Challenges and IBM Hardware Strategy to Cater With Clients Needs
summary
Claims that CPU P10/P11 and Z P11 chips embed AI accelerators to run inference at transaction time and cut latency into single-digit milliseconds for hundreds of thousands of events per second. Stack actually uses Onyx runtime and SNAP ML, Cloud pack for data, Spectrum storage and Rocket-built OpenC builds. Hardware ships to clients since 2020–2022 and early deployments report 10–50× acceleration, integration with COBOL/Java workloads and on-prem hybrid setups proving the approach in production.
AI-written summary · about
chapters (8)
- 0:00Enterprise AI at IBM
- 4:35IBM's holistic AI approach
- 11:01Real-time AI transactions
- 15:33Hardware for in-core AI
- 20:13AI software stack on IBM
- 27:36AI deployment strategies
- 33:37IBM storage for AI pipelines
- 35:27Q&A
related talks
25mAI-powered digital transformation
15mKeynote: Fujitsu's Open Source Vision and Initiatives
05mKeynote: The Value of Open Source for the Enterprise
41mHow InnerSource Is Beginning to Change Software Development at IBM. Lessons Learned...
08mIke Nassi (TidalScale) interviewed at O'Reilly Software Architecture Conference NY 2018
31mScaling AI for Industry
12mDigital & Data/AI Strategy and Transformation
24mAI Implementation: Navigating the Enterprise Lifecycle for Technical Leaders