The acqui-hire is replacing the traditional startup exit in AI
Between March 2024 and January 2026, Google, Microsoft, Amazon, and Meta collectively spent more than $20 billion acquiring AI startup talent and technology — without a single one of those deals being structured as a traditional acquisition. Microsoft licensed Inflection AI's models for $620 million and hired roughly 70% of its staff. Google paid approximately $2.7 billion to license Character.AI's technology while its co-founders returned to DeepMind. Amazon licensed Adept's agent technology for around $25 million and hired 66% of its team, then separately paid Covariant AI $380 million to license its robotics models and hire a quarter of its staff, including all three co-founders. None of these were called acquisitions. All of them functioned like one.
This structure — pay to license the technology, hire the key people, leave a legal shell of the original company behind — has a name now: the reverse acqui-hire. And it has become the dominant exit path for AI startups that built something a hyperscaler wants, but don't want to trigger the merger review that a formal acquisition would require.
Why companies structure deals this way
A licensing-plus-hiring deal isn't legally an acquisition, and in most jurisdictions that means it doesn't automatically trigger the same antitrust review a straight buyout would. The Federal Trade Commission has taken notice: the Amazon-Adept deal drew formal FTC attention, and the German Federal Cartel Office and UK Competition and Markets Authority both scrutinized the Microsoft-Inflection deal, with the German regulator concluding it constituted a reportable concentration but declining jurisdiction due to insufficient local nexus. Regulators are catching on to the pattern, but enforcement is still playing catch-up to deal structure — which is exactly why the structure keeps getting used.
For the acquiring company, there's a second advantage beyond antitrust: they get the people and the technology without inheriting the acquired company's cap table, existing contracts, liabilities, or investor board seats. It's a cleaner transaction than a merger, executed faster, with fewer strings attached.
What happens to the company left behind
The fate of the original startup varies, and not always well. Covariant AI, after Amazon's $380 million deal and the departure of its three co-founders to Amazon Robotics, was described in a 2025 whistleblower complaint as a "zombie company" — an entity that continued to technically exist primarily to receive a second $20 million payment a year after the initial deal closed. Inflection AI shifted to a smaller enterprise-focused operation after roughly 70% of its staff left for Microsoft. Adept AI continued independently with its remaining 34% of staff, though its former CEO David Luan — who had gone on to lead Amazon's AGI Lab — announced his departure from Amazon in February 2026, underscoring that even the "winning" side of these deals doesn't guarantee long-term retention of the acquired talent.
What this means for founders raising money today
If you're building an AI startup with technology a hyperscaler might eventually want, the reverse acqui-hire is now a realistic, well-precedented exit path — and it changes how you should think about cap table structure and investor terms from the start. Early Inflection investors reportedly saw a 1.5x return; late-stage investors saw only 1.1x, a modest outcome for what was, at the time, one of the most heavily funded AI startups in the world. That gap matters: reverse acqui-hires typically pay out on terms structured to compensate the company and its investors for IP licensing and a talent-hiring waiver, not on terms that reflect a full acquisition premium. If your cap table assumes a traditional acquisition multiple, model what a licensing-plus-hire structure actually pays before you assume it's equivalent.
The regulatory clock is still running
Congressional scrutiny of these deals persists into 2026, and it's reasonable to expect the FTC or a comparable regulator eventually closes this loophole — probably by defining a hiring-plus-licensing threshold that automatically triggers merger review regardless of legal structure. Until that happens, expect the pace of reverse acqui-hires to continue, particularly as hyperscalers shift AI infrastructure spending toward consolidation rather than a "best of breed" software acquisition strategy. For founders, the practical move is to negotiate licensing and hiring-waiver terms as explicitly as you would negotiate an acquisition price, because in substance, that's exactly what they are.