Equity

Reverse Acqui-Hires: Who Gets Paid, and What's Left to Value Afterward

In a reverse acqui-hire, the buyer licenses the tech and hires the team. Who actually gets paid, why employee options miss out, and how to value what's left.

By 409.AI Team - 2026-09-01

# Reverse Acqui-Hires: Who Gets Paid, and What's Left to Value Afterward

On July 11, 2025, Google agreed to pay $2.4 billion for a non-exclusive license to Windsurf's AI coding technology and for the right to hire its two co-founders and much of its research team, [Reuters reported](https://www.reuters.com/business/google-hires-windsurf-ceo-researchers-advance-ai-ambitions-2025-07-11/). It did not buy the company. Windsurf was still there the next morning: same entity, same charter, same cap table, roughly 200 of its 250 employees still on payroll, and no acquirer.

Three days later, Cognition [bought what remained](https://techcrunch.com/2025/07/14/cognition-maker-of-the-ai-coding-agent-devin-acquires-windsurf/).

Those three days are the part worth studying, because that is the window in which a reverse acqui-hire actually lands on a company. The value walks out as people and a license. The legal entity stays. And every question a normal acquisition answers cleanly, including who gets paid, whose equity accelerates, and what the common stock is worth on Monday, has to be answered from scratch.

The structure, and why it keeps happening

The pattern is common enough now that three US senators have put it in writing. In a [February 4, 2026 letter](https://www.warren.senate.gov/imo/media/doc/final_-_warren_wyden_blumenthal_letter_to_the_department_of_justice_and_the_federal_trade_commission_on_big_tech_reverse_acqui-hires.pdf) to the DOJ Antitrust Division and the FTC, Senators Warren, Wyden and Blumenthal described "reverse acqui-hiring" as Big Tech firms "swooping in to hire star talent and license technology ... discarding the rest by the wayside."

The letter names three deals. Meta paid more than $14 billion for a 49% stake in Scale AI in June 2025 and hired its CEO. Google did the Windsurf deal a month later. In December 2025, NVIDIA announced a $20 billion non-exclusive license for Groq's inference chip technology, and Groq's founder and president went to NVIDIA. The senators argue all three function as de facto mergers, and the FTC chair had already said in January 2026 that the agency would investigate acquihires.

The antitrust fight is not what this article is about. The structural fact underneath it is. A buyer that wants a team and a technology can get both without a merger agreement, without buying a single share, and therefore without triggering any of the machinery that a stock or asset purchase sets in motion. The choice gets made for regulatory reasons, but it lands on the cap table all the same.

All of it is happening inside the busiest startup acquisition market on record. Crunchbase counted [$119.8 billion spent on private, venture-backed U.S. companies](https://news.crunchbase.com/ma/2026-mergers-acquisitions-record-cursor-spcx/) by late June 2026, already on pace to beat 2025's record. Plenty of that is conventional M&A. A growing share is not.

Who gets paid, and in what order

The Windsurf numbers are the clearest public illustration. TechCrunch [reported](https://techcrunch.com/2025/08/01/more-details-emerge-on-how-windsurfs-vcs-and-founders-got-paid-from-the-google-deal/) that the $2.4 billion split into roughly $1.2 billion flowing to investors and founders, and roughly $1.2 billion in compensation packages for the approximately 40 people Google hired. Investors left more than $100 million of capital behind in the company. Employees who had joined in the prior year got nothing from the Google transaction, and about 200 people were not hired at all.

That outcome is not a matter of anyone behaving badly. It falls out of the structure, and it falls out the same way every time.

Money paid to the corporation for a license is corporate cash. It reaches shareholders only if the board moves it, and whichever route the board takes, a dividend or a wind-down, the preferred stock's rights in the charter typically come ahead of common. Money paid to individuals for going to work somewhere else is compensation for future services. It is ordinary income to those people, it comes from the acquirer's payroll, and it is not consideration for anybody's shares. Neither pot is an equity payout, so neither pot flows to option holders as option holders.

Put numbers on it. Say a company raised $180 million across three rounds, all non-participating preferred with 1x preferences, and holds $40 million of cash. A buyer pays $300 million: $150 million as a license fee to the company, $150 million in pay packages for the 35 people it hires. The company now sits on $190 million against $180 million of liquidation preference. Wind it up tomorrow and distribute everything, and preferred takes its $180 million while the $10 million left over spreads across 12 million common shares at about $0.83 apiece. The 4 million outstanding options carry a $3.20 strike from the last 409A. They are worth nothing, because $0.83 is not $3.20, and $300 million changing hands did not move them an inch. This is the same [liquidation preference waterfall](https://409.ai/articles/liquidation-preferences-waterfall-common-stock-exit) that governs an ordinary exit, applied to a transaction most employees never think of as an exit at all.

Nothing in the plan documents fires

Founders often assume acceleration protects the team here. Usually it doesn't.

