The new IBM z17 truly is a modern mainframe designed for the AI age. It's packed with exciting functionality built to drive business value for enterprises across industries with more than 250 AI use cases including, mitigating loan risk, managing chatbot services, supporting medical image analysis and impeding retail crime, among others.
Here are three standout ways in which the new mainframe maximises the potential of AI on IBM Z.
1. On platform AI-inferencing at scale
The IBM z17's predecessor, the z16 mainframe, was lauded for introducing the Telum processor, IBM's first to have an on-chip AI accelerator designed for real-time AI inferencing at scale. It was responsible for the z16's widely publicised ability to detect payment fraud for up to 10,000 to 50,000 transactions per second in real-time.
To put this in context: the IBM z16 could use anti-fraud technology with in-transaction AI inferencing to identify fraud before payments are authorised - to stop the fraud before it happens.
Now the z17 brings in the next generation Telum II processor with an on-chip AI accelerator that delivers more compute power for in-transaction inferencing. Plus, it optionally has a separate PCIe card AI accelerator called Spyre that complements the Telum II to add additional AI inferencing capabilities - and bring generative AI capabilities to the mainframe., including running AI assistants and using enterprise data contained in the system. In fact, even without the Spyre AI accelerator the z17 has the ability to process 50 per cent more AI inference operations per day.
Again, the quickest way to appreciate why this is so important is to consider it in the context of banking, cards and payment fraud prevention and detection.
As research company Celent describes in its recent report, fraudsters are now using advanced AI and automation tools to commit increasingly sophisticated fraud at a bigger scale than ever before. But with IBM z17 and the Telum II and Spyre, financial institutions can now run bigger, more complex AI models and even run multiple models to identify and prevent fraud in-transactions at high volume with greater precision.
Being able to run the AI inferencing on z/OS, using MLz native on z/OS, or other AI models running in zCX or Linux on IBM Z, virtually eliminates network latency allowing AI infusion to transactional workloads in real time, it also eliminates the need to copy data off the platform to implement AI, allowing customers to keep their valuable or sensitive data within the secure confines of the mainframe.
The superior AI inferencing capability of Telum II and Spyre Accelerator also opens the door to new use cases involving multiple AI models, including traditional machine learning models and large language models (LLMs), to enhance the performance and accuracy of AI predictions compared to relying on a single model.
2. Using AI to enhance developer and IT operations productivity
The ability to run Generative AI and LLMs on the z17 means customers are going to be able to deploy tools such as IBM watsonx Code Assistant for Z and watsonx Assistant for Z on-premises to enhance developer and IT operations productivity.
Christian Jacobi, IBM Fellow and CTO of IBM Systems Development, has suggested that mainframe shops will likely want to use generative AI for tasks such as code assistance and general systems administration.
Many of them run massive applications extending to 100s of millions of lines of code, so applying generative AI through IBM watsonx Code Assistant for Z on-premises will help them understand, refactor, validate, transform, explain and optimize that code. By streamlining code analysis, the code assistant can help organizations ensure that applications remain robust and adapt to evolving business needs.
This will be particularly valuable for addressing the challenges posed by the complexity of legacy systems as seasoned developers retire, leaving gaps in expertise and documentation.
Similarly, watsonx Assistant for Z, a conversational AI assistant, is designed to enhance the IBM Z user experience for new and experienced users while driving up productivity and reducing the time it takes to onboard users who are new to Z.
Not only can this assistant give users accurate, up-to-date answers to their IBM Z queries, but it also provides "automations" to allow mainframers at all levels of experience to execute complex or repetitive tasks to improve productivity. These automations are created by codifying the knowledge or Z experts.
Customers can also customize watsonx Assistant for Z by ingesting documentation related to their own configuration for a personalized and closely tailored experience.
3. AI powered security
Security has always been a paramount concern for mainframe customers. A recent report, "Mainframe as Mainstays of Digital Transformation", reveals that 82 per cent of the customers think it's 'very important' or 'extremely important' that the mainframe supports AI capabilities for monitoring, analyzing, detecting and responding to cyber threats.
So, it's not a surprise that IBM z17 has been designed to use AI in innovative ways to make a difference in this area.
For example, Sensitive Data Tagging for IBM z/OS - a new solution that's going to be available for the z17 - relies on AI with natural language processing (NLP) to distinguish between sensitive and non-sensitive data.
In other words, it can help organizations overcome the labour-intensive and error-prone process of manual data classification. They can benefit from automated tagging and classification using AI-driven technology and ensure that any "crown jewels" data is encrypted and protected appropriately.
Another AI-powered security feature, IBM Threat Detection for z/OS, runs routine scans and uses AI to identify potential threats or malicious anomalies that might be the result of a cyberattack. Customers can view a dashboard to help them understand and diagnose the issue allowing them to detect and take action to prevent (or limit) the potential damage.
This tool is also designed to help organisations identify and issue alerts for "anomalous ICT-related incidents" in line with the requirements in regulations like the EU Digital Operational Resilience Act for financial institutions.
Conclusion: A transformational AI platform to drive business value
With its next-generation Telum II processor, on-chip AI accelerator and PCIe attached Spyre AI accelerator with the potential to fun AI-powered assistants and security tools, the IBM z17 really is a modern mainframe for the AI age. For enterprises that are facing growing fraud risks, IBM Z talent shortages and strict security and regulatory demands or those with their own unique AI-specific use cases, the z17 can be a transformational AI platform designed to drive business value today and well into the future.