Three AI Questions Buyers are Asking
Artificial intelligence (AI) has largely evolved from an experimental sandbox into a core, operational technology across global businesses and daily consumer products. Enterprises are embedding AI directly into daily workflows, core operating models, and customer-facing platforms. And with that evolution AI has become a fundamental value driver for technology companies. AI-enabled tech companies are receiving much higher M&A deal valuations than traditional SaaS companies.
Buyers of AI-enabled tech companies have evolved too. They no longer ask generic questions like "do you have AI?" Rather, those questions are more specific and insightful. We spoke to Dr. Ivan Ruzic, Corum's resident AI authority about these questions. He cited three that buyers are typically asking.
1. Is your AI something you own, or something you’re renting?
Buyers are putting a premium on tech companies whose core operations are built around proprietary AI-native architectures that rely on proprietary domain data rather than those providing thin wrappers on top of generic foundational models. Ruzic notes that providing a thin wrapper around a third-party API has a margin problem baked in. That's because its cost structure is completely controlled by an external provider, while its pricing is constantly pressured by open competition and rapid technological commoditization. Ruzic adds that if the buyer can’t tell what is proprietary and what is simply an outbound API call, they’ll price the whole thing as a "rental". And that puts a drag on the company's valuation. If the company's core AI functionality is seen as rented, that is, its AI functionality is through a third-party API wrapper, its valuation suffers an immediate compression or deal repricing.
2. What’s your headcount conversion story?
Replacing human headcount with agentic workflows is an increasingly significant aspect of AI-enabled software company acquisitions. In fact, Ruzic says that every buyer’s model now carries a line called "headcount conversion" ‒ the cost and timeline to replace roles with agentic workflows. If a seller can't answer the headcount conversion question, it's likely the buyer will "fill in their own worst-case number and price you against it," says Ruzic.
The conundrum here is that deep headcount replacements via agentic workflows do not necessarily mean better returns. Global research firm Gartner found that companies that cut deepest posted virtually the same returns as the ones that cut least, and some of the lighter cutters did better. Ruzic stresses that layoffs create budget room, they don’t create return.
3. Where did the savings actually go?
Tech companies actively turn to AI to reduce costs. But if those costs are indeed reduced, buyers want to know where those savings went. Did they go toward increased margins or were they plowed into something that does not contribute to the bottom line? Ruzic points to the classic case of SAP which cut costs through a hiring and travel freeze, and then used the savings to fund their AI buildout. "That’s not savings reaching the bottom line," says Ruzic. "That’s your operating budget being recycled into somebody else’s capital expenditure." Ruzic points out that a sophisticated buyer asks directly, "Did the headcount savings widen your margin, or did it just fund the compute bill?" Savings that are applied to buildouts rather than the bottom line are what Ruzic calls the savings-to-capex loop. It’s the fastest way, he says, to lose a repricing argument in due diligence.
What this means for you
Ruzic notes that these three questions together constitute the checklist buyers of AI-enabled tech companies are running today. If you're the founder of a tech company that has clear convincing answers for these three questions, buyers are willing to pay very high premiums for your company. Ruzic points to numbers that bear this out. "In the first quarter of this year, he says, “AI-enabled software companies traded at 8.4 times revenue. The SaaS median ‒ the non-AI legacy comparison ‒ closed at about 3.2 times revenue. That 2.6 times spread is the widest capability-driven valuation split in tech M&A since the cloud migration wave of 2012 to 2015."
Ruzic also underscores the fact that there is a closing date to these high premiums. It may last into through 2027 or even 2028, he says. However, for tech companies that simply provide a product wrapper around someone else’s foundation model, do not base it on proprietary data, and have shallow workflow depth, that premium is already compressing toward 2.5-to-3.5 times revenue.
No matter where your tech company fits, it is important to understand that buyers are already asking these questions. "The thing left to decide, "says Ruzic "is whether the answers in the data room are yours or theirs."