Qualcomm Unveils AI200 and AI250 Data Center Chips, Stock Rises 20 Percent

Qualcomm on Monday introduced two new artificial intelligence processors aimed at data centers, marking its most substantial move to date beyond the smartphone segment that has long defined its business. The company said the AI200 and AI250 chips are engineered to handle inference workloads, offer expanded memory capacity and integrate with commonly used AI software frameworks. Qualcomm expects the AI200 to reach commercial availability in 2026, followed by the AI250 in 2027.

The announcement drove a sharp market response. Qualcomm’s shares advanced about 20 percent after the products were revealed, reflecting investor confidence in the firm’s strategy to tap demand for data-center hardware that can run large language models, chatbots and other generative AI systems. Spending on such infrastructure has accelerated as cloud providers and enterprises seek alternatives to Nvidia, the current leader in AI accelerators.

According to Qualcomm, both chips support popular AI development tools out of the box, which the company says will help customers migrate existing models without extensive rewrites. Executives highlighted potential cost savings for enterprises that adopt the new hardware, although detailed pricing was not disclosed. Qualcomm also introduced complete racks built around the chips, positioning itself to supply end-to-end systems rather than individual components—a strategy similar to recent offerings from Nvidia and rival Advanced Micro Devices (AMD).

The product launch extends Qualcomm’s diversification plan. The San Diego-based firm remains the largest supplier of modem chips for smartphones, but handset demand has slowed and the company has faced pressure from key customers. It lost significant orders from Chinese device maker Huawei following U.S. trade restrictions, while Apple continues to develop its own cellular and application processors. To offset that exposure, Qualcomm has spent the past two years expanding into personal computers, where it supplies processors for Windows laptops in competition with Intel and AMD.

Competition in data-center AI silicon is intensifying, yet Nvidia’s dominant market share presents obstacles for newcomers. Cloud operators and enterprises that already run Nvidia hardware often face high costs to switch providers because of software investments and established performance benchmarks. Qualcomm executives acknowledged those challenges but argued that demand for efficient, scalable alternatives is large enough to support multiple suppliers.

Initial adoption commitments surfaced alongside Monday’s unveiling. Humain, an artificial intelligence startup backed by Saudi Arabia’s Public Investment Fund, plans to deploy 200 megawatts of data-center capacity based on Qualcomm’s new racks beginning in 2026. Qualcomm described the project as evidence that customers are seeking to diversify their computing options, particularly in regions investing heavily in digital infrastructure.

Imagem: financial planning 62

Industry analysts noted that the Humain agreement could broaden Qualcomm’s footprint in the Middle East and demonstrate real-world performance of the AI200 before the higher-capacity AI250 ships the following year. While volume details beyond the 200-megawatt installation were not provided, Qualcomm indicated that discussions with additional hyperscale and enterprise buyers are ongoing.

The rollout underscores Qualcomm’s effort to balance its legacy handset revenue with emerging categories such as data centers and personal computing. Management has previously signaled that it views AI as a multiyear growth driver, and Monday’s stock reaction suggested investors are receptive to that outlook despite the substantial capital required to enter a market dominated by an established incumbent.

Crédito da imagem: Reuters