Looking at Nvidia’s current public equity holdings makes me feel nearly uneasy. It’s not because the amounts are small—roughly $1.8 billion is substantial money by any standard—but rather because of the way it’s set up. Just two businesses hold nearly three-quarters of that total. Two. In a time when diversification is taught as gospel in every first-year finance course, Nvidia has practically done the opposite with what seems to be total intentionality. Depending on the analyst desk you’re at, that may come out as visionary or obstinate.
With between 50 and 60 percent of the entire portfolio, Intel holds the biggest position. It takes a second to land when you pronounce that number out loud. Intel has spent the majority of the last few years navigating terrible mistakes, giving up significant territory to AMD in data center CPUs, and attempting to carry out a foundry turnaround that has strained investor patience at almost every turn. Nvidia’s stated reasoning is simple: support a local semiconductor manufacturer that can produce cutting-edge AI chips, strengthening the U.S. supply chain to reduce the industry’s reliance on TSMC facilities in Taiwan.
That argument has a certain consistency from a strategic standpoint. The logic seems credible, even persuasive, to many experts who cover semiconductor supply chains. The issue is pricing; according to multiple estimates, Intel’s stock is priced with optimism that hasn’t yet been rewarded by the performance.Supporting Intel could be just what the American chip industry needs, or it could be a very costly expression of faith in a turnaround that keeps getting delayed by a quarter.
A different kind of story is told in CoreWeave, the second component of the focus. It is smaller than Intel at 10 to 15 percent of the portfolio, but it may be more illuminating. CoreWeave, a dedicated cloud service designed from the ground up around high-performance GPU workloads and one of the first operators to deploy Nvidia’s Grace Blackwell GB200 systems at scale, is one of those businesses that emerged from the AI infrastructure boom looking indispensable.
Nvidia’s ownership of CoreWeave is comparable to a winery investing in the top restaurant that serves its bottles. It makes sense in ways that go beyond simple profit-seeking. In a market where hyperscalers like Amazon, Google, and Microsoft are covertly developing their own custom silicon and may eventually become completely independent of third-party cloud providers like CoreWeave, the partnership may be as much strategic as financial.
It’s difficult to ignore how Berkshire Hathaway was discussed for years when Apple grew to dominate its portfolio as all of this is taking place. Critics said that there was a concentration risk. The defenders said, “Brilliant discipline.” The argument was never completely settled; performance over time provided an answer. Although Nvidia’s circumstances are different, the dynamic is similar. Concentrated confidence begins to look less like recklessness and more like the kind of wager that only insiders feel entitled to make when you’re operating from a position of profound technical knowledge—when you may know the AI chip ecosystem better than anyone else on the planet.
However, it’s unclear if the structural challenges each company faces will be overcome by that insider confidence. The years of delays and cost overruns that have accompanied Intel’s foundry aspirations are true. Although CoreWeave’s growth is real, capital-intensive cloud enterprises are in a risky position when they compete with trillion-dollar digital firms. The portfolio’s flavor is not significantly altered by the lesser assets, such as Synopsys, Coherent, and Nokia, which feel almost like footnotes. These two lines, along with the roughly $1.3 billion in conviction they stand for, tell the story. Smart or careless? As is often the case with concentrated bets, the solution won’t become apparent until it has.