Analyst(s): Brad Shimmin
Publication Date: June 2, 2026
Starburst introduced its Enterprise Intelligence Platform at the AI & Datanova event, delivering native capabilities designed to execute AI directly on distributed, governed data. By introducing the general availability of AIDA, AI-Ready Data Products, and Managed Icehouse, Starburst is targeting the costly requirement of replatforming or moving data into centralized silos in order to support agentic AI workloads. This release highlights a very realistic architectural philosophy that prioritizes data gravity, semantic trust, and open composable ecosystems to accelerate enterprise AI initiatives.
What is Covered in This Article:
- Starburst’s launch of the Enterprise Intelligence Platform, a four-layer architecture designed to unify hybrid data environments without widespread data replication.
- The introduction of Managed Icehouse, featuring Icehouse LakeOps and Ingest to automate the lifecycle management of Apache Iceberg open table formats.
- The rollout of AI-Ready Data Products, including “Data Products as Code,” which allows data teams to govern semantic context using software engineering principles.
- The general availability of AIDA (AI Data Assistant), functioning as a conversational interface and governed agentic control plane.
- The economic implications of localized AI workflows and the strategic integration of GPU-accelerated infrastructure to bypass public cloud API costs.
The News: On May 28, 2026, Starburst unveiled its Enterprise Intelligence Platform, an architecture engineered to support the rigorous demands of enterprise analytics, AI and agentic AI workloads. The platform brings AI directly to the data, mitigating the governance blind spots and latency issues associated with centralized cloud data lakes. The release introduces a four-layer architecture: a flexible federated data foundation, a performant analytics engine, an enterprise context layer built on AI-Ready Data Products, and a trusted agentic interface known as AIDA. Notably, the release includes Managed Icehouse to automate Apache Iceberg maintenance and a new Bring Your Own Cloud (BYOC) deployment model that allows enterprises to maintain full sovereignty over their data and network architectures while leveraging a managed software plane.
Starburst’s Intelligence Platform Sets Out to Tame the Multi-Cloud AI Wilderness
Analyst Take: For the better part of two years, the enterprise technology narrative has been utterly consumed by the capabilities of frontier AI models such as Anthropic Claude, Google Gemini, and OpenAI GPT. The market has collectively operated under the assumption that smarter algorithms, bigger context windows, and multimodality would magically unravel the complexities of enterprise decision-making. That assumption is now colliding with the physical realities of enterprise data.
Scaling artificial intelligence demands a lot more than sophisticated models. It requires securing access to trusted, governed data across highly complex hybrid environments. Organizations are actively embracing architectures designed to reduce data movement while rigidly preserving governance and business context. Starburst’s latest platform announcement reflects this structural progression, delivering a solution focused entirely on distributed data access, embedded intelligence, and enterprise-scale analytics performance.
The strategy makes immense practical sense. For years, the prevailing wisdom encouraged organizations to consolidate data into massive, centralized cloud warehouses and data marts. Today, organizations are adopting composable intelligence based on open standards, most notably Apache Iceberg and Model Context Protocol (MCP), to prevent vendor lock-in. Starburst’s introduction of Managed Icehouse capitalizes directly on this appetite. By automating the tedious background maintenance, compaction, and snapshot expiration required to keep Iceberg tables performant, for example, Starburst can lower the operational threshold for enterprises seeking to build scalable, open foundations.
The Agentic Economics Problem
The necessity of querying data in place becomes starkly apparent when we examine the evolving nature of artificial intelligence. Enterprises are graduating from conversational chatbots to autonomous, read-write agentic AI systems. Rather than waiting for a human prompt, these agents continuously retrieve context, evaluate conditions, and execute multi-step workflows.
This continuous operation consumes both data and compute tokens at a ferocious rate. When organizations run these constant, high-volume agentic reasoning loops against data consolidated in a centralized cloud repository, they trigger severe cost and performance volatility, often incurring massive data egress fees. The physical weight of the data (e.g., data gravity) makes it economically perilous to continuously shuttle information back and forth to an external frontier-scale language model.
This friction has forced a severe reappraisal of enterprise infrastructure. According to the Futurum Research 2026 Key Issues & Predictions report, 71% of CIOs are currently reevaluating their cloud workload placement directly due to AI cost structures and data gravity. But moving the data to the AI is proving to be a financial dead end. Instead, architectures that push the analytical processing down to where the data already resides offer a much more attractive alternative.
Dell and the Economics of Localized Processing
This Data FinOps pressure provides critical context for the deep engineering collaborations happening across the infrastructure landscape. Futurum recently covered one such development in a Futurum report entitled “Is the Cloud Too Expensive for Agentic AI? Dell Bets on Localized Tokens.” In that analysis, we explored how Dell Technologies paired its high-performance hardware with Starburst as the analytical core to bring GPU-accelerated SQL analytics directly to enterprise environments.
