Redpanda, creators of the Redpanda Agentic Data Plane (ADP) for enterprise AI, announced the general availability of the industry’s first adaptable data streaming engine — a single, multi-modal platform that allows enterprises to balance performance, safety, and efficiency at the topic level. Available in Redpanda Streaming 26.1, this release eliminates the need for separate, specialized clusters and provides a unified foundation for modern data and AI workloads.

Historically, organizations have been forced into “streaming sprawl” — managing a patchwork of disparate technologies to support different workloads. This often meant maintaining expensive, high-performance clusters for operational applications while standing up separate “diskless” systems for low-cost observability, analytics, or lakehouse ingestion. Redpanda Streaming 26.1 eliminates this tradeoff. By integrating Cloud Topics into its core single-engine architecture, Redpanda supports all workloads in a single environment, where developers can choose the precise behavior of their streaming data workloads within one platform.

“The delivery of the Redpanda One (R1) engine marks a fundamental shift in the streaming category,” said Alex Gallego, founder and CEO of Redpanda. “In completing this vision, we’ve built the world’s first engine that adapts to fit your workload, not the other way around. You no longer have to manage a dozen different environments just to get your data where it needs to go. Whether you are powering low-latency fraud detection alerts or petabyte-scale AI model training, the R1 streaming engine provides one platform, one security model, and zero trade-offs in data integrity, performance, and cost.”

Marketing Technology News: MarTech Interview With Fredrik Skantze, CEO and Co-founder of Funnel

One Engine, Every Workload

Redpanda Streaming 26.1 allows users to configure fit-for-purpose data streams using four distinct features to meet specific application requirements:

• Write Caching: Designed for ultra-low latency, providing an extreme speed boost for performance-critical applications with in-memory caching.
• Tiered Storage: Straddling the cost/speed tradeoff by spanning both local disk and cloud object storage, tiered storage balances performance and cost like a slider, according to precise user preferences.
• Iceberg Topics: A zero-ETL approach that automatically materializes selectable streams into Apache Iceberg™ tables for instant analytics.
• Cloud Topics (now generally available): A cloud-first approach that writes message contents directly to object storage for extreme cost-efficiency at massive scale, delivering over 90% savings in cross-AZ networking fees levied by some cloud providers.

Unlike “diskless” Kafka alternatives that rely entirely on object storage, Redpanda lets users combine these powerful features at will, tailoring each workload in a shared cluster, while retaining unique architectural advantages. In addition to supporting the full suite of Kafka semantics like idempotency, transactions and compaction with infinite retention, Cloud Topics leverages fast local networks and disks for metadata replication and read optimization, enabling faster reads of frequently used data while Raft-based consensus preserves the strict data safety guarantees customers expect from any data managed by the platform. Keeping it all in-cluster also eliminates reliance on an external control plane that moves critical private metadata outside customers’ sovereign cloud VPC, and becomes a single point of failure, the limitations of which were laid bare in several recent full-region outages in some hyperscalers.

Marketing Technology News: The Death of Third-Party Cookies Was Just the Start. Are You Ready for Consent Orchestration?

A Foundation for the Agentic Era

Redpanda Streaming is built for the Redpanda Agentic Data Plane, a new category of infrastructure designed to connect AI agents with live enterprise data, with full governance capabilities for explainability and compliance. The adaptable R1 engine enables AI for both classical offline training and real-time workloads in a single unified platform. Developers can use Iceberg Topics to ingest massive volumes of structured data needed for model training, efficiently retaining months of history in the lakehouse. Simultaneously, high-performance modes like Write Caching can power AI systems and agents requiring real-time access to constantly changing data.

More Innovations in Redpanda Streaming 26.1

Redpanda Streaming 26.1 includes additional platform enhancements:

• Group-Based Access Control (GBAC) to strengthen enterprise security and access management
• General Availability of Redpanda BYOVPC, which allows teams to meet strict security, compliance, and networking requirements while still using a fully managed Redpanda service
• User-controlled scaling for Bring Your Own Cloud (BYOC) deployments
• General availability of “unified identity,” fusing security across Redpanda Console and Redpanda Streaming, and through all layers of Redpanda BYOC’s sovereign-cloud architecture