Follow the Money: NVIDIA’s 13F Reveals the Next AI Winners

If you thought the AI trade peaked in 2023, think again.

While the hype around anything labeled “AI” has cooled since last year’s frenzy, the market is now shifting toward fundamentals — and the kingmaker is clearly NVIDIA $NVIDIA(NVDA)$ . Forget who’s talking about AI. The new question is: Who’s actually in NVIDIA’s inner circle?

NVIDIA just filed its latest 13F, revealing six investments that offer a blueprint for where it thinks the next wave of AI growth will come from. It’s a full-stack bet — from raw compute to real-world applications — and might give us a serious edge if we’re looking to “copy its homework.”

Let’s take a closer look at each name.

1. $CoreWeave, Inc.(CRWV)$

Key themes: Cloud compute | GPU infrastructure | NVIDIA’s top pick

NVIDIA stake: ~$897 million (78% of disclosed investments)

CoreWeave is the centerpiece of NVIDIA’s portfolio — accounting for nearly 80% of its reported holdings. It doesn’t design chips. Instead, it buys NVIDIA GPUs in bulk, builds data centers, and rents out AI compute power to companies like OpenAI and Anthropic. Think of it as a “GPU landlord.”

With 32 data centers and over 250,000 GPUs under management, CoreWeave is tightly integrated into NVIDIA’s ecosystem. It has long-term compute deals with OpenAI (worth $4B+) and backing from Microsoft, which is both a customer and an investor.

Why it matters: NVIDIA’s investment in CoreWeave is a power move — locking in control over downstream compute supply as AI demand surges. For investors, this suggests CoreWeave could become the AWS of AI infrastructure.

2. $ARM Holdings(ARM)$

Key themes: Chip architecture | Edge AI | Licensing play

NVIDIA once tried (and failed) to buy Arm for $40 billion. Now it’s circled back with a public market stake.

Arm’s IP is foundational to low-power chips used in everything from smartphones to smart cars. As AI expands to edge devices — wearables, autonomous vehicles, IoT — Arm’s architecture is positioned to ride that wave. The company is also developing its own AI-focused chip designs, aiming to move higher up the value chain.

Why it matters: While not a high-growth rocket ship, Arm is a “pick-and-shovel” play on AI hardware. It earns steady licensing fees and powers much of the AI edge future. NVIDIA’s move here shows it still wants in on the Arm ecosystem — even without an acquisition.

3. $APPLIED DIGITAL CORP(APLD)$

Key themes: Mid-tier AI compute | Data centers | Cost-efficient infrastructure

A former crypto miner turned AI data center operator, Applied Digital is a niche bet on scalable, lower-cost compute services. Like CoreWeave, it buys NVIDIA GPUs and offers compute-as-a-service — but with a focus on small to mid-size AI startups who can’t afford to rent from hyperscalers.

Why it matters: As AI models proliferate, not every company will use high-end compute from Microsoft or AWS. APLD is targeting the long tail of demand, and NVIDIA’s investment here shows it’s betting on that “middle market” of AI growth too.

4. $Recursion Pharmaceuticals, Inc.(RXRX)$

Key themes: AI + drug discovery | Biotech meets compute | Vertical AI application

Recursion applies AI to biology. Its platform uses high-throughput cell imaging and large models to discover new drug candidates — aiming to compress the drug development timeline. It’s already partnered with Bayer and Roche, and is a major customer of NVIDIA’s DGX platform and BioNeMo AI stack.

Why it matters: This is NVIDIA’s clearest bet on vertical AI — using GPU compute not just for language models, but for real-world breakthroughs. Recursion could become the poster child for AI-powered biotech.

5. $NEBIUS(NBIS)$

Key themes: Eastern European cloud | Affordable AI infrastructure

Born from the former Yandex AI team, Nebius is now a standalone cloud provider focusing on affordable compute solutions across Eastern Europe and Israel. Like others on this list, it’s building its stack around NVIDIA GPUs.

Why it matters: It’s a small bet, but reflects NVIDIA’s global strategy — embedding its tech into new markets and setting the stage for emerging-region AI adoption.

6. $WeRide Inc.(WRD)$

Key themes: Robotaxis | Autonomous driving | China AI exposure

WeRide is the first pure-play Robotaxi company to go public, focused on L4 self-driving systems. NVIDIA was an early investor and remains deeply integrated — WeRide’s AV systems rely heavily on NVIDIA’s Drive Orin chips.

Why it matters: For NVIDIA, this isn’t just about returns. It’s about closing the loop on its automotive AI ecosystem — from chips to algorithms to vehicle deployment. Think of WeRide as NVIDIA’s “reference customer” for robotaxi use cases.