Salesforce, Microsoft and SAP Race to Own AI’s Trusted Data Layer

Salesforce closed its $8 billion purchase of Informatica in November, then spent six months turning a back-office data vendor into the anchor of its entire AI pitch. By May, the combined company was shipping tools that let an AI agent clean, match and govern enterprise records with no person touching them.

The bet is that owning trusted, governed data now decides who wins the next phase of enterprise AI. Microsoft, SAP, Snowflake and Databricks are placing similar bets of their own, and the same month Salesforce showed off its new tools, SAP moved to absorb one of the last big independent master data management vendors. Consolidation is real. Whether it helps the enterprises paying for all of it remains unsettled.

Salesforce Turns an $8 Billion Bet Into a Live Product

Salesforce’s pursuit of Informatica had been rumored for more than a year, with a credible expectation of an $11 billion bid the year before, said Hyoun Park, chief executive and chief analyst at Amalgam Insights. The price that closed was lower. Salesforce signed a definitive agreement to acquire Informatica for approximately $8 billion in cash. The deal brought Informatica’s data catalog, integration, governance, quality and privacy, metadata management and master data management (MDM) services onto the Salesforce platform, according to Informatica’s November announcement confirming the deal’s completion on November 18, 2025.

“You have to get your data right to get your AI right,” said Marc Benioff, Salesforce’s chair and chief executive. He argued that without clean, connected data behind it, Agentforce, Salesforce’s AI agent platform, has nothing to reason with beyond guesswork. Salesforce told investors the deal would add to non-GAAP earnings within twelve months, a full year sooner than it had originally promised.

By Informatica World in Las Vegas this past May, the strategy had become software. Informatica from Salesforce rolled out fully headless data management, turning every data-management capability into a service any AI agent can call through the Model Context Protocol (MCP), the open standard that lets AI systems reach outside tools and data, detailed in Salesforce’s own announcement of the headless data tools. A new Agent Fabric Context Catalog created one destination to discover, govern and operate both data assets and AI agents together. Informatica also launched what it called the industry’s first agentic multidomain MDM system, letting AI agents cleanse and enrich master data in real time instead of waiting on manual review.

Edmond Tarée, an area lead for group risk and finance data at Rabobank, one of the customers named in the launch, said “data quality and the right platform are critical to move the data and make data ready for distribution.”

The Failure Numbers Behind the Rush

The vendors keep repeating the same argument because the survey data backs it up. Enterprises keep pushing AI pilots into production on top of data nobody cleaned up first, and the failure rate shows it.

• 76% of data leaders say governance has not kept pace with AI, per a 2026 survey of chief data officers cited by Informatica.
• 89% of data and analytics leaders say a strong data foundation is the most critical factor for successful AI adoption, according to Salesforce’s own State of Data and Analytics research.
• 60% of AI projects lacking sufficient data quality will be abandoned through 2026, Gartner has projected.
• 95% of organizations deploying generative AI saw zero measurable return on investment, according to MIT’s Project NANDA research.

Trust is having a moment across the industry this year, in more than one sense of the word. OpenAI added a trusted contact feature to ChatGPT this spring, a different wager that the same word could reassure a very different audience. Salesforce is betting it matters just as much to a data leader signing a multi-year contract.

Every Platform Giant Wants the Same Layer

Salesforce is not alone in chasing this. Every major enterprise software vendor is racing to become the layer that sits between raw data and the agents acting on it, even where each one uses different branding for the idea.

The Microsoft tie-up is the most concrete of the group so far. Informatica expanded its cloud data integration support for Microsoft Fabric, letting customers ingest billions of rows a month into Fabric OneLake or Fabric Data Warehouse while using change data capture to hold down compute costs, per Salesforce’s announcement of the deepened Microsoft collaboration. Salesforce’s own research found 89% of data and analytics leaders believe agent interoperability across vendors will soon be required just to do business.

The pattern is not confined to the giants either. Revvity’s Signals platform recently plugged into Claude through its own MCP connector, a life-sciences data business making the same bet that governed, agent-ready data is worth building toward, well outside the software giants fighting over the enterprise mainstream.

Who Loses When the Platforms Consolidate?

Consolidation always produces a shorter list of independent options, and enterprise data management is no exception. A sleepy, fragmented category is turning into a fight among a handful of platform owners.

Enterprise winners won’t be the ones with the best dashboards or biggest models.

Michael Ni, principal analyst at Constellation Research, made that call within weeks of the Informatica deal being announced, arguing the industry’s pecking order would flip within a year toward whoever owns the strongest data foundation. Gartner’s most recent Magic Quadrant for the category still lists five leaders, though the group looks different than it did twelve months ago.