Single and double-trigger acceleration hang on a defined term, and that term is almost always some version of a merger, a sale of all or substantially all assets, or an acquisition of a majority of voting power. A non-exclusive license plus a batch of resignations typically satisfies none of them. Which is precisely why exclusivity is not a drafting detail: an exclusive license of the company's core technology starts to look like a sale of substantially all its assets, and a non-exclusive one is much easier to argue is not. The same definitional question decides whether [double-trigger vesting](https://409.ai/articles/double-trigger-rsus-ipo-taxation-409a) means anything to your people.

Section 280G lands in the same place. The golden parachute rules apply to payments contingent on a change in the ownership or control of the corporation, or in the ownership of a substantial portion of its assets. Where a deal is built specifically to avoid being any of those, the [280G analysis](https://409.ai/articles/section-280g-golden-parachute-startup-exit-shareholder-vote) that would dominate a normal exit often has nothing to attach to. Good news for the people being hired, cold comfort for everyone else.

The one clock that does start is the least helpful one. Employees who are not hired are frequently laid off within weeks, and termination starts the post-termination exercise window, typically 90 days. For incentive stock options there is a matching statutory rule: [Section 422(a)(2)](https://www.law.cornell.edu/uscode/text/26/422) requires that the holder have been an employee at all times from grant until the day three months before exercise. Miss that and the option is no longer an ISO. So a departing employee gets three months to decide whether to write a check for shares in a company whose team just left, at a strike price set when the team was still there. [Extending the exercise window](https://409.ai/articles/extending-post-termination-exercise-window-iso-nso-409a) is a real option for the board, and it carries that ISO consequence with it.

Then someone has to value what's left

This is the part that gets deferred and shouldn't be.

A reverse acqui-hire is a material event by any reading. Treasury's 409A regulations presume a valuation is reasonable only if it is no more than 12 months old and does not "fail to reflect information available after the date of the calculation that may materially affect the value" of the company ([26 CFR 1.409A-1(b)(5)(iv)(B)](https://www.law.cornell.edu/cfr/text/26/1.409A-1)). Losing the founding team, licensing the core technology, and taking a nine-figure payment onto the balance sheet clears that bar three times over. Any grant issued off the old report is exposed, which is the whole reason [material events reset the 409A clock](https://409.ai/articles/409a-valuation-frequency-how-often-should-you-get-one) rather than waiting for the anniversary.

The harder problem is method. Most venture-backed 409As lean on a backsolve: take the price of the most recent preferred round, solve for the total equity value that makes that price consistent, then allocate across the classes. After a reverse acqui-hire, that round no longer describes the company. Nothing about a $900 million post-money from eighteen months ago survives the departure of the people who justified it, so the input that usually anchors [the allocation](https://409.ai/articles/409a-allocation-methods-opm-pwerm-backsolve) is gone.

What replaces it depends on what the company actually is now. The 409A regulations list the factors an appraiser is expected to weigh, starting with the value of tangible and intangible assets and the present value of anticipated future cash flows. For a company that is mostly a bank balance, a non-exclusive license it still owns, and a remaining team deciding what to do next, the center of gravity moves toward [what the assets are worth](https://409.ai/articles/asset-based-approach-409a-valuation) and away from a forecast nobody believes.

Run the earlier example through it. Equity value of roughly $190 million, almost all of it cash, with $180 million of preference sitting ahead of common. Common is not worthless, because the residual is real and because whatever the remaining team builds could still pay off, but its value is thin and largely option-like, and it carries a heavy marketability discount besides. A common FMV that was $3.20 can plausibly come back at a fraction of that. For the people still there, that is the good news and the bad news in one number: their old options stay stranded, and their new grants get a strike price worth having. Whether the report supports either conclusion is exactly what an [independent 409A](https://409.ai/products/409a) is for.

The accounting follows

Under ASC 718, compensation cost for awards that never vest because a service condition goes unmet gets reversed. A large layoff after a reverse acqui-hire therefore claws back a meaningful chunk of previously recognized stock compensation expense for unvested awards. Vested options that simply expire unexercised are different: that cost stays on the books, because the employee earned the award. Fresh retention grants at a lower strike bring in fresh, smaller expense. None of it is complicated, but it does mean the [stock compensation line](https://409.ai/articles/asc-718-stock-based-compensation-startup-guide) in the next set of financials will move in ways that need explaining to an auditor.

What to take from it

A reverse acqui-hire does not destroy a cap table. It strands one. The enterprise value converts into two things that employee equity cannot reach: cash inside a company where preferred stock stands first in line, and pay packages that belong to the individuals who receive them. Options are claims on residual equity, and this structure is very good at leaving almost no residual.

Cognition's response to the Windsurf situation is worth noting for exactly that reason. When it bought what Google left, it said [100% of Windsurf employees would participate financially in the deal and have vesting cliffs waived](https://techcrunch.com/2025/07/14/cognition-maker-of-the-ai-coding-agent-devin-acquires-windsurf/) for their work to date. That was a choice, not a mechanism. The mechanism, left alone, pays the preference stack and the people getting hired.

If a deal like this is on your table, order the new 409A before you promise anyone a retention grant, and read the change-in-control definition in your own plan documents before you tell your team what accelerates. Both answers will be less generous than everyone assumes, and both are much easier to explain in advance than in an all-hands afterward.

Related valuation reports