Fivetran’s CPO: Closed Data Stacks Won’t Survive the Agent Era

Why It Matters

Enterprises that cling to proprietary data stacks risk soaring AI‑driven analytics costs and sub‑par insights, while open, interoperable infrastructures can contain spend and improve performance.

Key Takeaways

• Agentic AI can issue 10‑100× more queries than traditional analytics.
• Closed stacks force all queries through costly compute, inflating AI expenses.
• Fivetran’s Open Data Infrastructure benchmark highlights vendor data access constraints.
• Interoperable data lakes and SQLMesh aim to lower AI workload costs.
• Open infrastructure plus semantic discipline unlocks productivity without budget overruns.

Pulse Analysis

The rise of agentic AI is reshaping enterprise analytics by turning a handful of human‑initiated queries into a torrent of automated requests. Each agent can probe data warehouses repeatedly, seeking the most cost‑effective path for a given question. In a closed data stack, every query is forced through the same high‑performance engine, driving up compute spend and creating latency bottlenecks that erode the value of AI‑generated insights.

Fivetran is positioning itself at the forefront of this shift with its Open Data Infrastructure initiative. By publishing a Data Access Benchmark, the company quantifies how different vendors throttle AI workloads, encouraging transparency and competition. Its recent work on Google Cloud’s data lake interoperability and the donation of SQLMesh to the Linux Foundation further lower barriers to moving data across platforms, enabling organizations to route cheap queries to lightweight engines while reserving premium resources for complex analytics. This modular approach aligns with the broader industry trend toward cloud‑native, polyglot data architectures.

For decision‑makers, the message is clear: embracing open, interoperable data layers and enforcing semantic discipline can prevent the "triple whammy" of poor AI answers, runaway costs, and wasted compute. Companies that act now will capture productivity gains from AI agents without inflating budgets, while those that double down on restrictive controls risk being left behind as the agent era accelerates. The strategic choice between lock‑in and openness will define competitive advantage in the next wave of data‑driven innovation.