• Salesforce (Informatica) – now the largest of the group, no longer sold as standalone software after being absorbed into Salesforce’s platform.
• Profisee – stays independent, closely aligned with Microsoft-centric environments.
• Reltio – acquired by SAP in May 2026, kept for its graph-oriented, API-first architecture.
• Semarchy – competes where data integration and MDM converge.
• Stibo Systems – holds ground in product information management and retail-focused deployments.

Two of the five leaders now sit inside larger software conglomerates. A category that spent two decades as an unglamorous back-office purchase is now something platform vendors pay billions to own outright.

Salesforce’s Own History Is the Fine Print

Salesforce has done this before, and the record is mixed. Salesforce’s $60 billion acquisition track record over the past decade has produced mixed results, according to BARC’s analysis of the acquisition’s customer risk, led by VP of Research Kevin Petrie. Analysts have pointed to the Tableau acquisition as a cautionary precedent, where standalone customers experienced pricing challenges and reduced product focus after Salesforce folded it into the broader platform.

Other analysts read the same history differently. VentureBeat, reporting on the deal shortly after it was announced, argued that Tableau, Slack and MuleSoft had all worked out well for Salesforce, growing and expanding under its ownership. Forrester’s Noel Yuhanna, vice president and principal analyst, called the purchase a “bold and highly strategic move”, telling another outlet it closed a “critical gap in its data management capabilities.”

A newer worry has nothing to do with acquisitions at all. Gartner predicts that by 2027, 40% of enterprises will demote or decommission autonomous AI agents because of governance gaps identified only after production incidents occur, per Gartner’s May 2026 warning on agent governance failure. Shiva Varma, senior director analyst at Gartner, said “enterprises are treating AI agent governance as binary, either locked down or fully trusted,” and called that the root cause of failure.

The failure mode is already showing up. In March 2026, an in-house agent at Meta posted incorrect technical information publicly without human approval, triggering two hours of unauthorized data exposure to employees who were not cleared to see it. It was the second agent control failure at the company within weeks, according to a governance analysis published by CloudEagle. Stanford’s 2026 AI Index found that 62% of organizations now cite security and risk as the primary barrier to scaling agentic AI, ahead of technical limitations and budget.

That tension mirrors a pattern researchers have flagged elsewhere: leaning on automated systems tends to dull the human judgment meant to catch problems before they spread, a dynamic researchers have already documented in how AI use reshapes everyday thinking. Salesforce’s answer is more automation layered on top, not less, betting that better tooling closes the gap machine autonomy keeps opening.

Deeper integration between Informatica, Data 360 and Agent Fabric is expected to continue through the rest of 2026, the same year Salesforce told investors its Informatica bet would start paying for itself.

Frequently Asked Questions

What Does “Trusted Context” Mean in Enterprise AI?

Trusted context is the data an AI agent draws on, plus enough meaning, history and lineage attached to it that the agent can act with confidence instead of guessing. Salesforce and Informatica describe it as raw enterprise data unified with the meaning, history and relationships that make it accountable and usable by AI. That is a different idea from a model’s context window, the technical limit on how much text an AI system can process in one request.

Why Did Salesforce Buy Informatica?

Salesforce needed data management tools it did not already own well enough to build itself. Pareekh Jain, principal analyst with Pareekh Consulting, expected new customers to come mainly from financial services, healthcare and airlines, sectors where Informatica already had a stronger foothold than Salesforce’s own sales force.

What Is Master Data Management, and Why Does AI Need It?

Master data management keeps one accurate, agreed-upon record for each customer, product, supplier or account instead of scattered, conflicting copies across systems. The category is growing fast on AI demand: the global MDM market is projected to reach $34.5 billion by 2027, growing at a 15.7% compound annual rate, according to MarketsandMarkets research.

Is Informatica Still an Independent Company?

No. Informatica now operates as part of Salesforce, continuing to support its existing data management solutions and partner integrations. Amit Walia, Informatica’s chief executive at the time of the deal, wrote that the combination would let AI agents operate safely and at scale across the enterprise once the transaction closed.

Which Other Companies Are Competing for the Same Data Layer?

Microsoft, Snowflake, Databricks, SAP and Oracle are all building comparable layers under different names. The Microsoft partnership already has specifics attached: Informatica customers can ingest billions of rows each month into Microsoft’s Fabric OneLake or Fabric Data Warehouse, using change data capture to hold down compute time and cost.

What Could Go Wrong With This Consolidation?

Two opposite failure modes, not one. Overly strict governance forces low-risk agents through compliance checklists built for high-stakes work, breaking delivery speed and pushing engineers toward unmonitored shadow AI instead, while overly loose governance lets highly autonomous agents reach further into systems than anyone approved. Gartner’s warning is that applying the same rulebook to both extremes guarantees one failure or the other.