The Zhitong Finance App learned that the two top Wall Street investors invariably made the same move — Stanley Druckenmiller and Dan Loeb both cleared Broadcom (AVGO.US) and instead took heavy stock of Alphabet (GOOGL.US). This position adjustment reflects the shift in AI investment logic in their eyes: from a chip supplier that “sells shovels” to an AI platform company with a complete closed loop ecosystem.
Stanley Druckenmiller and Dan Loeb are two of Wall Street's most high-profile investors. Drucken Miller is a macro strategy expert. He was a portfolio manager under George Soros and now heads his own family office. Loeb is the founder of hedge fund Third Point, known for aggressive investing and event-driven operations, and is good at capturing catalysts and driving management change.
The reason why savvy investors keep an eye on Druckenmiller and Loeb is because the two have repeatedly rotated capital to the next round of growth before a market consensus is reached. Unsurprisingly, the latest 13F position report submitted by the two shows a common shift: they both cleared Broadcom, and at the same time opened large positions on Alphabet.
The following will analyze how this rotation reflects the layout ideas of the two top Wall Street minds for the next stage of the AI revolution.
Exiting Broadcom at a strategic point in the AI chip cycle
Broadcom is undoubtedly the core supplier of AI infrastructure — the company designs customized accelerators and provides high-performance network chips to connect to processor clusters within data center servers. As AI hyperscale companies compete to expand computing power, demand for these components continues to drive Broadcom's revenue growth.
However, Drucken Miller and Loeb's decision to both clear their inventory shows that they believe other targets in the AI ecosystem have better risk-reward ratios. Despite the compression in valuations, Broadcom's current stock price still contains overly optimistic expectations. Meanwhile, Nvidia and AMD's next-generation GPU architectures continue to add new variables to the competitive landscape. Furthermore, once the supply of computing power is gradually matched with demand, the capital expenditure cycle may rapidly shift — making Broadcom's long-term prospects uncertain.
Alphabet's investment logic
Alphabet's appeal lies in its complete vertical integration capabilities. The company designs its own chip — the Tensor Processing Unit (TPU) — and operates vast data centers and optical networks. The Gemini series of AI models was developed through DeepMind Labs, and these capabilities were deployed on a large scale to global product matrices such as Google Search, YouTube, Android, and Workspace.
The core logic is: AI is not an additional function of Alphabet, but is deeply integrated into the underlying architecture of its various core businesses. With its near-monopoly position in online search, Alphabet has embedded AI capabilities more deeply into all of its major products — search now presents AI overviews to billions of users, and YouTube uses generative tools to optimize content recommendations and ad delivery.
Google Cloud is in a unique ecosystem: it is both a sales infrastructure (TPU) and enterprise services, and customer usage data can be directly streamed back for model iteration. The compound benefit effect brought about by this closed-loop model is difficult for cutting-edge model developers or commercial chip vendors to replicate on a large scale.
Alphabet's AI strategy has handed over an impressive report card. In the second quarter, the company's total revenue reached US$119.8 billion, an increase of 24% over the previous year. The Google Cloud business expanded at an accelerated pace, with revenue soaring 82% to US$24.8 billion, and the backlog of cloud business orders reached US$514 billion. Operating profit increased 30% year over year to US$40.8 billion, and operating margin increased to 34%.