SAP’s planned acquisitions of Prior Labs and Dremio highlight a major shift in enterprise AI. The next phase may not be defined by chatbots alone, but by structured-data intelligence, tabular foundation models, unified data platforms, and agentic workflows built directly into business systems.
Artificial intelligence has entered a new phase. The first wave was dominated by chatbots, and large language models that could write, summarize, translate, and answer questions with impressive fluency. That phase made AI visible to the public. But inside the enterprise world, the deeper transformation is happening somewhere less glamorous and far more valuable: structured business data.
Enterprises do not run only on text. They run on financial ledgers, procurement tables, customer records, supply chain movements, pricing data, inventory systems, HR databases, sales forecasts, invoices, and compliance logs. These are not casual documents sitting in folders. They are the operating memory of modern companies.
That is why SAP’s recent AI acquisition strategy matters. Its planned acquisition of Prior Labs, a German startup focused on tabular foundation models, and its announced intent to acquire Dremio, a U.S.-based data lakehouse platform, reveal a serious shift in enterprise AI. SAP is not merely adding another chatbot to its software suite. It is moving toward a deeper AI stack where structured data, governed data access, and agentic workflows come together.
The larger message is clear: the future of enterprise AI will not be won by the company with the loudest chatbot demo. It will be won by the company that can convert trusted business data into reliable intelligence.
Why Enterprise AI Needs More Than General-Purpose LLMs
Large language models have changed how people interact with software. They are powerful at understanding language, generating text, extracting meaning, and assisting with knowledge work. Yet business systems have a different challenge.
Most enterprise data is not stored as neat paragraphs. It lives in rows, columns, relationships, schemas, metrics, timestamps, and transaction histories. A procurement table does not behave like a blog post. A financial ledger cannot be treated like a casual document. A manufacturing forecast contains numerical patterns, missing values, categorical variables, and time-dependent relationships that normal text-first AI systems often struggle to interpret correctly.
This is the structured data challenge. Traditional LLMs can describe a spreadsheet if the data is converted into text. But that conversion often loses precision. Important relationships may be flattened. Numerical signals can become vague. Column-level meaning may be diluted. For enterprise AI, that is a serious limitation because business decisions require accuracy, context, traceability, and governance.
A CFO of the company needs a grounded explanation of what changed, where it changed, why it changed, and what may happen next, based on the financial data. A supply chain manager does not need generic suggestions. He needs early warnings based on inventory flows, vendor performance, logistics delays, and demand patterns. This is where SAP’s interest in tabular AI becomes strategically important.
Prior Labs and the Rise of Tabular Foundation Models
Prior Labs is focused on tabular foundation models, often called TFMs. These models are designed specifically to work with structured data. Instead of treating tables as awkward text inputs, tabular foundation models treat them as first-class data structures. This is a meaningful technical distinction. A tabular foundation model can learn patterns across rows, columns, categories, missing values, distributions, and relationships. It can support prediction, classification, forecasting, anomaly detection, and even synthetic data generation. In simple terms, it gives AI a stronger way to reason over the kind of data that enterprises actually depend on.
SAP had already been exploring this direction through its own work on SAP-RPT-1. By bringing Prior Labs into its ecosystem, SAP is strengthening its ability to build AI systems that understand enterprise data at a much deeper level. The strategic value is not only in the model. It is in the combination of model, domain context, customer access, and business process integration.
Imagine an AI assistant inside SAP S/4HANA that does more than answer questions. It could analyze procurement records, detect unusual supplier behavior, predict cash flow stress, identify demand changes, or recommend working capital actions. More importantly, it could do this using structured business data that already exists inside enterprise systems. That is a very different proposition from a generic AI chatbot.
Dremio and the Enterprise Data Foundation
If Prior Labs strengthens the intelligence layer, Dremio strengthens the data foundation. Enterprise AI cannot work properly when data is scattered across disconnected systems. Many companies use SAP for core business operations, but they also use cloud warehouses, data lakes, SaaS tools, third-party systems, spreadsheets, and industry-specific platforms. The result is fragmentation. Data lives everywhere, but intelligence struggles to reach it cleanly.
Dremio is known for its open data lakehouse capabilities and its work around Apache Iceberg. In practical terms, it helps organizations query and analyze data across different environments without constantly copying, moving, or transforming everything into one rigid location. That matters because data movement is one of the hidden costs of enterprise AI. Every time data is duplicated, companies introduce cost, latency, governance risk, and security complexity. For AI agents to work reliably, they need access to governed, contextual, and up-to-date data.
SAP’s intent to acquire Dremio fits into this larger objective. It can help SAP Business Data Cloud become a stronger platform for unifying SAP and non-SAP data. This is critical because the real enterprise environment is rarely clean. It is messy, distributed, and full of legacy systems. In that messy environment, the winner is not the company that promises magic. It is the company that reduces friction.
The Synergy: Data Lakehouse + Tabular AI + Agentic Workflows
The powerful part of SAP’s strategy is not Prior Labs alone. It is not Dremio alone. It is the possible synergy between them.
Dremio can help create a unified, governed data layer. Prior Labs can help build AI models that understand structured data more intelligently. SAP can connect both into real business workflows through its existing enterprise software ecosystem. That combination forms a practical enterprise AI stack.
