the second quarter, this figure had dropped to 1.9686 million shares.

In other words, within one quarter, H&H sold approximately 1.7374 million Google Class C shares, reducing its holdings by 46.88%. As of June 30, the market value of these holdings was approximately $696 million, and their proportion in the 13F portfolio dropped from 5.31% to 3.64%.

Although Duan Yongping's reduction was not entirely due to active selling, as he is very keen on managing his portfolio using put and call options, his attitude toward Google has still changed.

On August 9, in response to market rumors about Google's talent and management, Duan Yongping said, "It seems there's a cultural issue at Google; they struggle to retain top talent."

Bridgewater, the world's largest hedge fund, also reduced its Google stake.

Filings show that Bridgewater reduced its Google-A (GOOGL) holdings by approximately 675,500 shares in the second quarter, a 33.81% decrease, leaving it with approximately 1.3222 million shares, worth about $473 million, accounting for 1.94% of its 13F portfolio's market value.

Meanwhile, Bill Ackman also completely liquidated his Google holdings.

At the end of 2025, Pershing Square's Google holdings were worth approximately $2.1 billion.

By the first quarter of this year, Ackman had slashed his position by about 95%, leaving only 32,376 Google A shares and 311,726 Google C shares, with a combined value of less than $100 million.

By the second quarter, Ackman had sold off all remaining Google shares, completely exiting an investment he had held for over three years.

So, now we have an interesting situation surrounding Google:

On one side, Berkshire Hathaway and Li Lu are doubling down; on the other side, Duan Yongping, Bridgewater, and Ackman are reducing or liquidating their stakes to varying degrees.

Top investors are looking at the same company but reaching vastly different conclusions. So, why is this happening?

/ 02 /

Why Won't Google Lose?

Those bullish and bearish on Google are essentially debating one question:

In the AI era, is "asset depth" more important, or is "innovation speed"?

The bearish case for Google is straightforward: Google has fallen behind in the large model competition.

Because AI remains a rapidly evolving technological innovation race. Models iterate every few months, product forms are constantly being reshaped, and leading advantages can shift at any moment.

In such a competition, a large company's resource advantages do not necessarily translate into product advantages. On the contrary, a massive organization, complex interest relationships, and mature businesses can often slow down innovation.

The Transformer model was born at Google, and DeepMind has long been one of the world's top AI labs, but it was OpenAI that truly turned generative AI into a mass-market product.

Now, this concern has resurfaced.

After Gemini took an early lead at the end of 2025, it was surpassed by OpenAI and Anthropic in some model capabilities. Especially in the core area of coding, Google was completely absent.

Even more worrying is the departure of key technical leaders like Jeff Dean, Google's former chief scientist and Gemini co-lead, and Sanjay Ghemawat. To many, this signals a strategic retreat in cutting-edge large model development.

But those bullish on Google see a different battle.

They believe AI will ultimately not just be a model competition but a sustained, resource-intensive systemic war.

Models themselves will become increasingly commoditized—being first today and third tomorrow matters less. What will be truly scarce in the long run is who can simultaneously control chips, computing power, models, data, developers, enterprise clients, and user access points.

From this perspective, Google may be one of the most complete AI asset holders globally.

OpenAI has strong models but lacks its own chips, cloud, and operating system; Anthropic has strong models but relies more on partners for distribution and infrastructure; NVIDIA controls chips but lacks a consumer access point like Google Search.

Google holds nearly all the critical pieces, meaning it possesses a complete resource ecosystem for long-term participation in the AI war.

More important than the resources themselves is the unique approach Google has developed over the past two decades: both focused and exploratory. This is why some call Google the "Berkshire Hathaway of tech."

For over two decades, Google has remained centered on one core question: how to more efficiently organize, understand, distribute, and utilize information. Search, YouTube, Maps, Cloud, Gemini, TPU—these appear to be different businesses, but at their core, they all reinforce information processing capabilities.

From this angle, Google and NVIDIA even share some similarities. NVIDIA continuously bets on the long-term proposition of "accelerated computing," while Google continuously expands its capabilities around "information."

And Google can pursue this approach long-term because it has a sufficiently strong core business.

Search advertising generates massive cash flow, while YouTube and Cloud continuously form new growth pillars. These predictable businesses give Google the capital to bet on an uncertain future.

Waymo, quantum computing, DeepMind, smart glasses—any of these projects, taken individually, could require over a decade of investment before seeing financial returns.

These projects may seem inefficient on their own, but within Google's overall ecosystem, they function like a set of long-term call options that don't require additional payments.

Waymo is the best example. A project that began in 2009 and saw almost no financial returns for over a decade is now valued at $126 billion in its latest funding round.

This unique approach allows Google to excel at allocating today's earnings toward the most important technological directions a decade from now.

This is what makes the bull-bear debate on Google truly interesting.

If AI is a sprint, centered on model iterations, product innovation, and organizational execution efficiency every six months, then Google has indeed fallen behind. But if AI is ultimately a systemic war lasting a decade or even longer, then Google's advantages will only become more pronounced over time.

In tech, no company is "unbeatable." Google's true certainty lies in its ability to lose a battle but rarely exit the war due to a single defeat. This may be the real reason long-term investors remain bullish on Google.

By/ Qi