Jim Cramer reiterates his “own Nvidia, don’t trade it” stance, citing growing government demand for AI chips

Nvidia again made a central appearance on CNBC’s Mad Money, where Jim Cramer told viewers he is “not changing” a long-running personal rule for the stock: “own NVIDIA, don’t trade it.” The remark came as Cramer pointed to a tougher backdrop for markets, but he framed Nvidia as an exception because he expects demand to remain durable and widening.

Cramer’s argument emphasized that Nvidia’s customer base is not limited to the biggest hyperscale cloud providers. He said countries are buying Nvidia’s offerings too, describing them as not focused on quick turnarounds and arguing that such purchases can reduce dependence on a handful of large buyers.

In the segment, Cramer also cited a figure he associated with Nvidia’s business exposure to these government and “real country” buyers, stating they are “14% of the business.” He added that this share could grow quickly, and he suggested the impact could be larger when including what he called “what’s in the pipe,” with his view implying the ratio could be “substantially higher” by next year.

Cramer said he believes sovereign AI procurement would also ease fears that some hyperscalers do not want Nvidia, even if those hyperscalers remain a key part of the AI supply chain. He tied this to his broader question about payback and pricing, saying the answer likely depends on who is buying, but that sovereign buyers are purchasing Nvidia chips “in droves.”

For readers unfamiliar with the technology at issue, Nvidia’s core position is in accelerated computing for artificial intelligence. The company designs GPUs, or graphics processing units, and related software and platform capabilities used to train and run AI workloads, and it sells systems and services across data centers, gaming, and enterprise uses.

That context matters because Nvidia’s stock narrative has often turned on how concentrated demand is at the top of the market. Cramer’s “14%” framing is designed to address concentration risk by pointing to additional classes of buyers, including governments pursuing local or protected AI capabilities.

The one area the public segment did not clarify was how the “14%” number is defined, whether it is based on a specific reporting line, and what time frame or geographic scope it covers. It also did not provide new earnings figures, contract names, or a disclosure-backed breakdown of pipeline orders, so investors looking for precision would still need to reconcile the commentary with Nvidia’s latest filings and guidance.

Going forward, investors will likely watch whether Nvidia’s own disclosures begin to reflect a broader mix of customers tied to government or sovereign AI spending, and whether management comments on order visibility support the idea that “what’s in the pipe” is meaningfully large. In the near term, the market will also continue to weigh how much of Nvidia’s growth depends on hyperscaler capital spending versus procurement outside that core group. (Cramer’s comments are opinion and are not a substitute for company guidance.)