Nvidia CORP’s 13F: Doubling Down on AI Infrastructure Monopolies

With no trims in the top book, every incremental dollar of risk went into amplifying the existing thesis. The standout move is Coherent at 10.1% and $1.86B: a brand‑new, core‑sized position in photonics and optical components for high‑bandwidth networks and advanced manufacturing, added at an average cost that is already showing a 57.7% mark‑to‑market gain.

Just behind it, CoreWeave was pushed from a large to a defining position, now 19.91% of assets after a +94.5% increase in shares and a roughly $1.78B dollar add. That combination – lasers and optics via Coherent, GPU‑hungry AI workloads via CoreWeave – reads as a conviction call that the AI compute bottleneck is shifting from chips alone to networking, cooling, and specialized cloud operators who can actually monetize the hardware.

Around this, the manager has chosen to sit tight on its enabling infrastructure plays rather than press them: Synopsys at 10.4% and Nokia at 7.28% are unchanged. The message is that EDA, IP, and network equipment are still core ingredients, but the marginal dollar now chases the more operationally leveraged AI infrastructure nodes.

The 13F shows no trims among the top holdings this quarter, and that absence is itself informative. A book this concentrated, with Intel sitting on a 241.7% gain and Nebius up 698.4% versus average cost, could easily justify recycling winners; the decision not to touch them signals a willingness to let long-duration theses run.

The funding for Coherent and the CoreWeave add therefore has to be inferred as coming from outside the visible top‑50, or from new capital rather than internal rotation. That makes the move more aggressive: they’re not “upgrading” within the stack, they’re layering risk on top of already‑successful lines.

Unchanged stakes in Intel, Synopsys, Nokia, and Nebius – all held for six quarters – suggest that nothing in the core narrative has broken. Instead of dialing back volatility after an exceptional three‑year run, the manager is allowing concentration and path dependency to rise, effectively betting that the AI infrastructure cycle still has multiple turns left.

On paper, sector allocation is static: Technology remains 100.0% now, just as last quarter. Under the hood, however, the internal mix of Tech exposure is rotating from pure semis and software tooling toward the less crowded, more operationally complex plumbing of AI datacenters.

Intel remains the anchor semiconductor exposure at 51.58%, but the new Coherent position shifts roughly a tenth of the book into optics, lasers, and high‑speed interconnect – the gear that makes scale AI clusters physically possible. CoreWeave pushes the book further from components into a specialized cloud operator that monetizes GPUs and networking as a service.

Synopsys and Nebius continue to give the portfolio leverage to design‑time and software layers, while Nokia represents the broader telecom and networking buildout. The picture is of a manager intentionally clustering risk around every choke point in the AI compute supply chain – from chip design and manufacturing, through fiber and optics, up into GPU‑dense cloud infrastructure – rather than diversifying out of Tech cyclicality.

Taken together, this filing reads as a high‑conviction statement that the AI investment wave is still in its infrastructure phase, not its application phase. The firm is content to anchor half the book in Intel, leave long‑held winners untouched, and then lever up the “hard parts” of scaling AI – datacenter optics and specialized cloud capacity – via Coherent and CoreWeave.

The willingness to carry Synopsys at a modest unrealized loss, while adding to other capital‑intensive names, shows they are underwriting multi‑year returns across cycles rather than trading quarter‑to‑quarter sentiment. In that sense, the portfolio is less a basket of Tech stocks and more a systems‑level bet that AI compute intensity, and the capex to support it, will keep compounding.

For outside observers, the message is clear: Nvidia CORP is not hedging its dependence on the AI build‑out; it is concentrating into it. If the AI datacenter build slows, this book will feel it acutely – but if the cycle persists, this is exactly the kind of tightly focused, infrastructure‑first positioning that can continue to drive outsized returns.