As 2025 draws to a close, a striking pattern emerges from the year’s M&A landscape: companies didn’t acquire AI models or coding agents in 2025. They acquired the infrastructure that will power autonomous agents for the next decade.
With 33 major acquisitions totaling $157 billion or more in disclosed value, 2025 broke historic records for consolidation activity in the AI and data space. But the story is not about deal volume. It’s about strategic intent—and what that intent reveals about where enterprise AI is headed.
The Core Insight: Data Infrastructure Became the Competitive Moat
Through the first half of 2025, enterprise CIOs believed the constraint on AI deployment was model capability. Frontier labs were producing increasingly sophisticated LLMs, and enterprises assumed access to these models would unlock AI value.
By December 2025, that assumption had been shattered.
The realisation—crystallised through acquisitions from Salesforce ($8B for Informatica), IBM ($11B for Confluent), and more—was stark: Models don’t matter without trusted, governed, real-time data. Batch data pipelines updated weekly or monthly cannot support autonomous agents making real-time decisions. Data quality issues that humans could manually correct become unacceptable when agents operate at scale.
This is why 2025’s M&A activity was fundamentally different from 2023-2024 venture funding. Venture capital bet on models and talent. Strategic M&A bet on infrastructure—the unglamorous but essential layers that will determine which enterprises successfully deploy agentic AI at scale.
The Broader Pattern: Infrastructure Verticalisation
Beyond mega-deals, a second strategic pattern dominated 2025: infrastructure verticalisation. CoreWeave didn’t just provide GPU compute; it acquired Core Scientific ($9B) to own power infrastructure. AMD didn’t just sell processors; it acquired ZT Systems ($4.9B) to own systems design.
Each company recognised something fundamental: as AI scaling becomes constrained by power, networking latency, and system optimisation—not processor count—controlling the full vertical stack becomes a competitive advantage that cannot be outsourced.
The Talent Acquisition Reality: Speed > Hiring
Accenture made 23 acquisitions in 2025. OpenAI paid $6.5 billion for io (a startup that hadn’t shipped a product). Anthropic acquired Bun. These are talent consolidations executed through acquisition rather than hiring. In the tightest labour market in tech, acquiring boutique specialist firms is faster than competing for headcount on the open market.
The Bottom Line
2025 will be remembered as the year enterprises answered a critical question: What infrastructure do we need to actually deploy agentic AI at scale?
The answer wasn’t “better models.” The answer was data infrastructure, real-time serving, identity security, vertically integrated AI platforms, and autonomous remediation. With these 33 deals and $157 billion in commitments, enterprises have essentially said: We’re betting on the infrastructure that will let us deploy any frontier model safely, governably, and at scale.
Here are the top 16 acquisitions in the AI and Data space worth studying in 2025:
MEGA-DEALS (Strategic Industry Reshaping)
1. Google Acquires Wiz — $32 Billion
Announced: March 18, 2025
Status: Expected to close late 2025 (DOJ approved October 2025)
Strategic Intent: Cloud security and multi-cloud data visibility
Google’s $32 billion all-cash offer for Wiz represents Google Cloud’s most aggressive push to compete with Microsoft and AWS in enterprise security. At a 50x revenue multiple (Wiz’s ARR exceeded $700 million), the valuation appears rich by traditional metrics, but the strategic logic is sound: Wiz provides Google Cloud with real-time visibility into how enterprises manage data across multi-cloud environments—data invaluable for training enterprise-grade AI models.
Beyond security, this acquisition signals Google’s recognition that controlling multi-cloud observability is a prerequisite for selling cloud services in an era where enterprises refuse partner lock-in. Wiz’s platform sees data governance, security policies, and infrastructure decisions across thousands of organizations—a vantage point Google leverages to refine Google Cloud’s competitive positioning.
Impact: Accelerates consolidation in cloud security. Microsoft and AWS must respond with competitive acquisitions in the multi-cloud security space.
INFRASTRUCTURE VERTICALISATION (Ownership of Physical Substrate)
2. CoreWeave Acquires Core Scientific — $9 Billion
Announced: July 6, 2025
Status: All-stock transaction completed
Strategic Intent: Vertical integration of AI compute capacity
CoreWeave’s acquisition of its largest customer and long-standing data center partner eliminates $10 billion in prospective 15-year lease obligations while providing 1.3 gigawatts of gross power capacity. This transaction exemplifies the shift toward infrastructure verticalization: rather than renting compute resources from third parties, CoreWeave now directly owns and controls the physical substrate powering AI training.
