How IBM z/Architecture is pulling an entire generation of developers into the world's most resilient computing platform - with the tools they already love.

There is a quiet revolution happening inside some of the largest data centers on the planet. It does not announce itself with flashy product launches or venture capital press releases. It hums along at 99.999% availability, processing trillions of transactions each year while the world debates which JavaScript framework to adopt next.

The IBM Z mainframe - once dismissed as a relic of a pre-internet era - is having what might be its most important decade. Not because it changed what it is, but because it changed how it talks to the people who build on it.

A Platform That Has Earned Its Keep

The z/Architecture lineage traces back to the IBM System/360, announced on April 7, 1964 - a design so forward-thinking that modern IBM Z machines remain binary compatible with software written over sixty years ago. That is not a footnote; it is an engineering miracle. As Vincent Alloo, head of technology at software provider Zetaly, noted on the platform's 60th anniversary: "A program which ran 60 years ago, can run today without changing anything." [1]

Today, IBM Z handles roughly 70% of the world's transactional data by volume, including a lion's share of the more than 150 million worldwide credit card transactions banks handle each day. [1] When you swipe a card, check a flight, or transfer money between banks, there is an overwhelming probability that a mainframe touched that transaction. The platform's legendary reliability - five-nines uptime targeting less than 5.3 minutes of annual unplanned downtime, hardware-level encryption, and fault-tolerant design - makes it the backbone of global financial infrastructure, healthcare systems, and government operations. [2]

The "Z" in IBM Z stands for zero downtime, a design philosophy embodied in spare components capable of hot failovers and an operating system that supports dynamic changes without interruption. [2] The platform currently serves two-thirds of the Fortune 100, 45 of the world's top 50 banks, 8 of the top 10 insurers, and 7 of the top 10 global retailers. [3]


The Knowledge Transfer Crisis - and the Answer Hiding in Plain Sight

For years, the enterprise computing world has watched a slow-motion demographic cliff. The COBOL programmers, the JCL experts, the systems programmers who could tune a z/OS system with the confidence of a concert musician - they are retiring. Mainframe operation skills gaps have been a concern for over two decades, as seasoned experts reach retirement age. [1] The pipeline of replacements has been thin, because most computer science curricula do not mention mainframes at all.

"The mainframe was never the problem. The problem was that the mainframe spoke a different language than the generation being asked to run it."

IBM's answer was not to simplify the platform or strip away its power. It was to build bridges. Specifically, to bring Python, YAML, and Ansible natively onto z/OS - so that a developer who has never touched a 3270 terminal could still automate, orchestrate, and extend one of the most capable computing environments ever built.


Python on z/OS: Not a Port, a First-Class Citizen

When IBM introduced the Open Enterprise SDK for Python - now available as a no-charge license component bundled with z/OS 3.1 as of July 2024 - the goal was not a thin compatibility shim. [4] The runtime is built with awareness of z/OS's unique memory model and EBCDIC character encoding. Using Python's standard codecs package, developers can produce and consume data in EBCDIC, UTF, and several other encodings. [5] The SDK supports zIIP specialty processors to offload Python workloads that would otherwise consume general-purpose CPU cycles, reducing cost and freeing capacity for critical business applications. [5]

IBM also provides IBM Z Open Automation Utilities (ZOAU), a Python library that surfaces native z/OS capabilities - dataset management, JCL submission, operator commands - through clean, idiomatic Python APIs. A developer who has spent years writing data pipelines in Python can look at the following example and immediately understand it:

// Python · z/OS dataset access via zoautil_py

from zoautil_py import datasets

# Read a partitioned data set member
content = datasets.read("HLQ.PAYROLL.DATA(MONTHLY)")

for record in content.splitlines():
    process_record(record)

# Write back, z/OS handles the rest
datasets.write("HLQ.PAYROLL.AUDIT", output_lines)

The Python AI Toolkit for IBM z/OS further extends this, providing pre-built AI and machine learning packages - drawing on PyPI's thousands of community modules - enabling data scientists to run analytical workloads colocated with the enterprise data itself, eliminating the latency and security risks of offloading sensitive records. [5] IBM announced Python 3.14 support for z/OS in November 2025, demonstrating active parity with the open-source community release cadence. [6]

Ansible: The Hybrid Cloud Handshake

Perhaps the most strategically significant piece of this puzzle is Ansible. The Red Hat Ansible Certified Content for IBM Z - developed jointly by IBM and Red Hat following the 2019 acquisition - enables teams to write a single playbook that provisions a cloud VM, configures a Linux container, and then manages a z/OS LPAR, treating all three as peers in a unified automation pipeline. [7]

The ibm_zos_core collection, the flagship of the IBM Z Ansible portfolio, supports managing batch jobs, performing program authorizations, running operator operations, executing JES and MVS commands, data set creation, copying, fetching, and encoding, archiving, and volume management. [8] Organizations are applying this across real production scenarios: SSL certificate renewal, CI/CD pipeline integration for z/OS application deployment, z/OS middleware provisioning, and automated security configuration. [9]

// YAML · Ansible playbook excerpt for z/OS

- name: Submit a JCL job and wait for completion
  hosts: zos_host
  collections:
    - ibm.ibm_zos_core
  tasks:
    - name: Run the monthly reconciliation job
      zos_job_submit:
        src: HLQ.JOBS(RECON)
        location: DATA_SET
        wait_time_s: 120
      register: job_output

    - name: Print return code
      debug:
        msg: "RC={{ job_output.jobs[0].ret_code.code }}"

The IBM z/OS core collection 2.0, released in 2025, further advances this mission by positioning IBM Z as a governed, standardized endpoint within enterprise automation - applying the same automation governance, auditability, and change control practices to the mainframe as to the rest of the distributed cloud and hybrid estate. This eliminates the need for parallel or outdated automation processes specific to the mainframe. [9]

A DevOps engineer from any background can read this YAML and understand it. The fact that it is orchestrating a mainframe job stream is almost an implementation detail - and that is entirely the point. The cognitive wall between "cloud" and "mainframe" begins to dissolve when the tooling is the same.


