Starburst becomes the first open hybrid lakehouse platform optimized for NVIDIA's new inference compute platform, enabling customers to run real-time AI and analytics on governed, federated data at unprecedented speed

Starburst, a leader in data and AI platforms, today announced optimizations for NVIDIA Vera CPU, unveiled at NVIDIA GTC. Starburst customers will gain access to breakthrough query performance, lower-latency AI inference, and significant cost efficiencies from day one of Vera availability later in 2026.

Starburst's Trino-powered platform is uniquely positioned to capitalize on Vera, NVIDIA’s next-generation data center CPU designed for fast agentic reasoning and data analytics. While competing platforms require data to be centralized into proprietary warehouses before it can power AI, Starburst delivers hybrid, federated, and governed data access at inference speed directly where the data lives, across lakes, warehouses, and operational systems, without movement or duplication.

As enterprise demand shifts from model training to production inference powering agentic AI, retrieval-augmented generation (RAG), and real-time applications, NVIDIA Vera addresses the performance and cost constraints that have limited AI adoption at scale.

“The future of enterprise AI depends on fast access to governed data,” said Justin Borgman, Founder and CEO of Starburst. “With NVIDIA Vera, Starburst aims to bring real-time analytics and inference directly to where that data lives.”

Benefits for Starburst Customers

Early benchmark testing shows Starburst on NVIDIA Vera CPU delivering materially faster query performance and significantly higher CPU efficiency than comparable traditional x86 CPU–based configurations, while maintaining predictable performance for mixed BI and AI inference workloads. For customers, this translates into four core advantages:

• Faster time to AI value. Live, governed enterprise data feeds inference workloads, RAG pipelines, and agentic AI in real time, eliminating the ETL bottleneck that delays AI projects.
• Lower total cost of ownership. Vera's efficient architecture, paired with Starburst's federated model, which avoids costly data duplication, compounds savings across compute and storage.
• Predictable performance at scale. Deterministic throughput for concurrent analytics and AI workloads, even across complex, multi-source datasets.
• Built-in enterprise governance. Fine-grained access controls and policy enforcement travel with every query, ensuring AI applications consume only authorized, compliant data as organizations scale into regulated environments.

“The future of enterprise AI relies on securing instant insights from data that is often distributed across complex hybrid environments,” said Dion Harris, senior director, HPC, Cloud and AI Infrastructure, NVIDIA. “Our work with Starburst, optimizing their Trino-powered platform for NVIDIA Vera CPU, will deliver a foundational solution for real-time, federated data processing.”

“With Trino running on NVIDIA Vera CPU, we’re unlocking a new level of performance and efficiency for federated data access, enabling customers to run analytics and query data directly where it lives - the result is faster insights, lower infrastructure overhead, and a platform built to scale with growing enterprise needs,“ said Jitender Aswani SVP of Engineering, Starburst.

Competitive Differentiation

Starburst creates a position that neither proprietary cloud warehouses nor legacy platforms can replicate. Closed architectures require enterprises to copy data into a single system before AI can use it, adding cost, latency, and governance gaps. Legacy Hadoop and Spark ecosystems lack a native inference path entirely. Starburst eliminates both limitations with an open, hybrid, and federated architecture built on Trino, now optimized for the most advanced inference compute available.

Ecosystem Integration

Starburst's optimization extends across validated enterprise architectures, including the Dell AI Factory with NVIDIA and the Dell AI Data Platform, where Starburst serves as the analytics, data access, and governance engine. Customers deploying Vera within these configurations gain a fully integrated, inference-ready stack from infrastructure to insight.

Building on GPU-Accelerated Trino

In addition to support for Vera CPU, Starburst is developing GPU acceleration for Trino using NVIDIA CUDA, and NVIDIA cuDF for structured data. This future capability will bring large-scale parallelism and accelerated columnar processing to federated analytics and AI inference, powering the next generation of retrieval-augmented and agentic AI workloads.

The initiative reflects Starburst’s vision to unify compute across CPUs and GPUs within a single open engine–enabling enterprises to harness governed, high-performance data processing at the speed of innovation.

Availability

Starburst is exploring how NVIDIA Vera CPU architecture can benefit Trino-powered analytics and AI workloads over time. As Vera CPU moves toward general availability, Starburst intends to evaluate pre-production platforms, gather performance insights, and share guidance on how customers can take advantage of Vera’s efficiency gains once systems are broadly accessible.

Starburst plans to validate and publish updated reference architectures and best practices for customers interested in leveraging Vera for governed, federated analytics and AI.

About Starburst

Starburst is a leader in federated data and the foundation for enterprise AI, built for flexibility, delivering fast, secure access to all your data, wherever it lives. Whether on-premises, across clouds, or in hybrid environments, Starburst provides choice and control to your architecture. Built on an open data stack with Trino and Apache Iceberg, it unifies distributed data without complex or costly migrations, unleashing the full power of the data lakehouse for analytics and AI.

With our Lakeside AI architecture, enterprises gain federated access, governed collaboration, and full data lineage, laying the foundation for scalable, compliant AI innovation. Starburst empowers data-intensive and security-conscious organizations to unlock the full potential of their data while ensuring performance, governance, and control.

Enterprises in 60+ countries, including Comcast, Citigroup, and 4 of the top 5 global banks, rely on Starburst to power analytics and AI initiatives while maintaining flexibility and control. Our strategic partnerships with AWS, Dell Technologies, NetApp, and top cloud providers ensure seamless interoperability across environments.

From insights to action to AI, Starburst fuels innovation at every level. For more information, visit the AIDA, the AI Data Agent page. Explore everything Starburst at starburst.ai.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260316162054/en/