This type of partnership sharply improves query performance for data-heavy applications without the unpredictable metered economics of public cloud APIs. By utilizing advanced file readers and autonomous indexing locally, enterprises can offload heavy transformation tasks to GPUs sitting right next to the data.
Surviving the era of agentic AI requires taming the underlying infrastructure costs. Starburst’s federated query engine is a vital piece of that puzzle, allowing organizations to maintain massive datasets securely in place while accelerating the foundational SQL queries that feed retrieval-augmented generation and autonomous agent workflows.
Engineering Trust Through Data Products as Code
Raw data holds minimal value for AI without a proper semantic context. Exposing a language model to undecipherable database columns lacking strict business definitions is a guaranteed recipe for hallucination. To combat this, Starburst is heavily fortifying its enterprise context layer with AI-Ready Data Products.
These data products serve as reusable, trusted assets that package governed data, metadata, and precise business rules. In this way, data products function the same as code, providing a compelling solution where enterprises can manage data alongside code using the same operational practices. By allowing data teams to define and deploy these assets programmatically using declarative YAML files, Starburst is literally bringing data governance into the modern era of software engineering. Applying continuous integration and continuous deployment methodologies to data architectures eliminates the drift that causes AI systems to fail in production. When an agent requests a metric like customer lifetime value, it receives a mathematically consistent, version-controlled answer.
AIDA as the Control Tower
The final piece of Starburst’s puzzle is AIDA, the AI Data Assistant. AIDA serves dual roles, acting as a conversational analytics interface for business users while functioning deeply as an agentic control plane.
Enterprises cannot allow AI models unfettered authority to execute workflows or alter database records without strict, transparent oversight. AIDA is engineered for grounded reasoning, securely tethering its logical pathways to the governed data products housed within the platform. By utilizing the hugely popular MCP standard as an access method, AIDA can fluidly interoperate with external enterprise applications and pull context from unstructured repositories, all while respecting embedded access control policies. It keeps the AI securely constrained, ensuring that compliance remains inherent to the interaction.
Ultimately, with these announcements, Starburst is addressing the fundamental physics of modern data management. Synchronizing petabytes of distributed enterprise data into a single centralized repository consistently introduces latency and burns capital. By leveraging federation, automation, and strict programmatic governance, the company’s Enterprise Intelligence Platform offers a pragmatic pathway to operationalize artificial intelligence without breaking the bank.
What to Watch:
- Hyperscaler Retaliation: Watch how major public cloud providers respond to Starburst’s BYOC deployment model. As platforms like Starburst enable true compute sovereignty and allow enterprises to sidestep native cloud analytics tools, cloud giants may adjust (or absorb) egress pricing or introduce counter-federation services to keep workloads locked within their walled gardens.
- Data Engineer Adoption of YAML Governance: Keep a close eye on whether data products as code gains traction among data professionals. Data products have been on the docket for years, but have proven difficult to master. And it’s not getting any easier. Software engineering teams have long embraced declarative deployments, but data engineering teams must now also rapidly adapt their workflows to utilize these methodologies to keep pace with AI deployment timelines.
- Agentic Ecosystem Integration: Monitor AIDA’s integration capabilities via MCP. The long-term success of this control plane hinges on its ability to connect securely with external enterprise applications and third-party unstructured repositories without fracturing the underlying governance framework.
See the complete press release detailing the launch of the Enterprise Intelligence Platform on the Starburst website.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other Insights From Futurum:
Can Starburst’s AIDA Crack the Enterprise AI Data Access Problem?
Informatica Unbundles the Monolith: Headless Data Management Meets the Agentic Enterprise
Breaking Data Gravity: Google’s Play for a Composable Agentic Ecosystem
Author Information
Brad Shimmin is Vice President and Practice Lead, Data Intelligence, Analytics, & Infrastructure at Futurum. He provides strategic direction and market analysis to help organizations maximize their investments in data and analytics. Currently, Brad is focused on helping companies establish an AI-first data strategy.
With over 30 years of experience in enterprise IT and emerging technologies, Brad is a distinguished thought leader specializing in data, analytics, artificial intelligence, and enterprise software development. Consulting with Fortune 100 vendors, Brad specializes in industry thought leadership, worldwide market analysis, client development, and strategic advisory services.
Brad earned his Bachelor of Arts from Utah State University, where he graduated Magna Cum Laude. Brad lives in Longmeadow, MA, with his beautiful wife and far too many LEGO sets.
