Situation assessment: Enterprise AI has a data problem, and two of the biggest players just joined a group that wants to fix it.
Google and Microsoft have joined an industry group that’s working to standardize how enterprise data gets prepared for AI, according to The Information. The goal is a shared way to describe, structure and exchange business data so AI models and agents can actually use it. That can sound like dull plumbing. In practice, it’s the plumbing that decides whether corporate AI projects ship or stall.
The Information’s report gives only the headline facts, so the group’s full membership, technical scope and timeline aren’t clear yet. The direction is clear, though. The companies that sell AI models and cloud platforms are starting to agree on the data layer underneath them.
🎯 Why This Matters
Ask any company that’s tried to roll out AI agents what went wrong. The answer is rarely the model. It’s usually the data.
Most enterprise data is scattered across CRMs, warehouses, spreadsheets and legacy systems. Each one defines things its own way. One system’s “customer” doesn’t match another’s “account.” “Revenue” gets calculated three different ways depending on who you ask. People work around that mess on instinct. AI agents can’t.
A common standard would give AI systems one shared vocabulary for business data. Here’s what that means for you:
- Less custom integration work. Teams wouldn’t have to rebuild data connectors for every new AI tool.
- More portability. Data prepared for one vendor’s AI could work with another’s, which weakens lock-in.
- Fewer hallucinated answers. Agents that understand what a metric actually means are less likely to confidently report the wrong number.
- Faster deployment. Standardized data cuts down the long prep phase that sinks so many enterprise pilots.
📜 How We Got Here
The AI industry has been moving toward shared standards for about two years. The best-known example is the Model Context Protocol (MCP), which Anthropic created to connect AI models to outside tools and data sources. Competitors picked it up quickly. It turned into a de facto standard because nobody wanted to build and maintain a dozen incompatible connectors.
Data and analytics vendors have also started working together on shared definitions for business metrics and semantics. The message from customers has been consistent: they want AI that works across their whole stack, not inside one vendor’s walled garden.
Google and Microsoft signing on is a big deal. Microsoft runs Azure, Fabric and the Microsoft 365 data layer that holds a huge share of corporate information. Google runs Google Cloud and BigQuery. When two cloud giants back a standard, it has a real shot at becoming the default. Without them, it probably wouldn’t.
⚖️ The Strategic Read
This goes deeper than goodwill. It’s a strategic move.
Cloud providers make money when AI workloads run on their platforms. Messy data slows adoption, and slow adoption means less compute spending. Standardizing the data layer grows the whole market, and Google and Microsoft are positioned to take a large share of that growth.
There’s a defensive side too. If a standard is going to exist either way, it’s better to help write it than to have one written for you. Joining early gives both companies a say in how the specs develop.
What stands out is the competitive tension. These companies fight hard over enterprise AI contracts. Cooperating on data standards suggests they’ve decided the real fight is at the model and application layer, not over who owns the file format.
🛠️ What Practitioners Should Do
- Watch the spec. Once technical details come out, check how well they fit your current data models.
- Audit your definitions. Standards help most when your internal metrics are already consistent. Start cleaning those up now.
- Don’t overbuild custom glue. If you’re planning a big proprietary data layer for AI agents, think about whether a standard could make it unnecessary within a year.
- Ask your vendors. Find out whether your data platform plans to support the standard, and on what timeline.
🔭 What Comes Next
Standards bodies move slowly, and adoption isn’t guaranteed. The key signals to watch are which other major vendors join, whether an open specification ships, and whether AWS gets involved. A data standard without the largest cloud provider would have an obvious gap.
Still, the direction is clear. The enterprise AI race is moving from “whose model is smartest” to “whose AI can actually read your business data.” You’ll find more details in The Information’s original report.