With the release of the 13F document, the latest trends from the giants are here!
An investment account managed by Duan Yongping in the United States was previously dug up by netizens, and the institution was named "H&H".
The report submitted by H&H shows that as of the fourth quarter of 2025, it holds 14 U.S. stocks with a market value of US$17.489 billion, or approximately RMB 120.8 billion.
The latest position disclosure shows that Duan Yongping has begun to test the waters of AI-related stocks, and has entered three new emerging technology companies: CoreWeave, Credo Technology, and Tempus Ai, with positions accounting for 0.12%, 0.12%, and 0.04% respectively.
In the fourth quarter, Duan Yongping increased his holdings in Berkshire, Nvidia, Pinduoduo, Google C, and Microsoft; he reduced his holdings in Apple, Occidental Petroleum, Alibaba, Disney, and ASML.
Apple has always been Duan Yongping's top holding. As of the end of the fourth quarter, H&H, which he manages, holds Apple with a market value of US$8.797 billion, accounting for 50.3% of its holdings; Berkshire B, the second largest holding, has a market value of US$3.607 billion, accounting for 20.63% of its holdings.
As of the end of the fourth quarter of 2025, Duan Yongping managed 14 H&H holdings, including Apple, Berkshire B, Nvidia, Pinduoduo, Google C, Occidental Petroleum, Microsoft, Alibaba, TSMC, Disney, CoreWeave, Credo Technology, ASML, and Tempus Ai.

Judging from the three new companies, they are all related to AI.
CoreWeave is a company that sells “AI computing power”. While all companies are training large models and doing generative AI, CoreWeave specializes in building high-performance GPU clusters and rents computing power to AI companies and enterprise customers. It does not make models, but provides computing power to the models. The more popular AI is and the tighter its computing power, the stronger its bargaining power. But the problem is also very direct – this is an asset-heavy business with expensive equipment and a heavy reliance on upstream chips. While it expands rapidly, it also faces financial pressure.
Credo Technology is a company involved in data centers. If computing power is like an engine, then data transmission is like a highway. In an AI server, high-speed communication is required between a large number of GPUs, otherwise it will not run fast no matter how many computing power piles there are. What Credo does is high-speed interconnect chips and optical modules. It makes money not from AI itself, but from data center upgrades. It is highly bound to the iteration of AI servers. However, its core competitiveness lies in technical barriers and close ties with large customers. Once the pace of the industry slows down, performance fluctuations will also be obvious.
Tempus AI uses AI in precision medicine, especially in the field of oncology, to help doctors develop personalized treatment plans by integrating genetic and clinical data. It sounds grand, but the medical industry has a long commercialization cycle, strict regulations, and profitability is still in the verification stage. It is more like an endurance race.
From NetEase, Moutai to Apple, Duan Yongping’s investments have been evolving over the years.
In recent years, he has set his sights on the AI industry and mentioned artificial intelligence many times.
Duan Yongping also mentioned in particular: "In the new year, we need to seriously learn how to use AI, and hope that one day we will understand how to 'dare to make big bets.' The more AI is used, the more interesting it becomes, and I feel that the world is changing. He also lamented that it is difficult to imagine what the world will look like in 10 years, but the changes are likely to be greater than in the past 10, 20, or even 40 years."
Duan Yongping said: "AI is really powerful. It has greatly improved efficiency. You see, it has changed a lot of things in two years. I don't know whether it is a blessing or a curse. Now you can ask AI anything, and it can help in many places. Of course, you can check more, because AI sometimes talks nonsense and it will talk nonsense."
Regarding the AI bubble, he replied on social platforms: "The wheel of AI history is rolling forward and will crush all AI bubbles."
He also commented: "Nowadays, things that have anything to do with AI are skyrocketing. It is likely that 90% of these companies will be finished in the end. However, those that are not finished are likely to be the next Google, Amazon… It is an indisputable fact that AI is changing the world, right?! That's a fact!"
Duan Yongping's understanding of AI is simple and direct: "AI is actually nothing special. AI is what we originally called computer applications. AI is the product of greatly improved computing power, and quantitative changes lead to qualitative changes! I don't know what kind of changes AI will bring in the end, but there is a high probability that it will be greater than the changes brought by the Internet, and it is also likely to be greater than the changes brought by the industrial revolution. I think every rational person should not ignore AI. To take AI seriously, at least start by learning to use ingots."
Regarding AI-related companies, Duan Yongping mentioned:
1. I said before that in a few years, few people who make large models will be able to survive, but there is a high probability that Google will.
2. NVDA's supply shortage should continue for some time, and it seems that there will be no real replacement in a few years. But under the impact of all the giants, it is hard to say whether it will be broken through. Of course, the implicit substitution of TPU has already begun, but it is not a big threat. The changes that AI can bring may really be huge, and it may only have just begun.



