Key Takeaways
- Duan Yongping (段永平)’s disclosed U.S. portfolio, managed under H&H, reached $174.89 billion (¥1.208 trillion) in Q4 2025, with Apple remaining the dominant 50.3% holding.
- The portfolio reveals a strategic pivot into AI through new, small positions in three companies: AI cloud provider CoreWeave, data connectivity firm Credo Technology, and medical AI pioneer Tempus AI.
- Major moves included adding to Berkshire Hathaway, Nvidia, Pinduoduo (拼多多), Google, and Microsoft, while trimming Apple, Occidental Petroleum, Alibaba (阿里巴巴), Disney, and ASML.
- Duan’s public commentary frames AI as a transformative force on par with the internet and industrial revolution, but warns of a speculative bubble where “90%” of AI-related companies may fail.
- This evolution from consumer staples and tech giants to AI enablers offers a masterclass in adapting a value-investing framework to a disruptive technological paradigm.
A Legendary Investor’s Portfolio Evolution
The recent release of 13F filings for the fourth quarter of 2025 has offered a rare and invaluable glimpse into the strategic moves of one of China’s most revered investment minds. The portfolio managed by Duan Yongping (段永平) under the entity H&H in the United States reveals not just a colossal $174.89 billion (approximately ¥1.208 trillion) in assets, but a significant thematic shift that global investors cannot ignore. While Apple maintains its throne as the undisputed core holding, accounting for over half the portfolio’s value, the most compelling narrative lies in the new additions and the rationale behind them. This analysis delves into Duan Yongping’s latest strategic pivot into AI, unpacking the risks, opportunities, and profound market implications of his bets on the infrastructure and application layers of artificial intelligence.
For years, Duan Yongping (段永平) has been celebrated for his patient, value-oriented approach, famously applied to consumer icons like NetEase (网易), Kweichow Moutai (贵州茅台), and Apple. His public musings, often shared on social media, are studied for their clarity and long-term perspective. The Q4 2025 filings, therefore, are more than a snapshot of holdings; they are a ledger of conviction, marking a deliberate and calculated entry into the most disruptive theme of our generation. His move is a powerful signal to the market that artificial intelligence has matured from speculative hype to a investable megatrend with identifiable, cash-flowing infrastructure players.
Decoding the $120 Billion Portfolio: Core Stability and New Frontiers
The foundation of Duan’s H&H portfolio remains remarkably concentrated and consistent with his long-held beliefs. Apple’s position, valued at $87.97 billion, underscores a continued belief in the company’s ecosystem and cash-generation prowess. Berkshire Hathaway Class B shares, the second-largest holding at $36.07 billion, represent a pure play on the value investing philosophy of Warren Buffett and Charlie Munger, whom Duan has frequently cited as major influences.
Notable Adjustments in Q4 2025
The quarterly activity sheet reveals a investor carefully rebalancing and redeploying capital. Additions to Berkshire, Nvidia, Pinduoduo (拼多多), Google (Alphabet Class C), and Microsoft suggest a reinforcement of bets on diversified conglomerates, AI hardware leadership, Chinese e-commerce resilience, and cloud/software dominance. Conversely, the trimming of Apple, Occidental Petroleum, Alibaba (阿里巴巴), Disney, and ASML may indicate profit-taking, sector rotation, or a nuanced view on near-term challenges facing these specific names. It is crucial to interpret these trims not as a loss of faith, but as part of a dynamic portfolio management process where capital is recycled into higher-conviction ideas.
The complete list of 14 holdings, in descending order of portfolio weight, is: Apple, Berkshire Hathaway B, Nvidia, Pinduoduo (拼多多), Google C, Occidental Petroleum, Microsoft, Alibaba (阿里巴巴), Taiwan Semiconductor Manufacturing Company (TSMC, 台积电), Disney, CoreWeave, Credo Technology, ASML, and Tempus AI. The inclusion of the final three names, however small their initial allocations, is the story of the quarter.
The AI Trio: A Cautious but Calculated Infiltration
Duan Yongping (段永平)’s strategic pivot into AI is embodied by the three new positions, each representing a distinct layer of the AI value chain. With allocations of 0.12%, 0.12%, and 0.04% for CoreWeave, Credo Technology, and Tempus AI respectively, these are clearly exploratory “toe-in-the-water” investments. Yet, their selection is highly deliberate, focusing on companies that enable AI rather than those purely creating models, aligning with a classic “picks and shovels” investment strategy during a technological gold rush.
CoreWeave: Betting on the AI Cloud Infrastructure
CoreWeave operates as a specialized, high-performance cloud provider built specifically for GPU-accelerated workloads, primarily serving AI companies and large enterprises training massive models. Its business model is straightforward: as the demand for AI compute (算力) surges, CoreWeave rents out its clusters of advanced Nvidia GPUs. The investment thesis here is powerful—AI’s growth directly fuels demand for its services, potentially granting it significant pricing power during supply-constrained periods. However, Duan and his team are undoubtedly aware of the substantial risks. This is a capital-intensive business with enormous upfront costs for hardware, deep dependency on Nvidia’s supply chain, and fierce competition from hyperscale clouds like AWS, Microsoft Azure, and Google Cloud. The bet is that CoreWeave’s specialization and agility can carve out a profitable niche in an otherwise crowded field.
