SoundHound AI closed its acquisition of LivePerson on Thursday, two days after LivePerson shareholders approved it. The headline price was $43 million — a 22% premium to LivePerson’s 30-day weighted-average share price, and a slight premium to a market capitalisation that had fallen to roughly $40 million.
The enterprise value of the transaction is about $250 million. The difference between those two numbers is the deal.
LivePerson was not a company with no assets. It was a company whose assets were spoken for. Its digital messaging platform reaches 25 of the Fortune 100, and the combined patent portfolio now runs to more than 750 grants. What sat on top of that was a stack of secured notes large enough that the equity had been compressed to a rounding error. SoundHound issued approximately 36.9 million shares to retire those notes, and the combined company came out the other side debt-free.
The equity was the residual, not the price
This is a structure worth recognising because it keeps appearing. In a distressed capital structure, the acquirer is not really negotiating with shareholders — it is negotiating with lenders, and the shareholder vote is a formality attached to whatever is left. The $43 million is what remained after the noteholders were made whole in stock. Reported as a purchase price, it makes the transaction sound like a bargain-bin pickup. Reported as $250 million of enterprise value paid mostly in equity dilution, it looks like what it is: a sizeable bet funded by SoundHound’s share price.
The same asymmetry showed up in Félix’s $200 million raise, where only $87 million was the company itself. The headline number and the number that determines who bears risk are increasingly different numbers.
Our take: When you see a strategic acquisition priced below $50 million, look immediately at the debt line before concluding anyone got a deal. The interesting question here is not whether SoundHound overpaid — on enterprise value it plainly did not steal it — but whether paying in its own shares while its own shares carry a high AI multiple is a strength or a dependency. It is both, and which one it turns out to be depends entirely on the next two earnings reports.
What has to happen now
SoundHound is guiding to $350–400 million of combined 2027 revenue, with management suggesting it could exceed $500 million if cross-selling works. That spread — a potential 25% or more of forecast revenue riding on cross-sell — is where the thesis lives.
The logic is clean enough: SoundHound has proprietary voice agentic AI and a customer base weighted toward automotive and restaurants; LivePerson has enterprise digital messaging deployed at scale but has been losing ground to newer agentic entrants. Voice plus messaging under one contract is a genuinely better product than either alone.
The execution is the hard part, and it always is. LivePerson’s enterprise accounts did not stop renewing because the messaging was bad. They stopped because the roadmap looked slow, and a change of owner does not by itself change a roadmap.
What to watch
- Net revenue retention at the acquired accounts. The Fortune 100 logo count is the asset. If it erodes through integration, the $250 million bought a patent portfolio.
- Dilution. 36.9 million shares is the real consideration. Track it against share count, not against the $43 million.
- Whether the first cross-sell wins are named. Companies announce these when they happen. Silence through Q1 2027 would be informative.
- Competitive response. Every major conversational-AI vendor now has both modalities on the roadmap. The window on “voice plus messaging” as a differentiator is not wide.
Two days from shareholder approval to close is fast. That is what happens when the lenders have already agreed and the equity holders have very little to argue about.