Jim Cramer said investors should buy Intel stock, suggesting the company could be more instrumental to AI data-center workloads than Nvidia [1].

This perspective shifts the focus of the AI hardware race from GPUs to CPUs. While Nvidia has dominated the initial AI surge, Cramer said that the underlying infrastructure for data centers requires a different set of capabilities that Intel is uniquely positioned to provide [1, 2].

Speaking during the CNBC Investing Club Monthly Meeting on June 11, Cramer said the surge in CPU demand is driven by AI infrastructure [2]. He said that Intel's emerging foundry business, which allows the company to manufacture chips for other firms, could serve as a significant growth driver [1].

Intel stock has already seen a substantial rally this year, climbing 230% year-to-date [3]. Despite this increase, Cramer said the stock remains a buy and that he sees further upside ahead [3].

"If you don't own Intel, please buy it," Cramer said [2].

He said the company is his new favorite stock [3]. The push for Intel comes as the industry evaluates the long-term balance between specialized AI accelerators and the general-purpose processing power provided by CPUs in massive data-center environments [1, 2].

Cramer's analysis centers on the idea that as AI scales, the need for robust CPU integration and diverse manufacturing capabilities will make Intel's ecosystem indispensable [1]. This strategy relies on the successful execution of the foundry model to capture a wider share of the semiconductor market [1].

“"If you don't own Intel, please buy it."”

Cramer's endorsement reflects a broader market thesis that the 'AI trade' is evolving. While the first wave of AI investment focused on the chips that train models, the next phase may prioritize the infrastructure that hosts and manages these workloads at scale. By emphasizing the foundry business and CPU demand, Cramer is signaling that Intel's ability to both design and manufacture chips could provide a competitive edge in the physical build-out of global AI data centers.