The strategic win is operational: CoreWeave gains direct control over power distribution, cooling systems, and expansion capacity—factors increasingly critical for marginal unit economics in GPU-intensive AI infrastructure. Power is becoming the bottleneck resource in AI scaling; owning data centers provides CoreWeave with leverage in customer negotiations and the flexibility to prioritize power allocation for its most profitable customers.
Impact: Other AI infrastructure providers (Lambda Labs, Crusoe Energy) face pressure to vertically integrate. Traditional data center operators must differentiate on AI-specific services.
3. AMD Completes ZT Systems Acquisition — $4.9 Billion
Announced: March 31, 2025
Status: Completed; manufacturing divested to Sanmina for $3B
Strategic Intent: Rack-scale AI systems integration
AMD’s acquisition of ZT Systems represents the semiconductor company’s strategic pivot from selling individual processors to providing integrated rack-scale AI solutions competing directly with Nvidia’s offerings. By retaining ZT Systems’ design and customer enablement teams while divesting the manufacturing business, AMD gains engineering talent specialized in bundling processors, networking silicon, and system architecture into purpose-built AI infrastructure.
This move reflects a critical market shift: AI infrastructure customers increasingly prefer integrated solutions rather than component procurement. ZT Systems’ expertise in designing high-performance systems around AMD’s EPYC processors and custom networking is more valuable than the manufacturing capability itself.
Impact: Narrows the gap between AMD and Nvidia in enterprise AI infrastructure sales. Signals that competing with Nvidia requires vertical integration of systems design.
DATA INFRASTRUCTURE AS STRATEGIC MOAT (Governance, Serving, Integration)
4. Salesforce Acquires Informatica — $8 Billion
Announced: May 27, 2025
Status: Expected to close early FY2027 (Feb 2026)
Strategic Intent: Agent-ready data platform foundation
Salesforce’s $8 billion acquisition of Informatica represents the clearest statement that agentic AI requires trusted, governed data infrastructure. Marc Benioff explicitly framed the deal as creating an “agent-ready data platform” where autonomous agents can safely operate on enterprise data.
Informatica’s data integration, catalog, master data management, and data quality capabilities are non-negotiable for enterprise AI deployment. As agents become autonomous decision-makers, enterprises cannot tolerate data quality failures or governance violations. The acquisition consolidates Salesforce’s Data Cloud, MuleSoft, and Tableau with Informatica’s governance stack—creating an end-to-end platform where agents have clean, contextualized, compliant data.
Impact: Establishes data governance as essential infrastructure for AI. Competing platforms (Microsoft, Oracle) must accelerate their own data integration acquisitions.
ALSO READ: Salesforce to Acquire Informatica
5. Databricks Acquires Neon — $1 Billion
Announced: May 13, 2025
Status: Completed
Strategic Intent: AI-native database infrastructure
Databricks’ $1 billion acquisition of Neon, a serverless Postgres database, reflects an critical insight: 80% of new databases provisioned on Neon are created automatically by AI agents. This single statistic drove Databricks’ acquisition—the company recognized that agentic AI requires database infrastructure designed for autonomous workload management.
Neon’s architecture (serverless, auto-scaling, branching capabilities) is fundamentally different from traditional relational databases because it assumes machines, not humans, will be managing schema changes, compute provisioning, and concurrent access patterns. For Databricks, owning Neon means embedding database infrastructure into its broader AI platform.
Impact: Traditional relational database providers (PostgreSQL, MySQL) face pressure to adopt AI-native architectural patterns or risk obsolescence in agentic workflows.
ALSO READ: Databricks Agrees to Acquire Neon
6. Databricks Acquires Tecton
Announced: August 22, 2025
Status: Completed
Strategic Intent: Real-time feature serving for autonomous agents
Databricks‘ acquisition of Tecton (a real-time feature store) paired with the Neon acquisition creates an essential capability: serving fresh, contextual, sub-100 millisecond feature data to autonomous agents. Traditional batch ETL pipelines cannot support agents that must make real-time decisions.
Feature stores are specialized infrastructure that serve pre-computed features to ML models with minimal latency. Tecton’s focus on real-time serving (versus batch feature computation) makes it uniquely suited for agentic workloads where decision latency directly impacts user experience and business outcomes.
Impact: Establishes real-time feature serving as essential infrastructure for production AI. Competitors (Feast, Milvus) must accelerate feature-serving capabilities.
7. IBM Acquires Confluent — $11 Billion
Announced: December 8, 2025
Status: Definitive agreement signed; expected close mid-2026, subject to Confluent shareholder
Strategic Intent: Real-time data streaming backbone for enterprise agentic AI deployment across hybrid cloud environments
IBM’s acquisition of Confluent represents a watershed validation that real-time, event-driven data infrastructure is essential—not optional—for autonomous agent deployment. Unlike batch data pipelines that operate on stale information, Confluent’s streaming platform enables agents to access fresh, contextualized data at sub-100 millisecond latencies.
