IBM’s mainframe business has achieved a milestone it hadn’t seen in over two decades. The company reported its best fourth-quarter revenue in more than 20 years, with a 61% year-over-year increase adjusted for currency, significantly boosting the infrastructure segment by 17%.
This surge comes despite the widespread perception that mainframe technology is outdated. Contrary to this belief, IBM’s mainframes continue to serve as critical infrastructure, especially for industries requiring high levels of security and reliability. Data indicates that 71% of Fortune 500 companies rely on mainframes, with IBM systems dominating more than 90% of this installed base.
Mainframes Powering Critical Transactions
IBM mainframes are integral to the global financial ecosystem. Over 87% of worldwide credit card transactions are processed using these systems. Moreover, about 92% of large financial institutions and 63% of government agencies continue to utilize mainframes in their core operations. This underscores the ongoing dependence on the technology for mission-critical workloads.
IBM maintains relevance by continuously evolving its mainframe capabilities. The latest iteration, the IBM z17 mainframe system, launched in mid-May, is specifically engineered to support the growing demands of artificial intelligence (AI). This model can manage more than 250 AI use cases ranging from loan risk assessment to medical image analysis.
AI Inference Capabilities on the z17
The z17 model can process up to 450 billion AI inferencing operations daily, marking a 50% performance increase over its predecessor. It delivers an unprecedented one-millisecond average response time for AI tasks, making it highly suitable for real-time applications where latency is critical.
IBM offers additional modules like the Spyre AI accelerator, which can be integrated into z17 systems. This upgrade enables enhanced AI model execution, including IBM’s Granite AI models, further improving the mainframe’s AI processing performance.
Strategic Vision for AI Workloads
IBM CEO Arvind Krishna highlighted a notable shift in AI deployment trends. He forecasts that within three to five years, 50% of enterprise AI usage will migrate away from public cloud platforms to private clouds or on-premises data centers. This expectation arises from the high cost of running AI inference in public clouds and the efficiency gains of processing stable AI workloads locally.
Given IBM’s practice of introducing new mainframe models every 2.5 to 3 years, the z17 will likely dominate the company’s mainframe offerings until 2028. Future models are expected to escalate AI inferencing capabilities even further.
IBM’s Broader AI Ecosystem
Beyond mainframes, IBM’s AI efforts include consulting services and the watsonx AI platform. With a vast enterprise user base, the company leverages its mainframe stronghold to sell integrated AI solutions. IBM projects a revenue growth of at least 5% in 2026, alongside a 10% rise in software sales. The company also anticipates a $1 billion improvement in free cash flow from the reported $14.7 billion in 2025.
Although mainframes have a smaller share in IBM’s overall business than two decades ago, they remain a significant contributor to profits and cash flow. By evolving the mainframe architecture to excel in the AI era, IBM has transformed what many considered obsolete technology into a vital asset supporting next-generation workloads.
Read more at: www.fool.com
