Is this U.S.-China selloff a buy? A top Wall Street voice weighs in
ARMONK, N.Y. - IBM (NYSE:IBM), a prominent player in the IT Services industry trading near its 52-week high of $296.16, announced Tuesday that its Spyre Accelerator, designed for AI inferencing workloads, will become generally available on October 28 for IBM z17 and LinuxONE 5 systems, with availability for Power11 servers following in early December. According to InvestingPro data, IBM has demonstrated strong momentum with a 34.32% return year-to-date.
The hardware accelerator aims to support generative and agentic AI applications while maintaining security and resilience for enterprise workloads. According to the company’s press release statement, the Spyre Accelerator is designed to enable low-latency inferencing for AI models without compromising throughput on critical business operations. With a solid financial health score of "GOOD" from InvestingPro, IBM continues to demonstrate its capability to invest in innovative technologies while maintaining strong business fundamentals.
The accelerator is built as a system-on-a-chip with 32 individual accelerator cores and 25.6 billion transistors using 5nm node technology. Each unit is mounted on a 75-watt PCIe card, allowing for clustering of up to 48 cards in IBM Z or LinuxONE systems, or 16 cards in IBM Power systems.
"With the Spyre Accelerator, we’re extending the capabilities of our systems to support multi-model AI including generative and agentic AI," said Barry Baker, COO of IBM Infrastructure and GM of IBM Systems, in the announcement.
The product emerged from IBM’s Research AI Hardware Center, which was established in 2019. It underwent testing at IBM’s Yorktown Heights campus and with collaborators including the University at Albany’s Center for Emerging Artificial Intelligence Systems before being developed into an enterprise-grade product.
For IBM Power-based servers, the company states that Spyre customers will have access to a catalog of AI services that can be installed with a single click. When combined with the on-chip accelerator, the system reportedly enables the ingestion of more than 8 million documents for knowledge base integration in an hour with a prompt size of 128.
The accelerator is designed to allow organizations to keep mission-critical data on-premises while addressing operational and energy efficiency requirements. With revenue of $64.04 billion in the last twelve months and a track record of 29 consecutive years of dividend increases, IBM continues to demonstrate its commitment to both technological innovation and shareholder returns. For deeper insights into IBM’s financial metrics and growth potential, consider accessing the comprehensive Pro Research Report available on InvestingPro, which covers over 1,400 top US stocks with expert analysis and actionable intelligence.
In other recent news, IBM announced new AI productivity tools at its TechXchange 2025 event, aiming to enhance AI adoption and productivity across various business operations. The company introduced updates to its watsonx Orchestrate platform, including AgentOps for governance and agentic workflows for developers. Additionally, IBM has entered into a strategic partnership with Anthropic to integrate the Claude large language model into select IBM software products, reportedly improving productivity by 45% during internal testing. IBM also committed significant resources to support Datavault AI’s platform development, focusing on data authentication and security. In another collaboration, IBM partnered with AMD to advance quantum computing architectures, combining IBM’s quantum technology with AMD’s computing capabilities. Furthermore, IBM was among the foreign companies represented at a high-level meeting in Guangzhou with China’s vice commerce minister. These developments underscore IBM’s active role in advancing AI and quantum computing technologies in the enterprise sector.
This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.