The strategic win is Confluent’s TAM doubling from $50 billion (2021) to $100 billion (2025)—reflecting enterprise recognition that real-time data is becoming table-stakes infrastructure. This acquisition directly complements IBM’s earlier 2025 acquisitions of HashiCorp ($6.4B, February) and Seek AI (June), creating a vertically integrated enterprise AI data infrastructure stack: infrastructure provisioning → HashiCorp; data understanding → Seek AI; real-time data movement → Confluent; AI models → Watsonx; workflow automation → IBM Automation.
Impact: Confirms real-time data as essential for agentic AI, shifting enterprise budget allocation from batch ETL to streaming infrastructure. Positions IBM as most active mega-acquirer in 2025 ($27.8B across three deals), directly competing with Salesforce and cloud-native providers on complete AI infrastructure stacks.
ALSO READ: IBM in Talks to Buy Confluent for $11 Bn in Cloud, AI Push: Report
AI TALENT & PRODUCT ACQUISITIONS (Specialised Capability)
8. OpenAI Acquires io — $6.5 Billion
Announced: May 21, 2025
Status: Completed July 2025
Strategic Intent: Hardware-software integration and consumer AI devices
OpenAI’s $6.5 billion acquisition of Jony Ive’s io represents a fundamental strategic pivot toward hardware. Founded just one year prior by design legend Jony Ive and former Apple executives, io had released no products. Yet OpenAI valued the team at $6.5 billion, with Ive and his design firm LoveFrom assuming “deep design and creative responsibilities.”
This acquisition signals OpenAI’s ambition to create consumer-facing AI devices combining hardware elegance with frontier AI capability—directly competing with Apple’s potential consumer AI roadmap. The deal is fundamentally a talent and capability play: acquiring design excellence that can translate LLM breakthroughs into intuitive, hardware-integrated experiences.
Impact: Establishes hardware expertise as essential for consumer AI companies. Signals that API-only business models are insufficient for long-term AI competitiveness.
ALSO READ: OpenAI to Buy Jony Ive’s AI Devices Startup for $6.4Bn
9. Anthropic Acquires Bun
Announced: December 2, 2025
Status: Completed; Founder Jarred Sumner joins Anthropic
Strategic Intent: JavaScript runtime for Claude Code execution infrastructure
Anthropic’s first acquisition of Bun, a high-performance JavaScript runtime, represents vertical integration of code execution infrastructure. Claude Code reached $1 billion ARR in just six months; Bun provides the lightweight runtime (7M+ monthly downloads) executing all Claude-generated code.
Bun’s architecture (written in Zig, using JavaScriptCore) enables sub-second startup times and 4x faster application startup versus Node.js—critical for responsive experiences when Claude generates, tests, and iterates on code. By acquiring Bun, Anthropic gains direct control over the execution layer supporting Claude Code, enabling tighter Constitutional AI integration and security constraints.
Impact: Establishes code execution infrastructure as essential vertical integration for AI coding platforms. GitHub Copilot and other competitors must secure equivalent infrastructure.
ALSO READ: 6 Revolutionary AI Coding Models Transforming Developer Workflows in 2025
10. Google Acquires Galileo AI
Announced: May 19, 2025
Status: Completed
Strategic Intent: AI-powered UI design and code generation
Google’s acquisition of Galileo AI brings AI-powered UI design generation capabilities into Google’s ecosystem. As enterprise applications become increasingly AI-native, the ability to automatically generate user interfaces that match design systems becomes strategically important.
This represents Google‘s broader strategy of embedding AI across the entire product development stack—from foundational models to design tools to deployment infrastructure.
Impact: Establishes UI design automation as essential infrastructure for AI application development.
11. Accenture Acquires NeuraFlash
Announced: August 27, 2025
Status: Completed September 29, 2025
Strategic Intent: Salesforce and agentic AI consulting at scale
Accenture’s acquisition of NeuraFlash brought 510 Salesforce-certified professionals into Accenture’s workforce, enabling rapid scaling of agentic AI implementation services. NeuraFlash specialized in building conversational AI, digital customer support, and field-service automation.
Accenture made 23 acquisitions in 2025, but NeuraFlash represents the most significant for understanding the consulting industry’s response to agentic AI: acquiring specialized talent rather than competing for engineering time in an overheated labor market.
Impact: Establishes acquiring boutique consulting firms as the fastest path to scaling AI expertise within larger service providers.
