Nvidia Back in Focus and AI Infrastructure Acceleration

Each quarter, we break down the evolving world of semiconductors through the lens of Nvidia’s earnings and the broader AI investment landscape. This latest call came at a pivotal moment, with demand surging, CapEx accelerating, and the semiconductor ecosystem transforming faster than most expected.

Watch Video
Beyond Nvidia Earnings: AI CapEx Surge, Competition, and the Future of Semiconductors

Nvidia is “Back in the Game,” But It's Not Alone

While Nvidia remains the headline act, the semiconductor narrative is no longer a one-player story. After spending some time “on the bench,” Nvidia has returned to the spotlight, thanks to renewed demand, sovereign AI expansion, and robust margins on next-gen products. But the real story lies in the entire stack—from foundational GPU builders like Nvidia and Broadcom to the less visible but crucial chip designers in the fabless space.

What’s changed? For one, the pace. From model development to chip deployment, the speed of innovation and adoption, particularly across hyperscalers and enterprise infrastructure, has exceeded expectations. This confirms a core view we’ve held that diversified exposure, particularly through vehicles like the VanEck Semiconductor ETF (SMH) and the VanEck Fabless Semiconductor ETF (SMHX), is critical in capturing this sector’s momentum without betting on a single name.

1-Year Performance Snapshot: Nvidia Returns, but Broadcom Leads

Source: Morningstar, as of 6/2/2025 Past performance is no guarantee of future results. Index performance is not representative of fund performance. It is not possible to invest directly in an index. This is not an offer to buy or sell, or a solicitation of any offer to buy or sell any of the securities mentioned herein. Fund holdings will vary.

CapEx: Not a Bubble, But a Baseline

A key theme this quarter was capital expenditure—its scale, sustainability, and strategic importance. Nvidia’s call emphasized sovereign AI deals (notably in the Middle East), rising enterprise demand, and hyperscalers racing to future-proof infrastructure. CapEx tied to AI infrastructure isn’t showing signs of topping; if anything, forecasts may still be too conservative.

Why it matters: Token usage, a proxy for how heavily AI tools are being used, is doubling every six months. That translates directly to more compute, which requires more chips, more power efficiency, and more architectural innovation. In short, the rails are down, and the trains are speeding up.

CapEx Surge as Hyperscalers Ramp Spending into 2025

Source: Amazon, Microsoft, Alphabet, Meta as of 4/30/2025. ^*Past performance is no guarantee of future results. This is not an offer to buy or sell, or a solicitation of any offer to buy or sell any of the securities mentioned herein. Fund holdings will vary.

Beyond General-Purpose Compute: The Rise of Custom Silicon

The architecture of AI is shifting. Specialized chips—ASICs from names like Broadcom, Google (TPUs), and Amazon (Tranium)—are no longer secondary players; they’re core to how hyperscalers are optimizing cost and performance. This has implications up and down the stack. Nvidia’s pivot toward more tailored inference solutions is part of this evolution.

Meanwhile, fabless semiconductor designers, often behind the scenes, are developing chips that improve bandwidth, reduce power consumption, and unlock efficiency across the AI workflow. These innovators may not be household names yet, but their impact on AI deployment is foundational. It’s why we see SMHX as a space with significant long-term potential—it captures the "brains" behind the operation, not just the brawn.

Final Thoughts: AI Is the Story and Semiconductors Are the Backbone

If there’s one takeaway from this quarter’s discussion, it’s this: AI adoption isn’t slowing. It’s broadening. From sovereign initiatives to data centers to consumer internet, the demand for compute is feeding a flywheel of innovation, and semiconductors are at its center. Whether it's the industry giants or the rising specialists, this ecosystem is dynamic, diversified, and far from peaking.

We’ll continue to monitor the landscape closely each quarter. For now, it’s clear: AI and semis are inextricably linked, and the runway ahead is long.