Smart Money's Quiet AI Bet: Infrastructure and Software, Not Hype

While the market buzzes about chip stocks, the real smart money is quietly building positions in the AI infrastructure and software that makes the boom possible. The evidence from recent filings shows a clear playbook: bypass the most hyped names and accumulate in the essential, often overlooked, backbone of the AI revolution.

Major quantitative hedge funds have been quietly loading up on AI infrastructure. According to analysis of Q3 2025 13F filings, these firms increased their holdings in AI-related stocks by an estimated ~30–35%. The notable tilt? A significant bet on power and utilities. Data centers are the fuel for AI, and institutions are positioning for the massive energy demand they will create. As one report notes, the U.S. grid's growth has been flat for decades, but AI-driven demand is set to surge, creating potential bottlenecks that could reward early infrastructure bets.

This focus on the foundation extends to billionaire money. Ken Griffin's Citadel Advisors made a notable purchase of 388,000 shares of Palantir in Q3. While Palantir is a software play, its valuation is a stark warning. The stock has surged 2,200% since January 2023, trading at a premium of 96 times sales. This is skin-in-the-game from a proven winner, but it also highlights the risk of paying for hype. The smart money is not just buying growth; it's buying it at a price that still offers a margin of safety relative to the peak.

Then there's the bet on niche, critical software. Billionaire Israel Englander's Millennium Management purchased over 250,000 shares of Trimble. The company's geospatial software is now essential for autonomous vehicles and robotics, a key AI application. This move is a classic example of smart money finding value in a transition story-where a legacy hardware firm has pivoted to become a software enabler for next-gen tech. It's a quieter, less flashy accumulation than chasing the latest chip, but it represents a direct bet on AI's real-world deployment.

The bottom line is that institutional and billionaire money is accumulating where the real work gets done. They are building positions in the power grid, the data-center real estate, and the specialized software that runs the machines. This is the quiet accumulation of the smart money, a setup that often plays out after the initial hype has faded and the fundamentals of the infrastructure become undeniable.

Skin in the Game: Leadership Alignment and Valuation

The smart money's quiet accumulation is one thing. The real signal is whether the people running these companies are putting their own money where their mouth is. Recent insider transactions at two key AI infrastructure names suggest a surprising alignment of interest, even as the broader market remains skeptical.

At Salesforce, the move is both strategic and personal. Director David Blair Kirk, a former Nvidia architect, bought over 1,900 shares in mid-December, spending more than $500,000. This isn't a token purchase; it's a significant bet from someone who understands the AI stack. He's not alone. Activist firm ValueAct also acquired another $25 million worth of Salesforce stock in December, a powerful vote of confidence from a major shareholder. On the flip side, at SentinelOne, Director Mark Peek-a former CFO at Amazon and VMware-bought nearly $600,000 worth of shares in mid-December. These are high-caliber insiders with deep operational experience, stepping in to buy at a time when both stocks remain "beaten down."

This insider skin in the game is critical. When executives and directors are buying, it often signals they see a disconnect between the current price and the company's intrinsic value or future potential. It's a bet that the market is overlooking something-whether it's Salesforce's AI data platform or SentinelOne's niche security solutions.

The valuations support this contrarian view. Salesforce trades at a forward price-to-sales ratio of 4.7 times, a steep discount to its SaaS peers. SentinelOne is even cheaper, with a forward P/S below 4.5 times. For a company growing its revenue and making strategic AI moves, that's a compelling margin of safety. The smart money isn't paying for hype; it's buying the foundation at a price that suggests the market has written off the future.

The bottom line is a setup of alignment and value. Insiders are stepping up with real capital, and the numbers show these are not overvalued plays. This is the classic smart money signal: when the best minds are buying, and the price still looks low.

The Trap and the Opportunity: Catalysts and Risks

The smart money's quiet accumulation presents a clear setup: buy the foundation, not the flash. But for every bullish signal, there's a red flag. The biggest warning is the lack of widespread insider buying from AI executives. While a few high-caliber directors like those at Salesforce and SentinelOne are stepping up, the broader pattern suggests many insiders are not betting their own money. This is a classic trap for retail investors. When the people running the show aren't putting skin in the game, it often means they see a disconnect between the hype and the real value, or they are simply cashing out. The smart money is betting on the enabling layers-power, data centers, and niche software like Trimble's geospatial tools-because they know the end-user AI devices are a crowded, speculative race. This focus on infrastructure is a bet on the long-term plumbing of the revolution, not the short-term consumer fad.

The key catalysts that could validate this thesis are already in motion. First, watch Tesla's robotaxi service expansion. The company's move to unsupervised rides in Austin is a major operational milestone. If this service ramps up, it will drive demand for the very infrastructure-power, data centers, and specialized software-that the smart money is accumulating. Second, monitor the integration of strategic acquisitions like Apple's purchase of the AI audio startup Q.ai. Such moves signal that even the tech giants are quietly building their AI moats, which could benefit the underlying software and hardware providers in the ecosystem. Finally, the quarterly 13F filings are the ultimate pulse check. These reports, due in early February for the Q4 2025 period, will show if the institutional accumulation in AI infrastructure and software is accelerating or stalling. A new wave of buying would confirm the thesis; a wave of selling would be a stark warning.

The bottom line is a high-stakes bet on patience. The smart money is positioning for the long build-out of AI's physical and digital backbone, a play that often gets overlooked during the initial hype cycle. The risks are clear: the market could remain skeptical, and the catalysts may take longer to materialize. But for those willing to follow the filings, not the headlines, the opportunity lies in the quiet accumulation of the foundation.