Why This Matters
AI acquisitions are reshaping the technology industry’s structure more rapidly than any M&A cycle since the dot-com era. But unlike the 1990s, where acquisitions were primarily about revenue and market share, AI M&A is about three scarce resources: talent, compute, and data. The most valuable assets in AI — researchers who understand transformer architectures, engineers who can train models at scale, and datasets that provide competitive advantages — cannot be built organically fast enough to keep pace with the industry’s velocity. Acquisitions are how companies buy time.
The emergence of the “acqui-hire” as the dominant deal structure adds a layer of complexity. Big Tech companies have learned to hire entire teams, license their IP, and leave the corporate shell behind — structuring deals that function as acquisitions but avoid the formal merger review process. The FTC has flagged this practice as potentially anticompetitive, but enforcement has lagged behind the deal pace. Whether regulators can catch up will determine whether the AI industry consolidates around a few dominant players or maintains a competitive ecosystem.
2025 broke records for AI M&A: 33 major deals totaling over $157 billion in disclosed value. The deals span outright acquisitions, acqui-hires, strategic stakes, and infrastructure partnerships. 2026 is on pace to match or exceed it.
The most dramatic deal of the era — OpenAI’s attempted $3 billion acquisition of Windsurf — collapsed when Microsoft refused to waive its IP rights, leading to a 72-hour scramble where Google grabbed the talent and Cognition bought the product.
JustSaid tracks acquisitions from SEC 8-K filings, antitrust filings, press releases, and regulatory disclosures.
The Windsurf Saga
The defining acquisition drama of 2025. Here’s what happened:
- OpenAI pursues Cursor (Anysphere) — OpenAI approached Anysphere in 2024 and again in early 2025 about acquiring their AI coding editor Cursor. Anysphere declined; they were growing too fast to sell.
- OpenAI pivots to Windsurf ($3B) — OpenAI entered talks to acquire Windsurf (formerly Codeium) for ~$3 billion in May 2025.
- Microsoft blocks the deal — Under Microsoft’s partnership terms, it holds rights to all IP that OpenAI acquires. OpenAI asked Satya Nadella for a carve-out to wall off Windsurf’s IP. Nadella refused.
- Google grabs the talent ($2.4B) — Google DeepMind hired Windsurf CEO Varun Mohan, co-founder Douglas Chen, and top engineers, plus licensed core technology. Reported value: $2.4 billion.
- Cognition buys the product ($250M) — Within 72 hours of Google’s move, Cognition AI (makers of Devin) acquired Windsurf’s codebase, brand, enterprise customers, and ~210 remaining employees for $250 million.
Meanwhile, Cursor keeps winning. Anysphere (Cursor) hit $2B ARR and a $60B valuation in 2026 — vindicating its decision to stay independent.
Major Deals: Full List
Mega-Deals ($5B+)
| Date | Acquirer | Target | Value | Type | Status |
|---|---|---|---|---|---|
| Mar 2025 | Wiz | $32B | Acquisition | Closed Mar 2026 | |
| Jun 2025 | Meta | Scale AI (49%) | $14.3B | Strategic stake | Closed |
| May 2025 | Salesforce | Informatica | $8B | Acquisition | Closed early 2026 |
| Dec 2025 | IBM | Confluent | $11B | Acquisition | Closed |
| Feb 2025 | IBM | HashiCorp | $6.4B | Acquisition | Closed |
Large Deals ($1B–$5B)
| Date | Acquirer | Target | Value | Type | Status |
|---|---|---|---|---|---|
| Jul 2025 | Windsurf (talent) | $2.4B | Acqui-hire/license | Closed | |
| May 2025 | OpenAI | Windsurf | $3B | Acquisition | Collapsed |
| Jul 2025 | Databricks | Neon | $1B | Acquisition | Closed |
| 2025 | Databricks | Tecton | ~$1B | Acquisition | Closed |
Notable Deals ($100M–$1B)
| Date | Acquirer | Target | Value | Type | Status |
|---|---|---|---|---|---|
| Mar 2024 | Microsoft | Inflection AI (talent) | $650M | Acqui-hire | Closed |
| Jun 2024 | Amazon | Adept AI (talent) | ~$500M | Acqui-hire | Closed |
| Aug 2024 | Character.ai (talent) | $2.7B | Acqui-hire/license | Under FTC review | |
| Jul 2025 | Cognition | Windsurf (product) | $250M | Asset acquisition | Closed |
| 2025 | Anthropic | Bun | Undisclosed | Acquisition | Closed |
| 2025 | Salesforce | Convergence.ai | Undisclosed | Acquisition | Closed Jun 2025 |
| 2026 | Salesforce | Qualified | Undisclosed | Acquisition | Closed Apr 2026 |
Acqui-Hire Pattern
The “acqui-hire” has become the defining deal structure of the AI era. Big Tech hires the founders and key talent, licenses the IP, and leaves the corporate shell behind — avoiding formal antitrust review. The FTC opened investigations in 2025 and issued a staff report calling these “pseudo-acquisitions” that constitute unfair competition.
| Deal | Structure | What the acquirer got | What happened to the company |
|---|---|---|---|
| Microsoft ← Inflection | $650M licensing deal | CEO Mustafa Suleyman + core team | Shell company remains |
| Amazon ← Adept | Licensing + hiring | Founders + ~66% of employees | Shell company remains |
| Google ← Character.ai | $2.7B license | Co-founder Noam Shazeer + researchers | Continues under new leadership |
| Google ← Windsurf | $2.4B license | CEO + co-founder + top engineers | Product sold to Cognition |
Infrastructure Mega-Partnerships
Not technically acquisitions, but the scale makes them acquisition-equivalent:
| Date | Buyer | Seller | Value | What |
|---|---|---|---|---|
| Jan 2025 | OpenAI/Stargate | Oracle | ~$300B | Multi-year cloud + power capacity |
| 2025 | Microsoft | Constellation Energy | ~$16B | Three Mile Island nuclear restart, 20-year PPA |
| 2025 | Amazon | Talen Energy | ~$10B | Susquehanna nuclear, long-term PPA |
| 2025 | Kairos Power | Undisclosed | 7 small modular reactors by 2035 |
Salesforce’s Acquisition Spree
Salesforce was the most prolific acquirer by deal count in 2025, buying 12+ companies for over $10 billion to build its Agentforce AI agent platform. The strategy: assemble a full-stack AI agent platform spanning data, process intelligence, autonomous agents, and global AI talent.
