AI M&A Trends in Q2 2026: Who's Buying What and Why
The AI acquisition market has matured considerably since the frenzied acqui-hire days of 2023-2024. In Q2 2026, you're seeing more structured deals, higher scrutiny on revenue multiples, and a clearer strategic logic behind what the big players are buying. The acqui-hire as a strategy has mostly disappeared; buyers want products and customers now, not just talent.
Here's what's happening in AI M&A as of mid-2026, with specific deals, valuation patterns, and the strategic dynamics driving them.
The macro picture
Total AI deal volume in Q1 2026 was approximately $8.2 billion across 134 transactions, according to CB Insights. That's down from the $12.4 billion peak in Q3 2024 but up from the trough of $5.8 billion in Q2 2025. Deal count is actually up year-over-year, which tells you the market has shifted toward smaller, more disciplined deals rather than massive bets on pre-revenue companies.
The median AI acquisition in Q1 2026 was approximately $45 million, a stark contrast to 2023-2024 when $200-500 million deals for companies with minimal revenue were common. Buyers have gotten more disciplined. Sellers who raised at 2024 valuations are having difficult conversations.
One notable shift: the strategic vs financial buyer mix has changed. In 2024, financial buyers (private equity, growth funds) participated heavily. In Q2 2026, the acquirers are almost entirely strategic. PE has largely stepped back from early-stage AI, uncomfortable with the combination of high valuations and unclear moats.
What Salesforce has been building through acquisitions
Salesforce has been the most active large enterprise buyer in AI over the past six months. Their strategy is clearer than most: they're acquiring companies that extend Agentforce, their enterprise AI agent platform, into specific vertical domains.
The acquisition of Convergent AI (a manufacturing workflow automation company) for a reported $340 million in March 2026 was the most visible of these. Convergent had roughly $22 million ARR, making the multiple approximately 15x, which would have looked cheap in 2024 but is near the high end of current market. The rationale was clear: Salesforce doesn't have meaningful penetration in manufacturing, and Convergent had relationships with 200+ industrial customers that Salesforce couldn't have built organically in two years.
The smaller deals are more instructive. Salesforce has made six acquisitions in the $30-80 million range in early 2026, all of them targeting specific Agentforce capability gaps: better calendar and scheduling agents, improved document processing, better integration with ERP systems. These aren't headline-grabbing deals, but they show systematic portfolio-building strategy rather than trophy hunting.
Databricks and the data layer build-out
Databricks is the other company operating with clear M&A logic in 2026. After raising at a $62 billion valuation in late 2024, they've been deploying capital into acquisitions that reinforce their position as the data infrastructure layer beneath AI applications.
The acquisition of Einblick Analytics for approximately $180 million in February 2026 gave them a natural language-to-SQL interface that's now integrated into Databricks SQL. The strategic bet: if you can query your data with plain English questions rather than SQL, the addressable market for Databricks grows significantly beyond data engineers.
They've also made smaller bets in AI governance and observability, areas where they had product gaps relative to competitors. The pattern is classic horizontal platform acquisitions: buy the tools that belong in your ecosystem rather than letting them grow into independent category leaders who might compete with you.
Acqui-hires haven't disappeared, they've evolved
The pure acqui-hire, where you buy a company primarily for the team and shut down the product, is largely gone. But a variant has emerged: talent-forward acquisitions where the product is kept running but the team is the primary asset.
Google and Meta have both done these in early 2026, acquiring three-to-six person ML research teams building specialized models. The prices are typically $5-15 million, too small to generate headlines. These look like acqui-hires but the teams often retain more product autonomy than in classic acqui-hires. The acquirer wants the research direction, not just the people, and keeps the small team working on its problem with more resources.
This is particularly common in specialized model categories: code generation for specific languages, domain-specific retrieval models, efficient inference techniques. Labs and tech companies would rather buy a team that's already working on the problem than start from scratch.
Vertical AI getting acquired by incumbents
The pattern that's generating the most interesting deals is vertical AI companies getting acquired by established domain players rather than by tech companies.
The legal tech sector has seen several of these: established legal software companies (Thomson Reuters, LexisNexis, Wolters Kluwer) acquiring AI-native startups in contract intelligence, legal research, and document automation. The prices are typically $50-200 million for companies with $5-20 million ARR. The acquirers aren't buying for the AI technology per se; they're buying customer relationships and distribution in a vertical where they already have trust.
Similar patterns in healthcare: Epic Systems, Veeva, and others acquiring AI startups that have won early adopters in their customer base. The acquirer brings distribution to customers the AI startup couldn't have reached alone. The AI startup brings modern product thinking that the incumbent wouldn't have built fast enough internally.
These deals often work out better than pure tech-company acquisitions because the strategic fit is cleaner. The AI company gets customers immediately. The incumbent gets product credibility with a skeptical buyer base that trusts the incumbent's brand.
The Gemini 3 factor and big model economics
One dynamic accelerating acquisitions in May 2026 specifically: the release of Gemini 3 on May 17th has raised the baseline capability expectation for AI applications. When the underlying foundation models improve significantly, applications that were competitive three months ago can look thin today.
This creates acquisition pressure in both directions. Companies that built substantial differentiation above the model layer (proprietary data, workflow integrations, customer trust) become more attractive because their moat is clearly non-model. Companies that built primarily on model capability differentiation become less attractive because the differentiation eroded.
Buyers looking at targets right now are specifically asking: "Is this company's value in something that improves as base models improve, or something that gets commoditized?" The answers to that question are driving a lot of the current valuation negotiation.
Valuations in mid-2026
The revenue multiples for AI acquisitions have compressed significantly from 2024 peaks:
Strong ARR with high retention (120%+ NRR): 12-18x ARR. Reserved for the best businesses with clear defensibility.
Solid ARR with average retention (90-110% NRR): 6-10x ARR. The majority of deals in the current market.
Pre-revenue or early revenue with strong team: $3-15 million for talent-forward acquisitions. The pure technology bet has mostly left the market.
Distressed assets: Companies that raised at inflated 2024 valuations and failed to grow into them are being acquired at steep discounts, sometimes at or below the last funding round valuation. This segment is more active than the market discussion acknowledges.
The acquirers with the most disciplined approach are being transparent about what they'll pay for: recurring revenue, customer relationships, and specific technical capabilities. Teams without one of these three are finding it much harder to attract strategic buyers at meaningful valuations.
Who's likely to be acquired next
A few categories where acquisitions are probable in the next 6-12 months:
AI agents for regulated industries. Finance, healthcare, and legal AI startups with regulatory compliance baked in are hard to build from scratch. Established players in those verticals will keep buying rather than building.
Specialized retrieval and knowledge management. Enterprise knowledge management has been resistant to AI disruption because of data complexity and security requirements. Startups that have cracked enterprise data retrieval for specific industries are acquisition targets for productivity platforms.
Voice AI for customer interactions. The voice AI category is consolidating. After several years of distinct startups (call center AI, voice IVR replacement, voice synthesis), the use cases are starting to merge and the tools are getting acquired into contact center software platforms.
AI evaluation and testing infrastructure. As enterprises deploy more AI, they need evaluation tooling to verify outputs. This is a narrow but important category that several enterprise software companies will want to own rather than partner with indefinitely.
What's less likely to get acquired despite the volume of deals being done: general-purpose AI assistants competing directly with ChatGPT and Claude. The market has made clear that betting against OpenAI and Anthropic on pure model quality is a losing position, and acquirers aren't interested in companies whose primary differentiation is a frontend on top of someone else's API.