At the bottom, there is trusted business data. Above that, there is a data lakehouse layer that helps unify access across systems. On top of that, there are tabular foundation models that can interpret structured data. Finally, there are AI agents and business applications that turn insights into actions. This is where agentic AI becomes more meaningful.
Many companies talk about AI agents as if they are simply chatbots with extra tools. But in enterprise environments, a useful AI agent must do more than talk. It must reason over live business data, understand permissions, follow workflow rules, trigger actions, maintain audit trails, and explain decisions. For example, an agent working in procurement could detect that a supplier’s delivery reliability is declining. It could compare the trend with inventory buffers, check alternative suppliers, estimate the risk of stockouts, and recommend mitigation steps. In a mature system, it may even prepare a workflow for approval.
That is not science fiction. It is the direction enterprise AI is moving toward. But it requires the right foundation. Without unified data, agents become blind. Without structured-data intelligence, they become shallow. Without workflow integration, they remain demos. SAP appears to understand this.
Why Large Enterprise Firms Are Acquiring AI Startups
SAP’s acquisition strategy also reflects a broader industry pattern. Large enterprise firms are no longer relying only on internal AI research. They are acquiring specialized companies that bring technical depth, research talent, and speed. This is not unusual. In previous technology cycles, large firms acquired companies in cloud computing, cybersecurity, analytics, and automation. The same pattern is now unfolding in AI.
The logic is simple. Building every capability internally takes time. In AI, time matters. Specialized startups often move faster because they are focused on narrow, difficult problems. Enterprise giants bring distribution, trust, compliance experience, customer relationships, and the ability to embed technology into real production environments. A startup may have the breakthrough model. A large enterprise firm may have the customer base and platform depth. When structured correctly, the combination can create serious value.
This is especially true in enterprise AI, where customers care about security, compliance, integration, uptime, auditability, and long-term support. A powerful model alone is not enough. Enterprises need systems that work inside their existing operating environment. That is why acquisitions like these are not just financial transactions. They are capability accelerators.
The Technical Direction: From Data Access to Decision Intelligence
The technical direction behind SAP’s moves is worth noting. Enterprise AI is shifting from information retrieval to decision intelligence. The first stage was helping employees find information faster. The next stage is helping them understand what is happening. The more advanced stage is helping them decide what to do next.
For this to happen, AI must connect multiple technical layers. It needs a data layer that can access structured and semi-structured information. It needs semantic context so that business terms are understood correctly. It needs models that can reason over tables, trends, and exceptions. It needs agentic orchestration to execute tasks across workflows. It also needs governance, because enterprise decisions cannot be based on black-box outputs with no accountability.
Prior Labs and Dremio fit into this chain. Tabular foundation models can improve how AI interprets enterprise tables. A data lakehouse can improve how data is discovered, queried, and unified. SAP’s software ecosystem can connect these capabilities to business processes across finance, procurement, supply chain, HR, and customer operations. The result could be a more grounded version of enterprise AI. Less theatre. More utility.
The Risks: Integration, Regulation, and Market Concentration
Still, the opportunity comes with risks. The first risk is regulatory approval. Large acquisitions often face review, especially when they involve strategic technologies and data infrastructure. Timelines can shift. Conditions may be imposed.
The second risk is integration. Acquiring a startup is easy compared to preserving its innovation culture. Many promising acquisitions lose momentum when they are absorbed into slow corporate structures. SAP’s challenge will be to give teams like Prior Labs enough independence while still connecting their work to commercial products.
The third risk is execution. The promised synergy between tabular AI, data lakehouses, and enterprise agents sounds powerful, but customers will judge the outcome by practical results. Does it reduce implementation friction? Does it improve forecasting? Does it lower analytics cost? Does it make business users more effective? These questions matter more than press releases.
There is also a broader market concern. If the best enterprise AI layers are controlled by a few large platforms, customers may face deeper lock-in. Open standards, interoperability, and transparent data governance will become increasingly important.
What This Means for Enterprises
For enterprise leaders, SAP’s strategy offers an important lesson. AI adoption should not begin and end with chat interfaces. The real question is whether the organization’s data foundation is ready. A company cannot build reliable AI on fragmented, poorly governed, inconsistent data. It cannot expect agents to make useful decisions if they do not understand business context. It cannot scale AI if every use case requires manual data movement, custom integration, and one-off experimentation.
The enterprises that win will focus on architecture. They will unify their data where possible. They will preserve governance. They will invest in models that understand their domain. They will connect intelligence directly to workflows. Most importantly, they will stop treating AI as a separate layer and start treating it as part of the operating system of the business. SAP’s moves point exactly in that direction.
The Road Ahead for Enterprise AI
SAP’s planned acquisition of Prior Labs and its intent to acquire Dremio show where enterprise AI is heading. The future will not be defined only by bigger models or more polished assistants. It will be defined by the ability to turn structured business data into trusted intelligence. This is a more serious phase of AI. It is less noisy, less consumer-facing, and more deeply connected to how companies actually function.
The companies that master this layer will gain a durable advantage. They will not simply answer questions. They will help enterprises predict, decide, and act with more confidence .In that sense, SAP’s acquisition strategy is not just about buying AI startups. It is about rebuilding the enterprise AI stack around data, context, and business execution. And that may be where the real AI race begins.
