NextFin News - In a series of high-stakes portfolio adjustments revealed in its latest regulatory filings, semiconductor powerhouse Nvidia has completely exited its positions in Arm Holdings and Applied Digital. According to Nvidia’s Form 13F filing for the fourth quarter of 2025, the company liquidated 1,101,249 shares of Arm Holdings and 7,716,050 shares of Applied Digital. These divestments come at a time when both companies have seen their valuations reach historic highs, with Applied Digital alone surging between 500% and 1,000% over the past 18 months. While exiting these positions, Nvidia has seen its newest strategic investment—a $5 billion stake in Intel—double in value since the deal was finalized in late December 2025. As of March 3, 2026, Nvidia’s 214.7 million shares of Intel represent an unrealized gain of approximately $5 billion in just over 60 days.
The decision to dump Arm Holdings marks a significant symbolic and strategic shift for Nvidia. After a failed $40 billion attempt to acquire the UK-based chip designer in 2022 due to regulatory hurdles, Nvidia had maintained a minority stake following Arm’s 2023 IPO. However, with Arm’s price-to-sales ratio hovering near 29, the valuation appears to have reached a ceiling that no longer justifies the capital allocation for Jensen Huang’s firm. Similarly, the exit from Applied Digital—a specialized AI data center operator—suggests a tactical retreat from high-premium service providers in favor of direct infrastructure control. Applied Digital’s market capitalization of $8.1 billion against relatively modest sales created a "frothy" valuation environment that Nvidia’s management likely viewed as an optimal exit point for profit-taking.
The pivot toward Intel is perhaps the most consequential move in the current AI hardware landscape. By securing 214.7 million shares at an agreed price of $23.28 per share, Nvidia has effectively hedged its architectural bets. While Nvidia’s own Grace and Vera CPUs are built on Arm architecture, the massive investment in Intel signals a deepening commitment to the x86 ecosystem. This collaboration aims to create a new generation of AI data centers that pair Nvidia’s industry-leading GPUs with Intel’s x86-based central processing units. According to Nasdaq, this partnership is not merely financial; it is a structural alignment designed to ensure Nvidia’s networking solutions and GPUs remain the backbone of data centers, regardless of whether the underlying CPU architecture is Arm or x86.
From an analytical perspective, Nvidia’s portfolio rebalancing reflects a broader industry trend: the transition from AI training to AI inference and orchestration. As agentic AI environments—where AI models act as autonomous agents—become more prevalent, the demand for architectures that prioritize memory access and complex orchestration is rising. By diversifying its CPU roadmap across Arm, RISC-V, and now a heavily subsidized x86 partnership with Intel, Nvidia is insulating itself against architectural shifts. The doubling of the Intel investment in just two months provides Nvidia with a massive capital cushion, effectively funding its next phase of R&D through market gains alone.
Looking forward, Nvidia’s exit from Arm does not signal a divorce from the technology; rather, it signals a shift from being an equity partner to being a dominant customer. Nvidia will continue to utilize Arm’s Neoverse V2 architecture for its Grace CPUs, but it no longer needs to hold the stock to influence the roadmap. The capital freed from the Arm and Applied Digital sales is likely being earmarked for the "sovereign AI" movement, where nations are building localized data centers. As U.S. President Trump continues to emphasize domestic manufacturing and technological leadership, Nvidia’s deeper tie-up with Intel—the crown jewel of American semiconductor fabrication—aligns perfectly with the current administration’s industrial policy. This strategic positioning suggests that Nvidia is moving beyond being a component supplier to becoming the ultimate integrator of the global AI infrastructure stack.
Explore more exclusive insights at nextfin.ai.
