Enterprise AI Data Infrastructure: The New Strategic Battleground
Enterprise AI data infrastructure refers to the platforms, databases, and governance layers that prepare, store, and deliver company data so AI models and agents can operate reliably at scale and connect to real business workflows. As generative AI pilots move into production, spending is shifting from eye-catching models toward AI data platforms that can handle messy, high-volume, multi-format data. This is where ClickHouse and Databricks now sit: between raw corporate information and the AI systems executives want to deploy. Their recent growth shows that enterprises see data readiness, not model access, as the main bottleneck in AI. In this view, model quality and impact depend directly on the consistency, timeliness, and richness of the underlying data infrastructure, turning data engineering choices into core AI strategy decisions.
ClickHouse Revenue Growth Shows How AI Agents Are Changing Databases
ClickHouse’s tripling of annualised revenue run rate to USD 250 million (approx. RM1,150 million) signals how fast demand is rising for AI-optimized analytical databases. Its columnar, open-source engine is built to scan massive datasets at high speed, which suits AI agents that need fast, repeated access to logs, events, and metrics. More than 4,000 customers, including Anthropic, Meta, and Capital One, now rely on ClickHouse’s managed cloud service for these workloads. The company’s acquisitions also show where enterprise AI data infrastructure is heading. By buying six startups, including Langfuse, which tracks and evaluates AI agent performance, ClickHouse is moving from being a high-speed database to an end-to-end platform for AI-centric analytics. According to The AI Insider, ClickHouse is already planning for an IPO timeframe and has hired a CFO with Snowflake investor-relations experience to prepare for public markets.
Databricks Valuation Highlights the Shift in Enterprise AI Spending
Databricks’ USD 134 billion (approx. RM616 billion) valuation, backed by more than USD 7 billion (approx. RM32 billion) in its latest funding round, shows investors expect AI data platforms to capture a major share of enterprise AI spending. Built on Apache Spark, Databricks started by helping companies structure and analyze raw, often unstructured, data. Its Data Lakehouse now brings structured and unstructured data together on one platform, aligning closely with how AI models consume information. The company said its annual revenue run rate reached USD 5.4 billion (approx. RM25 billion), with fourth-quarter revenue up more than 65% year over year. That growth, unusual at this scale, reflects how enterprises now prioritize making internal data usable for AI over chasing the newest foundation model. The Databricks story suggests that control of data pipelines and governance is becoming more valuable than control of the models themselves.
Competing for the Same AI Data Layer: ClickHouse, Databricks, and Beyond
While they started from different places, ClickHouse and Databricks increasingly compete for the same enterprise AI data infrastructure layer: the environment where data is prepared, stored, governed, and made accessible to AI agents and models. Databricks aims to be the default lakehouse for both structured and unstructured information, while ClickHouse focuses on extremely fast analytical queries over large volumes of events and metrics. Both are racing, alongside Snowflake, to add capabilities through acquisitions, from AI agent observability to model customization and transaction-data streaming. For enterprises, this consolidation means AI decisions are also platform decisions. Choosing a data platform now shapes how new AI agents will connect to ERP, finance, supply chain, and customer systems. As platforms absorb more AI-native features, the line between “database” and “AI runtime” is starting to blur.
Why Data Infrastructure Is Becoming the AI Bottleneck—and the VC Bet
The funding surge into Databricks and the rapid ClickHouse revenue growth point to a shared lesson: enterprises have more access to models than to production-ready data. Most organizations struggle to combine structured ERP data with documents, messages, and logs in a way AI systems can use safely and consistently. This is why data readiness has become the unseen battleground for enterprise AI, and why venture capital now frames data platforms as the bottleneck to unlock further AI spending. In practice, this shifts attention from proof-of-concept chatbots to questions of lineage, governance, observability, and high-speed access. The message to CIOs and CFOs is clear: sustainable AI value will depend less on picking the “best” model and more on building an AI data platform that can support waves of agents, automations, and analytics over time.