IBM's leadership has championed an AI-on-the-mainframe future, with generative AI bridging the COBOL skills gap left by previous generations of technology experts. Arvind Krishna, IBM's chairman, president, and CEO, made these remarks as the company reported full-year results highlighting a significant increase in mainframe sales. Big Blue posted revenue from continuing operations of $19. 7 billion in the fourth quarter, marking a 12 percent year-over-year rise and resulting in net income of $5. 6 billion, up 91 percent. For the entire year, revenue increased 8 percent to $67. 5 billion, while net income soared 76 percent to $10. 6 billion. AI was naturally a key focus during the firm’s briefing to analysts on these figures. “Our cumulative GenAI book of business now exceeds $12. 5 billion, with Software accounting for over $2 billion and Consulting more than $10. 5 billion, both segments experiencing their largest quarterly growth to date, ” Krishna stated. He also described the technology as a “powerful productivity driver” internally. “In 2023, we set a target of achieving $2 billion in productivity savings by the end of 2024. Currently, we are well ahead, projecting $4. 5 billion in annual run-rate savings by the end of 2025. ” The firm’s infrastructure business showed the strongest growth, with revenues up 12 percent for the year and 21 percent in the fourth quarter. Software revenues rose 11 percent for the year and 14 percent in the fourth quarter.

Consulting lagged, growing just 2 percent over the year and 3 percent in the fourth quarter—translating to flat and 1 percent growth, respectively, on a constant currency basis. Much of the infrastructure growth stemmed from the launch of IBM’s z17 mainframe series. In his prepared remarks, Krishna noted, “Innovation value is evident in our IBM Z performance, up 48 percent this year, delivering the highest annual revenue for Z in approximately 20 years. ” IBM CFO James Kavanaugh highlighted the “record z17 launch, achieving the highest annual revenue for IBM Z in around two decades, and outpacing the z16 performance in the first three quarters of its program. ” Krishna emphasized that this was about more than replacing legacy systems. Sovereignty is increasingly critical, with “more clients realizing that for specific workloads, the mainframe represents the platform with the lowest unit cost economics, which is crucial. ” He also pointed out that GenAI facilitates leveraging and modernizing mainframes: “The GenAI tools provided through Watson Code Assistant for Z remove much of the burden. It can refactor COBOL into Java or assist in refactoring code if you wish to retain it as is. ” Krishna added, “I’m thrilled by our capability to execute AI inline. Performing AI directly in line with transactions introduces only a millisecond delay, compared to multiple seconds if processing is off-platform, which has been the norm. ” However, alongside the mainframe’s role in AI’s future, some persistent challenges remain. When asked if AI- and HBM-driven DRAM price increases posed an issue, Krishna responded, “I personally believe these pricing pressures will persist as long as this dynamic exists. ” He further explained, “There is no AI server without numerous CPUs adjacent to it. Thus, AI demand inherently fuels demand for standard servers, which then supply and load those AI servers. ” ®