Smart Money Recalibrates: Hedge Funds Trim AI Mega-Caps in 2026

By Vijay Venkatesh B

When Smart Money Retreats: Hedge Funds Trim AI Heavyweights in 2026 

Major investment firms, including Tiger Global and Adage Capital, have quietly begun cutting their positions in AI-focused mega-cap stocks. This is not a stampede out the door but the direction of travel carries a message every investor needs to understand. 

Something is changing quietly at the top of institutional finance. The hedge funds that rode the artificial intelligence wave to extraordinary gains are now, carefully and deliberately, beginning to reduce their exposure to the very stocks that powered that ride. Regulatory filings submitted to the U.S. Securities and Exchange Commission for the final quarter of 2025 reveal a pattern that is hard to dismiss: Tiger Global Management, Adage Capital Partners, D.E. Shaw, Berkshire Hathaway, and SoftBank Group have all trimmed their stakes in AI-linked mega-cap technology companies. The adjustments are measured, not panicked. But the fact that multiple sophisticated, independently operating institutions are all moving in the same direction at the same time is worth paying close attention to. 

Introduction 

Global equity markets are sitting at a crossroads. Over the past three years, a handful of technology companies commonly referred to as the Magnificent 7 dominated market performance, index weightings, and institutional portfolios. Their rise was tied inseparably to the artificial intelligence narrative: the belief that these companies, by virtue of their computing infrastructure, data advantages, and capital firepower, would be the primary beneficiaries of what many called the most transformative technology of the century. That belief drove valuations to extraordinary heights. Now, some of the world's most closely watched investors are quietly asking whether those valuations have run ahead of the underlying reality. 

The adjustments being made are not small. Tiger Global cut its Microsoft stake by approximately 16% in a single quarter, reducing its holding from 6.5 million shares to 5.47 million. Amazon was trimmed by more than 9%. Nvidia was also reduced. SoftBank went further still it dissolved its entire Nvidia position in October 2025, a complete exit from the company that had become one of the most prominent symbols of the AI infrastructure boom. These are not rounding-error adjustments. They are deliberate repositioning decisions made by institutions with access to the deepest research and the most sophisticated analytical frameworks in the world. 

 How AI Stocks Became Vulnerable to Repricing 

To understand why this matters, it is important to understand why these stocks were valued so highly in the first place. The Magnificent 7 did not simply benefit from AI enthusiasm they were perceived as the companies most likely to capture the economic value that AI would generate. Investors were willing to pay premium multiples because the assumption was that enormous capital spending on data centres, model training, and infrastructure would eventually translate into dominant revenue streams. The logic was compelling enough that questions about the timeline were largely set aside. 

That has changed. Investors are no longer content to pay for a promised future without seeing clearer evidence of when and how that future arrives. Microsoft's Azure AI revenues are growing, but not at a pace that obviously justifies the magnitude of capital being deployed. Meta faced criticism from its own shareholders over AI spending levels. Google's integration of Gemini into its core products moved more slowly than many had expected. The investment thesis is not broken, but the confidence that returns would materialise quickly and at scale has been shaken. When that confidence erodes, valuations built on optimistic assumptions become vulnerable. Markets don't wait for profits to actually disappoint they begin adjusting the moment the growth narrative becomes uncertain. 

 The Signals Hidden Inside the Filings 

What makes the current round of institutional repositioning particularly telling is not just what these funds sold it is what they bought instead. Adage Capital trimmed Microsoft, Amazon, Alphabet, and Nvidia between 1% and 3% each, but simultaneously increased its Oracle stake by nearly 19%.It doesn't dominate financial media in the way that Nvidia or Microsoft does. But it has quietly become one of the most important infrastructure providers for enterprise AI workloads, and it does so without the speculative premium that has been embedded in the Magnificent 7 valuations. Adage's pivot to Oracle reads as a search for AI exposure at a more defensible price the same structural tailwind, but without paying for perfection. 

D.E. Shaw's moves tell a similar story. The quantitative hedge fund reduced Nvidia, Micron, and Meta while adding to Amazon and AMD. This is not a withdrawal from the AI trade it is a more discriminating version of it. The implicit argument embedded in these transactions is that not every company benefiting from AI deserves the same valuation premium, and that the market has been indiscriminate in assigning it. SoftBank's decision to fully exit Nvidia and deepen its commitment to Open AI points in the same direction: away from the infrastructure hardware layer and toward the application and model layer, where the next wave of value capture may be more direct and more legible. 

 What This Means for Equity Markets 

The implications of coordinated institutional trimming extend well beyond the individual stocks involved. The Magnificent 7 carry enormous weights in major global equity indices. When large institutional holders begin reducing positions even modestly the effect on market structure is disproportionate. Index-tracking funds feel it. Passive investors feel it. And the psychological signal it sends to the broader market has a compounding effect that goes beyond the actual share of capital being moved. 

For equity markets more broadly, the current episode illustrates something important about how AI-driven narratives are being stress-tested. The 2023 and 2024 rally in technology stocks was powered largely by expectations about AI adoption curves, revenue uplift, and competitive moats. Those expectations remain directionally valid, but the timeline and magnitude are being revised. Multiple compression, where price-to-earnings ratios contract without a corresponding decline in earnings, is the mechanism through which markets absorb this kind of narrative correction. It is often uncomfortable but is ultimately healthier than a continued detachment between price and fundamental reality. 

The Path Forward for Investors 

For individual investors observing these developments, the most important takeaway is not that AI is a failed investment theme. It is that the easy, undifferentiated phase of that trade where buying anything with AI exposure produced strong returns is almost certainly behind us. The next phase rewards discrimination. It rewards understanding which layer of the AI stack is actually monetising, which companies have the earnings evidence to justify their capital spending, and which valuations are still living on narrative momentum rather than financial substance. 

Companies that are actively adapting deploying AI to improve their own operational efficiency, pivoting their revenue models to reflect a world where software and analytical work is increasingly augmented by machines, and demonstrating credible paths to margin improvement are the ones most likely to rebuild institutional confidence. Those that resist change, or that continue to rely on headcount-driven growth models that AI is already beginning to disrupt, may find their valuations under sustained pressure for longer than the market currently expects. 

 Conclusion: Markets Are Already Pricing the Next Chapter 

The point worth taking away from all of this is straightforward. The world's most sophisticated investors have looked at their AI heavy portfolios and decided that the balance of risk and reward has shifted. They are not abandoning the sector. They are recalibrating their exposure with the discipline that made them successful in the first place. Markets do not wait for certainty they price probability. Right now, the probability being priced is that the AI-driven earnings uplift for mega-cap technology companies will take longer and look different than the early optimism assumed. The investors who understand that distinction and position accordingly are likely to navigate what comes next far better than those still anchored to the narratives of 2023.