The Equation Series
Menu

Industry study 18 of 24 · AI Research Companion

M&A, Acqui-hires and Big Tech Partnerships

~7,000 words

In most technology cycles the big incumbents buy the start-ups that threaten them. In AI they mostly did not, at least at first. Between 2019 and 2025 the most important deals between Big Tech and AI labs were minority investments bundled with cloud contracts (Microsoft–OpenAI, Amazon–Anthropic, Google–Anthropic), and a new structure, the licence-and-hire or "reverse acqui-hire", in which a giant pays a start-up a large licence fee, hires its founders and most of its researchers, and leaves the company itself behind. Microsoft did this with Inflection (March 2024), Amazon with Adept and Covariant (2024), Google with Character.AI (2024) and Windsurf (2025), Meta in a different form with a $14.3bn stake in Scale AI (2025), and Nvidia with Groq (~$20bn, December 2025). These deals were designed to be fast and to avoid the merger review a full acquisition would trigger, and from 2024 competition authorities in the US, UK and EU began to treat them as mergers in substance. Then, from 2025, conventional M&A came back on a scale the industry had not seen: Google bought Wiz for $32bn, IBM bought Confluent for ~$11bn, SpaceX absorbed xAI and agreed to buy Cursor for $60bn, and Nvidia agreed to buy Hugging Face for ~$12.9bn (2026, reported). This section explains the deal types, the economics for founders, investors and employees, the antitrust response, and how an experienced dealmaker reads an AI partnership.

What the study covers

  1. 1Overview: why AI deals look different
  2. 2The deal types
  3. 3The licence-and-hire wave (2024–2025)
  4. 4Big Tech–lab partnerships
  5. 5Conventional M&A in AI
  6. 6Antitrust and foreign-investment scrutiny
  7. 7Integration, failure and what deals teach
  8. 8Key numbers
  9. 9Landmark companies and cases
  10. 10Pitfalls
  11. 11How an expert thinks about it
  12. 12Glossary