NVIDIA’s AI compute dominance faces a new test as storage rallies on supply tightness

NVIDIA, the company that has become synonymous with accelerated AI computing, is facing a different kind of competitive storyline in the market. A new analysis from Yahoo Finance frames the debate as a choice between two “winners” in the AI supply chain, one centered on GPUs and the other on data storage hardware, specifically NAND flash used in devices and servers.

The comparison highlights how dramatically storage shares have moved during a period when NAND supply constraints have tightened. According to the Yahoo Finance piece, SanDisk’s stock performance has been driven by a NAND shortage cycle, with the report pointing to gains of more than 600% over the course of the year it is discussing. The article uses that surge to argue that storage could plausibly capture a larger share of AI-adjacent spending, at least in the near term.

On the NVIDIA side, the same analysis underscores what has been a central theme in AI markets for the last several quarters: demand for AI training and inference has flowed through compute first. NVIDIA’s argument, broadly, is that AI systems require massive parallel processing, and its data center platform has been built around scaling that compute efficiently. The market has largely rewarded that role, even as competition and alternative accelerator designs continue to emerge.

The core question, as posed by the Yahoo Finance comparison, is whether storage can become an AI “winner” on a sustained basis rather than as a cyclical beneficiary of a supply-and-demand imbalance. In other words, the article suggests that the NAND-driven rally is not the same thing as a structural shift in how AI infrastructure budgets allocate long-run spending. For storage to match the outsized influence of compute, the market would need evidence that AI workloads are consistently driving durable storage capacity and memory-related demand across cycles.

Storage in AI systems is not just a background component, because modern AI deployments generate and move very large volumes of data. That said, the relative balance between compute intensity and data movement varies by workload type. Training can be compute-heavy, but inference and large-scale deployment also depend on how quickly models can be served, what data must be accessed, and how efficiently systems can store, stage, and retrieve information at scale.

NVIDIA’s broader position in the AI stack is shaped by this compute-first reality. The company has positioned its platforms to serve data center customers building training clusters and inference pipelines, and it has also pushed out related software and system-level integrations through its ongoing technology announcements. The industrial challenge for any storage contender is therefore to show that storage requirements are not only growing, but doing so in a way that sustains pricing, volume, and margins through the next industry cycle.

Still, the Yahoo Finance comparison does not settle the debate with specific disclosures from NVIDIA or SanDisk about medium-term storage demand for AI, nor does it provide detailed, workload-by-workload evidence. It essentially relies on the observed stock move and the supply tightness narrative to frame what could be an attractive investment theme. Without company guidance or segment-level disclosures tying AI directly to storage consumption at scale, investors are left to infer whether the NAND shortage-driven upswing can transition into a durable AI-driven growth story.

What to watch next is whether storage makers and AI compute vendors both update the market with clearer demand indicates. For storage, that would mean evidence that AI workloads are translating into steadier capacity expansion and not just a one-time pricing recovery. For NVIDIA, the key variable is whether AI compute continues to command the largest share of incremental infrastructure spending even as systems mature and customers optimize end-to-end performance.