Key deals: Informatica ($8B), Convergence.ai, Qualified, and others.
What’s Next
Companies most likely to be acquired or acqui-hired in 2026:
- Cohere — Struggling to compete with OpenAI/Anthropic/Google; enterprise pivot may not be enough
- Stability AI — Repeated restructuring; valuable IP, limited runway
- AI21 Labs — Strong tech, limited distribution
- Runway — Video generation leader, strategic target for media/cloud companies
Companies most likely to IPO in 2026:
- Databricks — $134B valuation, $4.8B ARR, 55% growth
- Anthropic — Reportedly discussing 2026 listing
- OpenAI — Converting to for-profit, IPO discussions underway
What We’re Tracking
JustSaid monitors AI M&A activity through systematic tracking of regulatory filings and public disclosures.
SEC 8-K filings. Every material acquisition by a public company must be disclosed via an 8-K filing with the Securities and Exchange Commission. The pipeline monitors EDGAR for 8-K filings from the 50 largest technology companies, flagging any filing that references AI, machine learning, or related terms. These filings provide authoritative data on deal terms, valuations, and structure.
Hart-Scott-Rodino (HSR) antitrust filings. Acquisitions above the HSR threshold (currently $119.5 million) must be reported to the FTC and DOJ for antitrust review. The pipeline tracks HSR waiting periods and second requests, which signal regulatory scrutiny. The FTC’s investigation of the Character.ai acqui-hire was first signaled through an extended HSR review period.
Press releases and earnings calls. Smaller deals and acqui-hires that fall below the HSR threshold are often disclosed only through press releases or management commentary during earnings calls. The pipeline monitors press release wires and earnings call transcripts from major AI companies for acquisition-related language.
Talent movement patterns. Acqui-hires are sometimes visible through LinkedIn data before they are formally announced — when a cluster of employees from a startup simultaneously update their profiles to show a Big Tech employer, that pattern often precedes the public disclosure of a structured deal.
Outlook
AI M&A in the second half of 2026 will be shaped by three forces.
Regulatory pressure will increase but not stop deals. The FTC’s staff report on AI acqui-hires and the EU’s merger control investigations signal growing regulatory attention. However, the acqui-hire structure has proven difficult to challenge legally — courts have been reluctant to treat hiring decisions as mergers, even when accompanied by multi-billion-dollar licensing payments. Expect more scrutiny but continued deal flow.
Vertical integration will accelerate. The next wave of AI acquisitions will focus on vertical integration — AI companies acquiring domain-specific data, distribution channels, and customer relationships. An AI lab acquiring a healthcare data company, a legal research platform, or a financial data provider would gain training data and distribution simultaneously. These deals will be smaller in dollar terms but potentially more strategically significant than the talent-focused acqui-hires of 2024-2025.
The IPO window will thin the acquisition target list. If Databricks, Anthropic, and OpenAI go public in 2026-2027, they move from potential acquisition targets to potential acquirers. The pipeline of independent AI companies available for acquisition will shrink as the most valuable ones either IPO or are absorbed by Big Tech. Companies that remain independent and private — particularly those with strong technology but limited distribution — become increasingly attractive targets.
Frequently Asked Questions
Why do AI acqui-hires cost billions when normal hiring is free? The premium reflects three things: speed (hiring a 100-person team takes months; acquiring one takes days), context (the team brings institutional knowledge, working relationships, and in-progress research), and competition (multiple Big Tech companies bidding for the same teams drives prices up). The licensing fees that accompany acqui-hires also reflect the value of the target company’s IP — training data, model weights, and proprietary techniques that the acquirer wants access to.
Is the FTC likely to block AI acqui-hires? The FTC has signaled concern but faces legal challenges. Traditional merger law requires a formal acquisition of stock or assets, and acqui-hires are structured to avoid both. The FTC’s 2025 staff report called these deals “pseudo-acquisitions” but stopped short of enforcement action. The agency could potentially challenge acqui-hires under Section 5 of the FTC Act (unfair methods of competition), but this theory has not been tested in court.
Which AI companies are most likely to be acquired next? Companies with strong technology but limited distribution or runway are the most likely targets. Cohere (enterprise AI platform struggling against larger competitors), Stability AI (valuable image generation IP, organizational turmoil), AI21 Labs (strong language model technology, limited market presence), and Runway (leading video generation, strategic value for media companies) are among the most frequently mentioned potential targets. Any of these could be acquired outright or acqui-hired by a hyperscaler within the next 12 months.
How do AI acquisitions affect the competitive landscape? Each major acqui-hire reduces the number of independent AI companies and concentrates talent in Big Tech. The Microsoft-Inflection, Amazon-Adept, Google-Character.ai, and Google-Windsurf deals collectively moved over 500 researchers and engineers from independent startups into three companies (Microsoft, Amazon, Google). This concentration effect is what the FTC is most concerned about — not any individual deal, but the cumulative impact of a pattern that is hollowing out the independent AI ecosystem.
Tracked from SEC 8-K filings, EDGAR, antitrust filings, press releases, and earnings calls. Last updated: May 2026.