Data Management & Governance

The three-horse race between Snowflake, Databricks, and Microsoft Fabric is intensifying on AI-native capabilities. Snowflake's AWS commitment and Databricks' real-time pipelines preview signal that infrastructure investment — not just feature parity — is now a competitive differentiator. Teams evaluating or re-evaluating data platform contracts should factor in the trajectory of each vendor's AI roadmap alongside cost and performance benchmarks.

The LLM market is bifurcating: frontier labs (Anthropic, Google, OpenAI) are moving up the stack into enterprise services, while infrastructure plays like OpenRouter — backed by essentially every major data platform vendor — consolidate model-routing. Enterprises should plan for a multi-model strategy rather than single-vendor commitment, and evaluate which routing layer they want to sit atop.

Agentic AI is production-deployed but nowhere near mature: even the best models score in the single digits on rigorous domain benchmarks. MCP is rapidly becoming the governance layer of record for agent-to-tool access. Teams deploying agents in regulated workflows should prioritise identity-aware MCP implementations and invest in human-in-the-loop checkpoints while model capabilities catch up.

Catalog vendors are under competitive pressure to demonstrate AI value-add beyond discovery — auto-documentation and active lineage are becoming table stakes. ServiceNow's governance pivot at Knowledge 2026 signals that IT service management players are encroaching on catalog territory from the policy and risk angle. Evaluate whether your catalog investment includes a credible AI-metadata roadmap.

The December 2027 high-risk deadline is a genuine reprieve but not a reason to pause compliance programmes — the runway is shorter than it appears when factored against typical enterprise implementation cycles of 12–18 months. Vendors offering pre-built policy packs (EU AI Act, NIST AI RMF, ISO 42001) are positioned to accelerate time-to-compliance for organisations still mapping their AI inventory.

Observability is expanding beyond structured pipelines into unstructured data and AI agent outputs — a necessary evolution as LLM-generated content enters downstream analytics and decision systems. Forrester's emphasis on heterogeneous ecosystem support aligns with enterprise reality: quality tooling must span cloud data warehouses, data lakes, streaming, and now AI-generated datasets.

Cyera's reported valuation trajectory signals that Data Security Posture Management (DSPM) has graduated from emerging category to essential enterprise capability. Capital concentration at the top of the DSPM market typically precedes either IPO or acquisition activity — watch for Wiz and Palo Alto as potential acquirers, and evaluate DSPM coverage in your own security stack.

The dbt Labs + Fivetran merger, if completed, creates the first integrated ELT-plus-transformation entity at meaningful scale (~$600M combined ARR). This consolidation will pressure standalone ingestion and transformation vendors and may change commercial dynamics for teams currently licensing both tools separately. Monitor for pricing, roadmap, and support changes post-close.

Power BI's Q&A retirement by December 2026 is an actionable deadline: organisations with embedded Q&A workflows need a migration plan to Copilot now. The broader shift from static dashboards to agentic analytics (Tableau Next, Copilot) is compressing the remaining useful life of traditional BI report investments — prioritise AI-native analytics capability in any BI platform evaluation.