Over two thirds of IBM’s clients running mainframes are increasing their workloads – and its z16 cycle was its “most successful program in history.”
That’s according to IBM’s CFO Jim Kavanaugh, reporting Big Blue’s FY2024 earnings late Wednesday. He spoke as its “transaction processing” revenues jumped 10%. (That covers IBM’s mainframe software offerings.)
Transaction processing accounted for approximately a third of IBM’s software segment revenues. Red Hat also performed strongly, with revenues up 17% (OpenShift ARR hit $1.4 billion up ~25%.)
(As The Stack noted in 2023, IBM revealed at an investor conference that year that its mainframe software licences have a magnificent profit margin of about “$0.80 to $0.90 on every dollar of growth.”)
Some “70% of the clients on mainframe are growing MIPS. That is a very different profile than where we were 10 years ago” he told analysts.
“We run [mainframe] workloads for 97% of the mission-critical transactional processing around… banking, retail, airlines, you name it.
“But we run it as a stack economic platform play…
See also: A $300 million cloud bill triggered a rethink - and a shopping spree on modular hardware
“We grew our transaction processing 10% in 2024, exiting at 11%... four points of that growth is due to the capitalisation of the underlying workloads that are driving mainframe. About three points of that is the investment in new innovation that we've been bringing to the mainframe platform… about three points of that is historical price optimisation.” Jim Kavanaugh
IBM’s infrastructure segment revenues fell and “z” mainframe revenues were actually down 21% for the quarter ahead of a refresh cycle.
(IBM will release its new z17 systems in “mid-2025” said Kavanaugh.)
Both hyperscalers and smaller players continue to try and peel mainframe workloads away from IBM-centric hardware and software.
The challenge for most CIOs isn't the hardware though – and although IBM licences might be expensive, the argument that the public cloud will be a lot cheaper if they can just be refactored for x86 carries less weight in many quarters than it used to. The issue is old applications, mostly...
"A lot of knowledge has disappeared..."
CIOs regularly tell The Stack that they have no problem with running workloads on mainframe. As one Group COO at a major bank told us just this month: “IBM does a very good job in offering a solid, resilient, modern architecture on mainframes. [The issue is] more the code…
“[When it comes to] old COBOL applications that may date back to the ‘80s of the last century… a lot of knowledge has disappeared; they’re not well documented, with hardly any inline comments, and sometimes not even fit for purpose anymore. Most were built in the days where the rhythm of the bank was defined by the opening hours of a branch.
COBOL, insist some advocates, is still very much fit for purpose.
As Misty Decker put it in The Stack's pages a few years back: "COBOL’s core design required it to be portable across computing devices.
"That core principle remains intact such that COBOL can even now compile and execute on any server environment.
"But COBOL hasn’t sat still for these past 60+ years. Continued investment and innovation from all manner of communities, private companies and individuals have enabled it to coexist and integrate with a wide array of contemporary technologies, including JVM, .NET, AWS, Azure and, Containers, as well as more established production server environments such as the Mainframe, Linux and AIX. Scalability and resilience of those systems is a key part of why they are adopted as production platforms, and COBOL applications can take advantage of such environments to provide the performance, scalability and resiliency that system needs."
