Board Case Study · Public Acquirer (M&A)
The $640M acquisition closes Friday. The target's valuation rests on an AI model diligence couldn't open, and its training data, IP, and vendor dependency are all unverified.
The situation
Your company, a public acquirer, is three days from closing a $640M acquisition of an AI analytics startup. The deal thesis, and most of the premium, rests on the target's proprietary model and its claimed accuracy. The CEO has championed the deal publicly and wants to sign Friday.
Technical diligence hit a wall: the target wouldn't expose training-data sources, the model's IP provenance includes possible open-source and scraped data, and the product depends on a third-party foundation model under a license that may not survive a change of control. The accuracy claims were validated only on the target's own benchmark.
If the model's performance or IP is not what's represented, the goodwill impairs, the premium evaporates, and the board approved it. A delay risks the deal and signals doubt to the market; closing blind risks a writedown the board will have to explain.
“The whole valuation is the model, and we were never allowed to open it. Do we close Friday on representations, or hold the line?”
Chair, Transaction Committee
The decision on the table
The full case, scenario architecture, board materials, and facilitation notes, is shared with boards and partners on request.