From Model Obsession To Data-First Enterprise AI
Enterprise AI spending is the redirection of corporate technology budgets from isolated experiments toward integrated systems that combine AI models, data platforms, and operational applications to deliver repeatable business outcomes. Databricks’ recent funding round has become a clear signal that this spending is shifting from headline-grabbing models to the quieter layer of data infrastructure. The company raised more than USD 7 billion (approx. RM32.2 billion), including USD 5 billion (approx. RM23 billion) in equity, at a USD 134 billion (approx. RM616.4 billion) valuation, showing how much investors value the data foundations behind AI. While OpenAI and other model builders dominate public attention, the competitive pressure in the enterprise market is now centered on who can best prepare, govern, and serve corporate data for AI model deployment. This marks a move from experimentation toward production-scale AI grounded in reliable enterprise data platforms.
Databricks, Snowflake And The New Data Platform Battleground
Databricks and Snowflake are fighting for control of the layer where enterprise data becomes usable for analytics, automation, and agentic AI. Databricks began with Apache Spark to handle unstructured data and then expanded into structured data with its Data Lakehouse, a single platform for multiple data types. A Microsoft agreement to bundle Databricks with Azure helped it reach a USD 100 million (approx. RM460 million) annual revenue run rate by 2018, and by February the company said its run rate had grown to USD 5.4 billion (approx. RM24.8 billion), with fourth-quarter revenue up more than 65% year over year. According to Tomasz Tunguz of Theory Ventures, “It’s exceptional that Databricks’ growth has been accelerating as it scales.” Snowflake, once dominant in structured cloud data, is racing to catch up on unstructured data and AI tools, while still growing about 30% annually, as enterprises align data infrastructure investment with AI ambitions.
Why Data Infrastructure Now Matters As Much As AI Models
Enterprises are discovering that without the right data architecture, even the most advanced AI models deliver limited value. Most organizational information is unstructured—text, documents, messages, images—and this is exactly the material AI systems need. Databricks’ and Snowflake’s evolution shows that enterprise data platforms must now support both structured ERP records and messy unstructured content, while enforcing governance, security, and privacy. For finance, supply chain, HR, customer operations, and industry-specific processes, AI model deployment depends on clean, timely, and well-governed data more than on any single model choice. This is pushing CIOs and ERP leaders to treat data readiness as the unseen AI battleground. Instead of focusing budgets purely on model licenses or training runs, they are directing enterprise AI spending toward platforms that can prepare, catalog, and operationalize data across departments so that AI applications can be embedded into real business workflows.
Acquisitions Show A Shift To Full-Stack Enterprise AI
The acquisition race between Databricks and Snowflake shows how enterprise AI strategies are maturing from proof-of-concept models to full-stack platforms. Databricks has bought MosaicML for about USD 1.3 billion (approx. RM6 billion) to bring in generative AI training and customization, along with Arcion, Neon, Mooncake Labs, and Tabular, the latter reportedly costing more than USD 1 billion (approx. RM4.6 billion). Snowflake, for its part, acquired Neeva for roughly USD 185 million (approx. RM851 million) and brought in its CEO, Sridhar Ramaswamy, to accelerate AI capabilities. These deals are not only about storage or querying; they expand model customization, transaction-level connectivity, and governance. For ERP vendors and systems integrators, this creates a new dependency: application data and process logic must align with whichever enterprise data platforms customers select. As a result, data infrastructure investment is becoming the backbone of sustainable AI model deployment, not a supporting afterthought.
Agentic AI Pushes Enterprises Toward Production-Grade Data Platforms
The next wave of enterprise AI is agentic AI, in which autonomous agents perform complex tasks across business systems. These agents cannot operate on disconnected or poorly governed data; they require live operational context, consistent access to structured and unstructured information, and safe connections into core processes. Databricks’ CTO of Neural Networks, Hanlin Tang, notes that companies are already gaining real value from agentic AI, a statement he says he could not have made a year earlier. Albertsons, for example, uses Databricks-enabled AI systems to optimize promotional spending, showing what is possible when data infrastructure and AI are tightly integrated. As more enterprises aim for similar results, they are shifting enterprise AI spending away from isolated model pilots toward enterprise data platforms designed for production-scale AI model deployment. This is repositioning data infrastructure from a back-office concern to a central pillar of competitive strategy.