Credo Technology: Wiring the AI Data Centers
If CoreWeave provides the AI “engine,” then Credo Technology builds the high-speed “highways” inside the data center. AI servers packed with GPUs require incredibly fast and efficient data connectivity between chips to function effectively. Credo designs high-speed connectivity chips and optical modules that enable this critical data movement. This investment is a play on the essential upgrade cycle within data centers (数据中心) driven by AI server adoption. Credo doesn’t profit directly from AI software but from the hardware modernization its adoption necessitates. Its success hinges on deep technical moats and securing design wins with major server OEMs. The risk is cyclicality; a slowdown in AI infrastructure spending could lead to pronounced earnings volatility.
Tempus AI: Applying AI to Precision Medicine
This position represents a venture further up the risk-reward curve, targeting AI application in a complex, regulated field. Tempus AI leverages artificial intelligence to analyze clinical and molecular data to assist in personalizing cancer treatments. The long-term potential is staggering—improving oncology outcomes through data-driven insights. However, the path to scalable profitability in healthcare is long, fraught with regulatory hurdles, lengthy sales cycles, and the need for robust clinical validation. This investment reflects a belief in the transformative potential of AI beyond tech, acknowledging it as a multi-decade “marathon” rather than a sprint.
From NetEase to Nvidia: The Evolution of an Investment Philosophy
Duan Yongping (段永平)’s investment journey provides essential context for this strategic pivot into AI. His legendary early bet on NetEase (网易) in the early 2000s was a deep-value play on a misunderstood internet company. His advocacy for茅台 (Moutai) and Apple reflected a shift towards high-quality companies with durable competitive advantages, strong brands, and predictable cash flows—cornerstones of classic value investing. His recent moves indicate a further evolution, applying the same fundamental principles of seeking wide moats and essential services to a new, technology-driven era.
Publicly, Duan has been increasingly vocal about AI. In social media posts, he has stated the need to “seriously learn how to use AI” and expressed a hope to one day understand it to the “敢下重注” (dare to make a heavy bet) level. He acknowledges AI’s power to dramatically improve efficiency and sees it changing the world at an accelerating pace. This learning-focused, first-principles approach is classic Duan: seeking to understand a technology deeply before committing significant capital. His current small positions in AI are the logical first step in that rigorous process.
Navigating the AI Frenzy: Bubble Warnings and Long-Term Conviction
Perhaps the most critical insight from Duan Yongping (段永平)’s commentary is his nuanced, bifurcated view of the AI landscape. He embodies both the exuberant believer and the cautious skeptic, a balance vital for navigating this market. On one hand, he unequivocally states: “AI is changing the world is an indisputable fact.” He has drawn comparisons to the transformative impact of the internet and the Industrial Revolution, suggesting the changes may be even larger. This underpins his strategic pivot into AI as a necessary adaptation for any long-term investor.
A Sobering Reality Check on Speculation
Concurrently, Duan offers a stark warning that serves as a crucial risk framework for investors following his lead. He has observed that “anything with a little connection to AI is rising chaotically” and estimates that “90% of these companies will eventually fail.” This echoes historical tech bubbles, where proliferation is followed by consolidation. His strategy of investing in enabling infrastructure (CoreWeave, Credo) rather than chasing every AI startup aligns with this cautious outlook. He is seeking the potential “next Google or Amazon” that survives the coming shakeout, not gambling on the entire sector.
His specific views on key players further refine this framework. He has expressed confidence in Google’s ability to survive the “large model wars,” citing its resources and technical prowess. On Nvidia, he acknowledges its sustained dominance and lack of near-term substitutes, while prudently noting the competitive threats from custom silicon like TPUs and the efforts of other tech giants. This balanced analysis suggests his additions to Nvidia and Google are considered bets on likely long-term winners, not mere momentum chasing.
Synthesis and Forward Guidance for Global Investors
Duan Yongping (段永平)’s Q4 2025 portfolio maneuvers offer a masterclass in adaptive, thesis-driven investing. The core lesson is not to blindly copy his holdings, but to understand the strategic reasoning behind his strategic pivot into AI. He is not abandoning value investing; he is redefining what constitutes a “value” and a “moat” in the age of artificial intelligence. The picks and shovels of this revolution—compute power, data connectivity, and specialized applications in fields like healthcare—are being evaluated through the same lens of business quality, competitive advantage, and essential utility that he applied to consumer goods and mature tech giants.
For institutional investors and sophisticated market participants worldwide, this portfolio shift is a significant data point. It validates AI as a fundamental, not faddish, driver of capital allocation. However, it also advocates for a highly selective, infrastructure-focused approach amid widespread speculation. The call to action is clear: develop a nuanced, layered understanding of the AI ecosystem. Look beyond the hype of model developers to the companies building the indispensable underlying platforms. Conduct deep due diligence on business models, capital intensity, and competitive threats. Most importantly, embrace Duan’s own mindset of continuous learning. As he advised, start by learning to use the tools themselves, cultivate a long-term perspective that looks past short-term volatility, and always balance transformative optimism with a sober assessment of economic realities and valuation. In doing so, you align your strategy with one of the most successful investment evolutions of our time.