Key Takeaways
- Stanley Druckenmiller bought Nvidia in late 2022, made hundreds of millions, but sold the position too early in 2024.
- He exited his Broadcom stake last quarter and redirected capital to Amazon and Alphabet, two hyperscalers investing heavily in AI‑focused data‑center infrastructure.
- Druckenmiller’s move reflects a broader trend: major cloud providers are asserting greater control over chip selection, favoring custom silicon (e.g., Amazon’s Trainium, Alphabet’s TPUs) over off‑the‑shelf solutions.
- Amazon and Alphabet currently trade at historically low forward P/E ratios (≈20.5 and 16.4), despite negative free cash flow from massive capex, suggesting a potential valuation‑driven entry point for long‑term investors.
- The article notes that “the market is offering investors the chance to buy Amazon and Alphabet at historically low P/E ratios,” highlighting a possible upside if the hyperscalers’ capital investments translate into strong returns on invested capital.
Early AI Chip Bet: Nvidia Purchase in 2022
Stanley Druckenmiller first recognized the upside potential of AI chipmakers when he added Nvidia (NASDAQ: NVDA) to his Duquesne Family Office portfolio in late 2022. “He bought Nvidia … and ultimately made hundreds of millions of dollars from that purchase,” the article notes, underscoring the timing of his entry just before the AI boom accelerated. The move exemplified Druckenmiller’s habit of positioning ahead of major secular shifts, even if valuation was not his primary filter.
Selling Too Soon: The Nvidia Exit in 2024
Despite the sizable gains, Druckenmiller admitted he sold the Nvidia stake too early, fully disposing of the position in 2024. He told reporters, “I sold the stock too soon,” reflecting a common investor dilemma: capturing profit while potentially leaving further upside on the table. The premature exit highlights the challenge of timing even when a thesis proves correct.
Broadcom Exit and the Shift to Hyperscalers
In the most recent quarter, Druckenmiller also sold his entire holding in Broadcom (NASDAQ: AVGO), another AI‑related chipmaker. The decision came as he observed a strategic shift among the largest cloud providers. “He purchased relatively large stakes in two other AI giants further down the AI compute supply chain: Amazon (NASDAQ: AMZN) and Alphabet (NASDAQ: GOOG) (NASDAQ: GOOGL),” the piece explains, indicating a reallocation from chip suppliers to the firms that design and deploy their own silicon.
Why Valuation Took a Backseat
Druckenmiller’s investment philosophy places less emphasis on traditional valuation metrics when a macro trend is strong. The article states, “valuation is a lesser concern for Druckenmiller,” noting that Broadcom’s rise to a P/E of 40 did not heavily influence his sell decision. Conversely, he had initially bought Broadcom when its forward P/E sat below 30, viewing it as a good value at that time.
Hyperscalers’ Push for Custom Silicon
The shift toward Amazon and Alphabet reflects a broader industry development: hyperscalers are asserting greater control over which chips populate their data centers. Amazon CEO Andy Jassy said, “the largest number of the new chips going into its data centers this year will be its own custom Trainium chips, not Nvidia GPUs or any other off‑the‑shelf solution.” Alphabet similarly reported strong demand for its TPUs and is even selling TPU systems to external customers, while also diversifying suppliers with a new deal with Marvell for inference‑focused TPUs.
Multiple Sourcing and the Decline of Sole‑Source Chip Dependence
As cloud giants design their own accelerators, they are also showing a willingness to use multiple sources for those chips. The article notes, “the hyperscalers are showing greater willingness to use multiple sources for those chips.” This strategy reduces reliance on any single supplier—such as Broadcom or Nvidia—and strengthens the hyperscalers’ bargaining power, aligning with Druckenmiller’s view that long‑term value resides with those who control the chip supply chain.
Market Opportunity: Low P/E Ratios Amid Heavy Capex
Despite the aggressive capital spending that has pressured earnings, Amazon and Alphabet currently trade at depressed multiples. Amazon’s forward P/E stands at roughly 20.5, while Alphabet’s is about 16.4. The article explains, “Those earnings multiples are depressed due to fears related to both companies’ capital spending.” Amazon’s free cash flow turned negative over the past twelve months, burning $7.6 billion, and Alphabet posted its first‑ever negative free cash flow of ‑$5.9 billion. These figures reflect the massive outlays for new data‑center infrastructure needed to sustain AI workloads.
Backlog and Future Revenue Visibility
The negative cash flow is offset by substantial contracted revenue backlogs, providing visibility into future earnings. “Alphabet ended the quarter with $514 billion in remaining performance obligations. Amazon’s backlog ballooned to $496 billion,” the report highlights. Such backlogs suggest that, once the new data centers come online, both firms can expect a steady stream of high‑margin revenue, bolstering utilization rates for their custom chips and driving strong operating results over the long haul.
Inference Workloads and the Hyperscaler Advantage
While third‑party providers have stepped in to fill temporary capacity gaps, the article anticipates that the majority of inference workloads will eventually run on hyperscaler‑owned servers, where data and applications already reside. “As a result, Amazon and Alphabet should be able to maintain very high utilization rates for their chips, producing strong operating results in the long run.” This dynamic reinforces the strategic advantage of owning both the hardware and the software stack that powers AI services.
Conclusion: Aligning with Druckenmiller’s Lead?
The piece closes by posing the question of whether individual investors should follow Druckenmiller’s reallocation from Broadcom to Amazon and Alphabet. It notes that while the billionaire may not be overly concerned with valuation, buying at historically low P/E multiples offers a margin of safety. If the hyperscalers’ capital investments translate into the anticipated returns on invested capital, the stocks could deliver outsized gains—mirroring the early‑stage success Druckenmiller enjoyed with Nvidia, albeit with a different risk‑return profile.
Quoted excerpts are taken verbatim from the original source to preserve the author’s voice and ensure journalistic fidelity.
https://currently.att.yahoo.com/att/billionaire-stanley-druckenmiller-sold-broadcom-202000494.html