When Salesforce paid $8 billion for Informatica in November 2025, it signaled a decisive acceleration in enterprise AI M&A. While the headlines fixated on the price, The more instructive detail was buried in the press release: the acquisition’s primary purpose was to give Agentforce — Salesforce’s autonomous agent platform — clean, governed, standardised data to act on. Not a new model. Not a better interface. Data plumbing. That deal, alongside more than 33 others totalling at least $157 billion in disclosed value, tells a specific story about where enterprise software incumbents believe the durable competitive advantage in AI actually lies. Source: index.dev
The latest wave of enterprise AI M&A is about buying data infrastructure, not just models. The IBM and Confluent acquisition is a clear example of how streaming data has become a strategic asset.
That consolidation pressure is visible across the market. The enterprise AI vendor consolidation trend suggests buyers are favoring fewer, larger suppliers, which raises the stakes for every enterprise AI M&A deal that follows.
For buyers, the key question is whether each enterprise AI M&A adds capability or simply adds another contract to manage.
How Enterprise AI M&A Reshapes Infrastructure Buying
The largest enterprise AI M&A deals of 2025 were not bets on frontier model capability. They were bets on the infrastructure that makes models useful at enterprise scale.
IBM’s $6.4 billion acquisition of HashiCorp, completed 27 February 2025, gave the company Terraform — the dominant infrastructure-as-code tool for provisioning multi-cloud environments where generative AI applications run. Source: TechCrunch IBM then followed with an $11 billion acquisition of Confluent in December 2025, adding real-time data streaming infrastructure that autonomous agents need to make decisions without waiting on batch pipelines. Source: index.dev Databricks acquired serverless Postgres company Neon for approximately $1 billion in May 2025, explicitly because more than 80% of databases provisioned on Neon were being created automatically by AI agents — the platform was already operating at machine speed. Source: CNBC
The common thread across recent enterprise AI M&A: none of these acquisitions improves a model’s reasoning capacity. All of them improve what a model can see and act on. Incumbents have concluded that model performance is converging — and that the durable moat is governed, real-time enterprise context.
The Enterprise SaaS Arms Race
ServiceNow committed at least $12 billion to acquisitions and strategic investments in 2025, more deal-making than in any prior year of the company’s history. Source: Bloomberg Its $2.85 billion acquisition of Moveworks, completed in December 2025, illustrates the logic precisely: Moveworks had solved enterprise AI search — helping employees find information across systems. ServiceNow then bolted that capability onto its workflow execution platform, so an agent can not only locate a policy document but act on it. Source: ServiceNow Newsroom ServiceNow’s $7.75 billion Armis acquisition extended that into security identity data — knowing not just what an employee needs, but what access they hold. Source: NowBen
Salesforce ran a parallel strategy at larger scale, completing 12+ acquisitions totalling more than $10 billion in 2025. Source: Medium The Informatica deal was its centrepiece: Informatica’s data catalog, master data management, and governance tooling provides the metadata layer that Agentforce agents query before acting. Source: Salesforce investor relations The strategic frame in Salesforce’s own announcement was direct: agents need trusted data or they will act on bad data — and bad data at agent speed produces enterprise-scale errors.
Google’s $32 billion Wiz acquisition, completed in March 2025 — the largest in Alphabet’s history — was framed as a cloud security play. But it also gave Google the most comprehensive visibility layer across hybrid and multi-cloud customer deployments, precisely the position a cloud provider needs to surface contextual data to enterprise agents wherever those agents run. Source: index.dev
Novel Deal Structures and Regulatory Blind Spots
The Windsurf episode exposed a structural feature of 2025–2026 enterprise AI M&A that regulators are still catching up with. OpenAI agreed to acquire AI coding assistant Windsurf for $3 billion in May 2025. The deal collapsed — reportedly due to conflicts with investor Microsoft — and Google then hired Windsurf’s CEO and key R&D staff while licensing certain Windsurf technologies in a package valued at roughly $2.4 billion. Source: Bloomberg / CNBC
The result was functionally indistinguishable from an acquisition — talent transferred, IP licensed, startup left as a rump entity — but structured to remain below Hart-Scott-Rodino merger notification thresholds. The FTC has opened inquiries into Microsoft’s arrangement with Inflection AI and Google’s with Character.AI, which followed the same template. Source: American Action Forum / Skadden The European Commission launched a comprehensive review of its merger guidelines in May 2025, the first substantial revision in over two decades, partly in response to these structures. Source: Goodwin Antitrust Year in Review 2025
The regulatory gap in enterprise AI M&A is real: a $2 billion talent-plus-IP transaction can currently avoid the scrutiny that a $2 billion direct acquisition would trigger. That asymmetry is structuring deal teams’ behaviour.
