The CDP Is Dead, Long Live the Agentic Data Platform: How Databricks and Braze Are Rewriting Customer Data Infrastructure

The standalone Customer Data Platform, a category that generated billions in venture funding and spawned dozens of point solutions, is being absorbed into something larger. In June 2026, two announcements from opposite ends of the data stack revealed the same conclusion: the future of customer data infrastructure is not a dedicated CDP product. It is an agentic layer embedded directly in the systems where data already lives and engagement already happens.

Databricks Enters the Marketing Stack

On June 16, 2026, Databricks announced CustomerLake at its Data + AI Summit. The product is an agentic CDP built natively on the Databricks lakehouse, bringing identity resolution, audience building, campaign automation, and activation directly into the platform where enterprise customer data, AI models, and governance already reside.

“Marketers need to reimagine their entire foundation,” said Ali Ghodsi, Co-founder and CEO of Databricks. “With CustomerLake, customer data, AI models, and agents live in one governed platform.”

The architectural significance is substantial. Traditional CDPs required enterprises to copy data from warehouses into yet another system, creating synchronization challenges, governance gaps, and latency that undermined real-time personalization. CustomerLake eliminates that copy by embedding CDP capabilities where the data already sits.

CustomerLake introduces what Databricks calls “infinity campaigns,” continuous agent-driven engagement loops that replace one-off batch campaigns. Profile agents transform raw customer data into unified records. Campaign agents handle audience building and personalization. Agentic identity resolution combines deterministic rules with AI-powered probabilistic matching.

HP, Circle K, AB InBev, and Getnet by Santander are among the early adopters in private preview. The partnership ecosystem includes Adobe, Meta, Braze, Bloomreach, Iterable, LiveRamp, The Trade Desk, and 16 other activation and data partners.

Braze Extends From Engagement Layer to Data Platform

The same day, June 16, 2026, Braze announced a major expansion of its Data Platform. The update deepens Braze’s positioning as more than a messaging tool: it is now a composable data platform purpose-built for customer engagement, with the data warehouse serving as the source of truth and Braze operating as the engagement layer directly on top.

Key capabilities include a redesigned Cloud Data Ingestion interface that separates source and sync configuration, a CDI SQL Editor for querying warehouse tables directly, and zero-copy Canvas Triggers that let marketers launch experiences using warehouse data without moving it into Braze first. In 2025, Braze processed 16.6 billion zero-copy updates and streamed 13 trillion events through its Currents data export tool.

The enhanced Databricks partnership creates a bi-directional data loop: CustomerLake handles identity resolution and audience intelligence while Braze handles engagement execution, with signals flowing both ways. Engagement data (clicks, opens, conversions) flows back to the warehouse for ML model training, while enriched segments from warehouse models flow into Braze for activation.

NAVER WEBTOON’s marketing team reported saving 640 work hours monthly using Braze’s data platform, syncing up to 5 million users in just 10 minutes with a single click.

Why This Matters: The Consolidation Pattern

The standalone CDP market has been consolidating for months. ActionIQ was acquired by Uniphore. mParticle was acquired by Rokt for 300 million dollars in January 2025. Segment remains embedded within Twilio’s broader platform. The independent CDP as a category is contracting.

What is expanding is the concept of CDP capabilities distributed across the data stack. Databricks approaches this from the infrastructure layer upward. Braze approaches it from the engagement layer downward. Both arrive at the same architectural principle: zero-copy data access, agent-driven orchestration, and governance embedded at the platform level rather than bolted on.

The Agentic Difference

The word “agentic” is doing real work in these announcements, not serving as marketing polish. In a traditional CDP, a marketer defines a segment, schedules a campaign, and evaluates results days later. In an agentic data platform, AI agents continuously monitor customer signals, build and adjust audiences in real time, trigger personalized experiences without batch scheduling, and feed engagement outcomes back into the models that drive future decisions.

Databricks’ infinity campaigns embody this: rather than discrete send-and-measure cycles, marketing becomes a continuous loop where agents respond to customer context as it changes. Braze’s Account Objects (planned for early fiscal 2027) extend this to hierarchical relationships, enabling account-level personalization for families, households, and enterprise buying groups.

What Enterprise Marketers Should Evaluate

For organizations currently running a standalone CDP, the strategic question is whether that product will remain viable as data platforms (Databricks, Snowflake, BigQuery) embed the same capabilities natively. The operational question is whether existing workflows can transition from batch-oriented, human-driven campaign management to continuous, agent-driven orchestration.

The answer depends on data maturity. Organizations with clean, governed data in a modern warehouse are positioned to adopt lakehouse-native CDP capabilities quickly. Those still wrestling with identity resolution across fragmented systems may find that a transitional period with dedicated CDP tooling remains necessary, even as the long-term direction is clear.

Braze’s supported data warehouse integrations (Databricks, Snowflake, Google BigQuery, and AWS Redshift) reflect where enterprise data gravity has shifted. Rather than asking marketers to move data into a CDP, the platform meets data where it already resides and provides the engagement execution layer on top. User Profile Streaming, a new capability announced alongside the platform updates, enables Braze to stream user attributes and custom data to external destinations, creating a true bi-directional data architecture rather than a one-way ingestion model.

The CDP is not disappearing. Its capabilities are becoming infrastructure expectations rather than standalone product features. The platforms that win will be those that embed customer intelligence directly where data lives and where engagement happens, governed by agents rather than managed by campaigns. For marketing leaders evaluating their data architecture strategy, the decision is no longer which CDP to buy. It is which data platform already holds your governed customer data and which engagement layers can operate on it natively.

Related: the agentic CDP arriving on two fronts | the standalone CDP disappearing