Sanjeev Mohan, SanjMo | IBM Think 2024
summary
Predicts a complete rewrite of the data stack within two years with unified storage for structured and unstructured data, a semantic metadata layer, and disaggregated multi‑engine compute (Starburst, Trino, Dremio, Pandas, Spark, Ray, LLMs) using common table formats like Iceberg and Delta so compute becomes a commodity. Hyperscalers concentrate power while IBM positions as a super cloud glue via Red Hat OpenShift, Ansible, Terraform and HashiCorp. Claims LLMs can handle roughly 80% of COBOL migration and that agents will automate DevOps, testing and integration.
AI-written summary · about
