Editorial
Guide
Is AI Making Your Organization More Productive or More Dependent?
I recently sat down with Bill Hostmann at Dresner Advisory Services to discuss where organizations should be storing their data, especially their data that powers AI. His answer, at its core, was "it depends" — but the nuance behind his answer should matter to data leaders.
Hostmann suggested that the analytic data infrastructure (ADI) market has moved past a warehouse vs. lake debate. No single storage architecture dominates, but data warehouses remain the most widely preferred option.
Rather than describing a market in transition from one architecture to another, Hostmann said the data shows something more settled: a mix of data architectures in which each architecture — warehouse, lake and lakehouse — serve defined roles, have well defined use case fit and have defined places in an organization's maturity progression.
As a part of this, data virtualization serves as a necessary layer for hybrid or complex environments, particularly when accessing data outside the primary lakehouse structure, though Hostmann stopped short of calling it a top priority for ADI buyers, at least not yet given that many analytic use cases today are prioritizing data gravity (i.e., data in one place) rather than diversity/virtualization of data sources.
AI Maturity Redraws the Architecture Map
What Hostmann said is reshaping the market, however, is AI maturity. As organizations advance from nascent through emerging and intermediate to advanced AI capability, the architecture mix shifts dramatically:
- Warehouse usage declines from 65% to 38%
- Lake adoption surges from 42% to 62%
- Streaming-first architecture grows from 8% to 50%
Put simply, organizations with advanced AI capabilities are increasingly turning to data lakes.
Related Article: Master Data Management: The Quiet Foundation Your AI Strategy Can't Afford to Ignore
Production AI Breaks the Warehouse-Only Model
Production AI demands the kind of data diversity, velocity and volume that warehouse-only infrastructure cannot meet — making AI maturity, not vendor preference, industry or organization size, the primary forcing function for storage architecture decisions.
Framed through this lens, Hostmann sees AI-driven use cases as a primary catalyst pushing customers toward a lakehouse architecture. To be clear, this architecture can handle a mix of structured, governed data and unstructured or streaming that AI use cases often need.
On the competitive side, the ADI market has clear leaders but also meaningful challengers:
- AWS anchors the market at 27%
- Microsoft holds 17%
- Google has 10%
- Databricks and Oracle each hold 9%
The latter two hold anchor shares not on the strength of ecosystem breadth, but through specialized capabilities that have earned them a defined place in the ADI stack.
What Data Leaders Should Do Next
Next, I asked Hostmann for actionable guidance for data leaders. He structured his recommendations around where an organization sits in their architectural journey, providing a set of principles that apply universally.
For Warehouse-Anchored Organizations
For organizations anchored firmly with traditional warehouses, Hostmann’s prescription is to extend before replacing.
He said near term invest priorities should focus on extending existing analytic data warehouses with a semantic layer and a data lake-house / iceberg layer to support data virtualization from new data sources (including streaming and unstructured data) needed for agent-driven AI use cases. Leveraging existing data infrastructures will be the better investment compared to a rip and replace of existing analytic data infrastructure investments.
From there, Hostmann urges investment in governance and cataloging, which he framed not as a housekeeping exercise but as the structural bridge to the architectural flexibility needed as part of an AI maturity roadmap.
For Hybrid Organizations
For organizations in the hybrid middle, Hostmann argued the lakehouse investment case is clear. These organizations are typically running warehouse-native governance requirements and lake-scale ingestion requirements at the same time, and the lakehouse is built to support both.
His practical cautions are two:
- Prioritize platforms with proven, out-of-the-box connectivity to your existing anchor.
- Do not defer semantic layer and observability investment to a later time.
These are the components that make lakehouse infrastructure accessible to business users and monitorable for data quality; treating them as optional add-ons is a common and costly mistake.
For Lake-Native Organizations
For organizations in the lake-native tier, your challenge is governance. It represents your primary operational risk, and your destination should be a composable, modular architecture designed for flexibility at scale.
For Everyone: 3 Principles to Follow
For all organizations, Hostmann suggests three standing principles.
- Align storage infrastructure investment with your AI maturity and roadmap.
- Enforce SQL portability across every storage component.
- Treat the three-to-five-year ADI implementation window as an investment discipline, not a timeline to compress.
The organizations reporting the highest success with their business intelligence initiatives are not those that pursued the most architecturally ambitious roadmaps. Instead, they have made phased, sustained infrastructure investments over time. Simply put, this allowed them to industrialize the data they need for BI and AI success.
Related Article: From Dashboards to Agents: The Future of Self-Service BI
Let AI Maturity Drive the Decision
The warehouse vs. lake debate asks the wrong question. Hostmann's research shows it was never about which architecture wins — it was about which one fits your organization's stage of maturity.
What makes this moment different from prior infrastructure cycles is the forcing function. Past shifts in storage architecture were driven by cost, vendor preference or ecosystem consolidation. This one is driven by what production AI actually demands.
The strategic implication for data leaders is clear, even if execution isn't. Organizations pulling ahead have aligned infrastructure investment to where their AI capabilities are heading, built governance and observability in from the start and refused to let vendor preference substitute for strategic clarity.
Why Some Dealers Are Pulling Ahead With AI
Register
Content Leaders Collective: Proving Content's Business Impact Starts With the Right CCMS
Register
From Automation to Intelligence: How Leading Teams Are Rethinking Operations
Register
Ai4 2026
Register
The lakehouse isn't a silver bullet. But AI maturity is the clearest signal available for where your storage infrastructure needs to go and the organizations that read it early will have the flexibility to act when it matters.
Learn how you can join our contributor community.