12. Cognition Acquires Windsurf
Announced: July 14, 2025
Status: Completed; CEO hired by Google simultaneously
Strategic Intent: AI coding assistant integration
Cognition’s acquisition of Windsurf represents a multi-sided acquisition battle: OpenAI bid $3B, Google hired the CEO for DeepMind, and Cognition ultimately acquired the remaining assets. The deal illustrates intense competition for specialized AI talent.
Windsurf’s lightweight coding assistant integrated naturally with Cognition’s Devin AI coding agent, enabling Cognition to offer complementary tools to developers using its platform.
Impact: Demonstrates that frontier AI labs will aggressively acquire talent and specialised AI capabilities in competitive categories.
ALSO READ: LLM Developers Building for Language Diversity in 2025
EMERGING PATTERN: INFRASTRUCTURE FOR AGENTS
13. Hugging Face Acquires Pollen Robotics
Announced: April 13, 2025
Status: Completed
Strategic Intent: Embodied AI and robotics
Hugging Face’s acquisition of Pollen Robotics signals an emerging intersection of AI models and embodied robotics. Pollen’s Reachy 2 humanoid robot, deployed at Carnegie Mellon and Cornell, represents the physical instantiation of AI agents.
By acquiring Pollen, Hugging Face gains research assets and customer relationships in embodied AI—positioning the company to lead the integration of foundation models with physical robotics.
Impact: Establishes robotics as an emerging category where foundation model providers must develop embodied AI capabilities.
14. Amazon Acquires Bee
Announced: July 21, 2025
Status: Completed
Strategic Intent: Wearable AI assistant
Amazon’s acquisition of Bee (wearable AI listening device) represents the company’s competitive response to OpenAI’s consumer AI ambitions. As AI assistants move from screens to form factors like wearables and ambient devices, owning specialized hardware becomes strategically important.
This acquisition signals Amazon‘s intention to compete in consumer AI across multiple form factors (Echo devices, Alexa, now Bee wearables).
Impact: Establishes wearable form factors as essential for consumer AI strategies.
15. IBM Completes HashiCorp Acquisition — $6.4 Billion
Announced: February 27, 2025
Status: Completed
Strategic Intent: Multi-cloud infrastructure automation
IBM’s $6.4 billion completion of its HashiCorp acquisition represents infrastructure automation as the strategic foundation for IBM’s cloud offerings. HashiCorp’s Terraform and Vault products are essential for enterprises managing infrastructure across multiple cloud providers.
As AI workloads increasingly require multi-cloud deployment for flexibility and cost optimization, owning infrastructure-as-code tools provides IBM competitive advantage.
Impact: Establishes infrastructure-as-code and secrets management as essential for multi-cloud AI deployment.
ALSO READ: Geopatriation for Cloud Sovereignty: Why 75% Are Moving Home by 2030
16. Stripe Acquires Metronome — ~$1 Billion
Announced: December 3, 2025
Status: Completed
Strategic Intent: Usage-based billing for AI consumption
Stripe’s $1 billion acquisition of Metronome (usage-based billing platform) represents an essential shift in AI monetization: from subscription to consumption-based pricing. Metronome’s platform powers billing for NVIDIA, OpenAI, Anthropic, and other AI infrastructure providers.
As AI agents become more autonomous and unpredictable in compute consumption, consumption-based billing becomes essential for both providers and customers. Stripe’s acquisition of Metronome signals the payment infrastructure company’s recognition that AI infrastructure monetization differs fundamentally from traditional SaaS.
Impact: Establishes consumption-based billing as essential infrastructure for AI platform monetization.
Final Takeaway
2025 was the year of strategic positioning for the agentic AI era. Companies that won—Google (Wiz), Salesforce (Informatica), , Anthropic (Bun) —all made structural investments in infrastructure rather than feature-level acquisitions.
The winners in 2026 will be those that:
- Own data infrastructure controlling training and serving of AI models
- Control security from identity to AI-specific threat defense
- Vertically integrate infrastructure from power to networking to compute
- Acquire talent faster than competitors in scarce specialisations
- Build consumer capabilities for hardware-integrated AI
The following will be at a loss:
- Point-solution providers that remain unbundled
- Companies acquiring for financial engineering rather than strategic capability
- Providers unable to move fast enough in recruiting AI talent
- Infrastructure providers unable to integrate vertically
The M&A activity in 2025—$146B+ in disclosed deals—is less about consolidation and more about strategic repositioning for agentic AI dominance. The second half of 2026 will reveal whether these acquisitions created competitive moats or expensive distractions.
ALSO READ: The Unspoken Prerequisite by AWS: Enterprise AI Must Solve Modernisation First
