Artificial intelligence chipmaker Nvidia recently boosted its dividend, surpassed earnings and revenue expectations, and reported an impressive 75% adjusted gross margin. Despite these strong financial results, the stock experienced a dip in extended trading following the release of its quarterly figures on Wednesday.
The current phase of AI development, often referred to as the "agentic" phase, is shifting focus back towards central processing units (CPUs) and more decentralized system architectures. This trend has benefited other chipmakers like Intel, Micron, and AMD, which have seen significant gains in recent months, while Nvidia's performance has been comparatively more modest.
During Wednesday's earnings call, analysts were keen to understand if Nvidia can maintain its competitive edge as the demand for chip types diversifies. CJ Muse, managing director at Cantor Fitzgerald, directly questioned Nvidia CEO Jensen Huang about the impact of the upcoming Vera Rubin AI system on Nvidia's market share in the inference sector. Inference refers to an AI's capability to respond to queries and execute tasks for users and businesses.
Huang emphasized multiple times that Nvidia's market share in inference is indeed growing rapidly. However, the landscape of AI chip production is evolving. While Nvidia has long been synonymous with graphics processing units (GPUs), the new "agentic" AI paradigm, which involves semi-autonomous bots performing tasks independently, necessitates a more distributed processing design relying on a greater number of traditional CPUs alongside GPUs.
Although Nvidia remains the dominant player, this shift in infrastructure is intensifying analyst interest in the competitive dynamics. It remains unclear whether a single architecture will dominate the market or if chipmakers will increasingly specialize in different areas. Potential disruptors include rival custom silicon products like Alphabet's Tensor Processing Units and Amazon's Trainium chips, as well as offerings from AMD and the newcomer Cerebras.
Despite potential competition, Huang indicated that production capacity for Nvidia's forthcoming Vera Rubin chip platform might be fully booked even before its launch later this year, stating, "My sense is that we'll be supply constrained throughout the entire life of Vera Rubin."
Analysts are observing a sustained upward trend in CPU and memory demand. Nicolas Gaudois, an analyst at UBS, noted on Wednesday, "Server CPU demand is strongly inflecting up, driving demand … for both conventional server as well as AI servers (CPU head nodes). Agentic AI is a key driver."