What Acquirers Are Paying — and Why
Analysis of more than 90 enterprise AI M&A deals by Finro Financial Consulting found an average EV/Revenue multiple of 25.8x across the sector in 2025. LLM vendors commanded 54.8x; data intelligence companies reached 41.7x; cybersecurity AI targets averaged 20.4x. Source: Finro For context, the broader SaaS sector traded at roughly 6–8x revenue during the same period — making the AI premium stark.
The premium reflects two overlapping pressures. First, genuine scarcity: the number of companies that have solved enterprise-grade data governance or real-time streaming at scale is small, and each one has network effects from customer integration. Second, competitive urgency: 83% of buyers reported paying higher multiples for AI-native targets, with 86% expecting those premiums to persist through 2026. Source: Development Corporate The paradox is that 63% of targets actually evaluated by buyers had only limited AI integration — meaning buyers are paying AI-native premiums for companies that are, at best, AI-adjacent.
Private equity is also active across enterprise AI M&A, though with a different posture. Thoma Bravo completed acquisitions of Verint Systems (26 November 2025) and PROS Holdings (approximately $1.4 billion), in both cases merging the acquired AI-enabled platform with an existing portfolio company to create a combined SaaS entity targeting a specific enterprise workflow. Source: Thoma Bravo PE’s thesis differs from strategics’: build scale within a vertical rather than extend a horizontal platform.
What to Watch
The agentic orchestration layer. No incumbent has yet bought a clear leader in multi-agent orchestration — the systems that coordinate dozens of specialised agents across enterprise workflows. Enterprise search company Glean and AI coding tool Cursor are frequently cited as 2026 acquisition targets by investors, precisely because they have demonstrated product-market fit in high-value segments. Source: Fortune
Regulatory crystallisation. The FTC’s ongoing review of licensing-plus-hire structures will determine whether the Windsurf/Character.AI template remains viable. If the Commission moves to expand HSR thresholds to cover de facto acquisitions, deal structures will shift — and valuations for acqui-hire targets may reprice downward.
The observability gap. Monitoring AI agents in production — understanding why an agent took an action, auditing decisions, detecting drift — is not yet owned by any major incumbent. Companies like Datadog and Sentry are positioned at this intersection, identified by investors and founders as likely 2026 acquisition targets. Source: Fortune The category that M&A activity has not yet reached may be the one most urgently needed.
Microsoft’s next move. Beyond the Osmos acquisition (January 2026, integrating agentic data engineering into Microsoft Fabric), Microsoft has been comparatively less acquisitive in disclosed AI-specific deals than Salesforce or ServiceNow in the enterprise AI stack. Its deep investment in OpenAI — whose total valuation reached approximately $135 billion after October 2025’s recapitalisation — leaves the company dependent on a single model provider for the AI capabilities it bundles across 365, Copilot, and Azure. Source: Microsoft Blog That concentration is a strategic risk, and an acquisition in AI coding tools or enterprise search could address it.
Further Reading
- ServiceNow’s $12bn Buying Spree Echoes CEO’s SAP-Era Playbook — Bloomberg
- Salesforce Completes Acquisition of Informatica — Salesforce investor relations
- AI M&A Valuation Multiples 2025 — Finro Financial Consulting
- FTC Eyes Reverse Acquihires in AI Sector — American Action Forum
- AI Startups Founders and VCs Say Could Be Acquisition Targets in 2026 — Fortune
This article was produced with AI assistance and reviewed by the editorial team.
4 thoughts on “Enterprise AI’s $157bn Shopping Spree Targets Data Pipelines, Not Models”