A Generation Inheriting Something Worth Inheriting

There is something quietly significant about this moment in enterprise computing. The architects who designed System/360 in 1964, betting IBM's survival on a 5-billion-dollar gamble, could not have imagined Python or Ansible. Yet the decisions they made - around reliability, security, scalability, and backward compatibility - created a platform robust enough to absorb these modern tools without losing any of its character.

For the younger engineers encountering IBM Z for the first time through a Jupyter notebook or a Git-managed Ansible repository, they are not being handed a burden. They are being given access to infrastructure that processes more critical work in a day than most systems do in a year. The learning curve toward JCL, RACF, and SMF records is real - but the on-ramp has never been gentler.

A recent Forrester State of Mainframes report found that nearly two-thirds of hardware decision-makers leaned on mainframe compute, with enterprise mainframe budgets growing in line with other infrastructure spending. Mainframe usage is growing particularly in business services, transportation, construction, and manufacturing - well beyond its traditional financial services stronghold. [1] Meanwhile, 75% of more than 2,500 global IT executives surveyed by the IBM Institute for Business Value rated mainframes equal to or better than cloud computing in total cost of ownership. [10]

The Quiet Confidence of a Platform That Has Nothing to Prove

What is most striking about the IBM Z evolution is its lack of defensiveness. The platform does not need to compete with Kubernetes on Kubernetes's terms. It does not need to win a benchmark war with commodity x86 clusters. It knows what it is excellent at - and it has simply decided to be more accessible while staying excellent at it.

That is a harder thing to do than it sounds. It requires respecting the existing community of COBOL and assembler experts who built careers on z/OS, while simultaneously opening the door to developers who learned their craft on GitHub and Stack Overflow. The IBM z17, announced in April 2025 and powered by the Telum II processor with 43 billion transistors, now integrates on-chip AI acceleration for real-time transactional inferencing - proof that the platform continues to absorb the best of contemporary computing without discarding what makes it irreplaceable. [2]

The mainframe is not a museum piece waiting to be decommissioned. It is an active, evolving ecosystem - one that is quietly making sure the next generation of engineers can walk in, open a terminal, type python3, and feel at home.

References

[1]  CIO Dive. (2024, April 12). As mainframes turn 60, skill gaps threaten the enterprise workhorse. https://www.ciodive.com/news/ibm-mainframe-sixty-years-anniversary-cloud-skills/712325/

[2]  Grokipedia. (2026). IBM Z. https://grokipedia.com/page/IBM_Z

[3]  TechChannel. (2025, January 17). How IBM z/Architecture Has Evolved to Address Client Needs. https://techchannel.com/ibm-z-mainframe/zarchitecture-evolution-for-client-needs/

[4]  Planet Mainframe. (2024, December 4). IBM Open Enterprise SDK for Python 3.13 and z/OS 3.1 integration. https://planetmainframe.com/2024/12/gigaom-radar-report-recognitions-ibm-open-enterprise-sdk-for-python-3-13-and-more/

[5]  IBM. (2024). IBM Open Enterprise SDK for Python. https://www.ibm.com/products/open-enterprise-python-zos

[6]  IBM Community. (2025, November 11). Announcing IBM Open Enterprise SDK for Python 3.14 for z/OS. https://community.ibm.com/community/user/blogs/jc-yao/2025/11/10/announcing-ibm-open-enterprise-sdk-for-python-314

[7]  IBM. (2024). Red Hat Ansible Certified Content for IBM Z. https://www.ibm.com/support/z-content-solutions/ansible/

[8]  GitHub - ansible-collections/ibm_zos_core. Red Hat Ansible Certified Content for IBM Z. https://github.com/ansible-collections/ibm_zos_core

[9]  IBM. (2025). Announcing IBM z/OS Core Collection 2.0 for Red Hat Ansible Automation Platform. https://www.ibm.com/new/announcements/announcing-ibm-z-os-core-collection-2-0-for-red-hat-ansible-automation-platform

[10]  SiliconANGLE. (2025, October 14). System Z reinvented: Why IBM mainframes dominate enterprise IT. https://siliconangle.com/2025/10/14/system-z-reinvented-ibms-mainframe-owns-enterprise-core-ibmzinfrastructure/

[11]  Wikipedia. (2025). z/Architecture. https://en.wikipedia.org/wiki/Z/Architecture

[12]  Wikipedia. (2026). IBM Z. https://en.wikipedia.org/wiki/IBM_Z

[13]  IBM GitHub. Red Hat Ansible Certified Content for IBM Z - Documentation. https://ibm.github.io/z_ansible_collections_doc/

[14]  IBM. (2024). IBM Open Enterprise SDK for Python - Documentation. https://www.ibm.com/docs/en/python-zos/3.11.0