Nvidia (NASDAQ:NVDA) finds itself in an unusual position: the world's leading AI chipmaker delivered another blockbuster quarter, yet its stock is trailing the broader semiconductor rally. While the company reported 85% revenue growth and guided for approximately $91 billion in next-quarter revenue, its shares have risen only about 12% in 2026. In contrast, the VanEck Semiconductor ETF has surged more than 80% this year, and the PHLX Semiconductor Index has posted historic gains.
This divergence is not a reflection of weak execution. Nvidia remains dominant, with CEO Jensen Huang describing the AI buildout as "the largest infrastructure expansion in human history." Instead, the market is signaling that the AI hardware trade is broadening beyond Nvidia, and investors are becoming more selective about valuations.
Memory and Infrastructure Names Take the Lead
The semiconductor rally has shifted focus to memory, storage, and networking companies. Micron and Sandisk have emerged as new favorites as analysts bet that AI data centers will require massive amounts of memory and storage capacity alongside GPUs. This rotation reflects a maturing view of the AI supply chain, where infrastructure components beyond processors are gaining attention.
According to Barron's, Nvidia now trades at about 20.2 times forward earnings, a discount to the semiconductor index average of 26.8 times. That valuation gap suggests investors respect Nvidia's dominance but are no longer willing to pay peak multiples for it. The rally has broadened into custom silicon, networking, and other infrastructure plays, as seen in recent gains for companies like Marvell, which Goldman Sachs analyst James Schneider sees benefiting from higher hyperscaler capital expenditure and optical networking demand.
Hyperscalers Build Their Own Alternatives
The biggest risk to Nvidia's premium valuation is the growing push by its largest customers to develop proprietary chips. Microsoft and Meta are working on in-house AI processors to reduce dependence on Nvidia. Amazon's Trainium and Google's TPUs are gaining traction as cost-effective alternatives for certain workloads. While these efforts are unlikely to displace Nvidia overnight, they give hyperscalers negotiating leverage and signal a more diversified sourcing strategy.
Wedbush analyst Matt Bryson noted to Barron's that there is "growing demand for alternative solutions" because "there are all these alternatives." Mizuho's Lloyd Walmsley struck a more cautious tone, saying "a lot remains unknown today" about how far in-house silicon can go. Still, the trend is clear: Nvidia's customers are actively building options.
Bull Case Remains, But With Caveats
UBS analyst Timothy Arcuri argues that Nvidia has built a formidable moat around its hardware, software, and networking ecosystem. However, even he flags that investors may be underestimating the opportunity for AMD and server CPUs as agentic AI shifts computing workloads. The bull case is still powerful, but it now comes with a caveat: Nvidia's dominance is no longer unchallenged, and the AI hardware boom is no longer its alone to command.
For investors tracking the broader semiconductor landscape, the rotation into memory and custom silicon names is a key theme. For more on Nvidia's recent moves, see Nvidia Dips 0.18% Despite China Vera CPU Orders, Global AI Infrastructure Push and Nvidia Stock Rebounds 2% on South Korea AI Deals; CEO Calls Selloff a Discount. Meanwhile, the rise of custom silicon is also benefiting companies like Cerebras, as discussed in Cerebras Surges 17% as Analysts Back Its Nvidia-Challenging AI Chip Strategy.
This article is for informational purposes only and does not constitute financial advice.
