Billionaire investor Stanley Druckenmiller recently pivoted his AI investment strategy, selling his stake in chipmaker Broadcom to invest in AI giants Amazon and Alphabet instead. This move reflects a recognition of hyperscale cloud providers increasingly developing their own custom AI chips and taking greater control over their data center infrastructure. Despite heavy capital spending impacting short-term free cash flow, Amazon and Alphabet offer compelling long-term value given their massive revenue backlogs and strategic position in the evolving AI landscape.
Legendary investor Stanley Druckenmiller, known for his uncanny ability to spot major market trends, recently made headlines with a significant shift in his artificial intelligence (AI) investment portfolio. Once an early proponent of chipmakers like Nvidia (NVDA), from which his Duquesne Family Office reaped hundreds of millions before he admitted selling too soon in 2024, Druckenmiller is now reallocating his bets.
In the last quarter, Druckenmiller fully divested from another prominent AI chip stock, Broadcom (AVGO), completely exiting his position. Concurrently, he initiated substantial new positions in two other AI powerhouses further up the computing supply chain: Amazon (AMZN) and Alphabet (GOOG, GOOGL).
This strategic pivot suggests a deeper understanding of the evolving AI landscape. While traditional valuation metrics might not always be Druckenmiller's primary concern, the shift reflects an acknowledgment of a critical industry trend: hyperscalers are increasingly taking command of their AI infrastructure.
Druckenmiller often identifies investment opportunities born from profound global economic and market shifts. The explosion of AI chipmakers exemplifies such a trend. Despite Broadcom's valuation hovering around a P/E of 40 at one point last quarter, it likely wasn't the sole factor in his decision to sell, especially given he bought it at a more favorable forward P/E below 30 previously.
Instead, his focus appears to have shifted to the growing dominance of hyperscale cloud providers. These tech giants are aggressively investing in new data centers and internal chip development to meet surging AI compute demand. They are raising substantial capital through debt and equity to fund these massive expenditures.
While this might seem to initially benefit companies like Broadcom, which produces networking chips and custom AI accelerators (including Alphabet's Tensor Processing Units or TPUs), the reality is that major customers like Amazon and Alphabet are asserting greater control over their data center hardware choices.
Amazon CEO Andy Jassy highlighted this, stating that the vast majority of new chips entering their data centers this year will be their proprietary Trainium chips, not off-the-shelf solutions from vendors like Nvidia. Similarly, while Alphabet has historically collaborated with Broadcom for its TPUs, it recently forged a deal with Marvell for specialized inference TPUs.
The increasing preference among AI customers for hyperscalers' custom silicon, combined with the hyperscalers' willingness to diversify their chip suppliers, points to a long-term trend. The power to dictate which chips are made and by whom is consolidating with the large hyperscale providers, making them strategic long-term bets.
For investors, this shift offers a compelling opportunity. Despite Druckenmiller's less valuation-centric approach, Amazon and Alphabet currently trade at historically attractive forward P/E ratios of 20.5 and 16.4, respectively. This relative undervaluation is largely due to concerns over their substantial capital expenditures.
Both companies have reported negative free cash flow recently – Amazon burned $7.6 billion over the last four quarters, and Alphabet registered negative $5.9 billion in its most recent quarter. These capital outlays are set to intensify before yielding returns.
However, both possess immense backlogs of contracted revenue, with Alphabet reporting $514 billion and Amazon $496 billion in remaining performance obligations. This provides clear visibility into future demand and expected returns on their significant investments.
As these hyperscalers continue to build out their infrastructure and exert greater control over their AI supply chains, they are strongly positioned to deliver robust operating results and impressive shareholder returns in the years to come, solidifying Druckenmiller's strategic move.
