This article represents the perspectives of two co-founders with very different vantage points on M&A transactions. Tony has spent decades in structured finance, assessing assets and designing valuation frameworks from the buy-side. Mark has spent 30 years building businesses and selling them, understanding the seller's challenge of articulating value to buyers. When it comes to valuing AI capability in an M&A context, these two perspectives diverge more than they align — and that divergence reveals the real challenge.
I have been on the buy-side of countless transactions. The mechanics are straightforward: identify the target, assess value, identify risks, adjust the offer accordingly. AI capability complicates this process in ways that previous technology acquisitions did not.
When a PE buyer or strategic acquirer looks at an AI-enabled target, they are essentially asking four questions:
First: Is the AI capability defensible? Can we actually operate the AI systems post-acquisition, or are they dependent on individuals or external providers who will leave or become unaffordable?
We look at model governance documentation. Has the company documented how models are trained, validated, tested, and deployed? Can a new team inherit the capability? Or does it live entirely in the head of a single ML engineer?
We assess data dependencies. Are the datasets proprietary or can they be replicated? Are they clean and structured, or will we inherit a data quality problem? What happens if the external data sources disappear?
We evaluate vendor dependencies. If the AI system depends on a specific foundation model provider's API, what is our switching flexibility? If the provider changes pricing or availability, can we migrate to an alternative?
Second: What is the competitive moat created by AI? Does the AI capability make the company harder to compete with, or is it a replicable feature that any well-funded competitor could build?
A company that has used AI to build a proprietary dataset has a moat. A company that has used AI to improve customer churn prediction has built organizational capability. A company that has bought access to a commercial foundation model and wrapped it in a UI has not — the feature is replicable by any competitor with funding.
We conduct reference calls with customers. Do they stay because the AI capability is genuinely superior, or because they have switching costs? Would they leave for a marginally better AI product? Or would they require a substantial improvement to justify switching?
Third: What is the embedded talent risk? The AI capability might be strong, but if its creators leave post-acquisition, can it be maintained and improved?
We look at retention agreements. Are key ML engineers and data scientists locked in? What are the financial incentives for them to stay? What is our realistic timeframe for developing internal AI expertise if the current team leaves?
We assess team documentation and knowledge transfer. If key people leave, how much institutional knowledge leaves with them?
Fourth: How do we price this capability? This is where the measurement frameworks matter most.
We want to understand what portion of enterprise value should be attributed to AI capability specifically. Is it 10% of the EBITDA multiple? 20%? Is the AI driving a disproportionate share of gross margins or customer retention?
This is where most buyers and sellers diverge. Sellers want to claim that AI is worth a massive multiple premium. Buyers want to isolate the specific, measurable value attributable to AI vs. other factors (brand, market position, sales execution, etc.).
When we put a number on AI capability, we typically use three approaches in parallel:
Excess earnings attribution. If a company's EBITDA is £5 million, and we can isolate which portion is directly attributable to AI-driven efficiency, cost reduction, or revenue uplift, we can value that contribution using multi-period excess earnings (MPEE). If AI is driving £1 million of the £5 million EBITDA (through faster customer service, better churn prediction, or superior product personalisation), we can value that £1 million using a perpetuity or terminal value approach.
Relief from royalty. For proprietary AI models or unique algorithmic approaches, we apply relief-from-royalty (RFR) valuation. What would the company need to pay if it licensed an equivalent AI capability from a third party? If the answer is "they could not — there is no equivalent available," the RFR approach gives a premium value. If the answer is "they could license something similar for 3-5% of revenue," that sets an upper bound.
Comparable transactions. We look at recent exits in the same sector. Did similar companies with equivalent AI capabilities sell at higher multiples? By how much? This provides a market check on AI premium pricing.
| Approach | Strengths | Limitations |
|---|---|---|
| Excess earnings attribution | Ties value to actual business impact | Requires isolating AI contribution from other factors |
| Relief from royalty | Defensible for proprietary models | Hard to price if no market comparables exist |
| Comparable transactions | Market-based | Limited comparable AI-enabled exits |
The challenge is that these three approaches often yield different answers. When they diverge significantly, we dig deeper into the seller's claims about AI impact.
In my experience, the sellers typically overestimate AI contribution by 20-30%. A seller claims AI is worth a 25% EBITDA premium because it drives cost reduction and churn improvement. The buyer does the work and finds that AI contributes to 15% of incremental EBITDA. The valuation gap becomes a negotiation point.
There are specific signals that make a buyer discount AI-enabled acquisitions:
The most successful AI-enabled acquisitions I have been part of are those where the seller has done the hard work of documenting, measuring, and defending their AI claims before we engage.
From my 30 years of advising founders preparing businesses for PE exit, I can tell you this: buyers and sellers are having completely different conversations about AI capability, and that gap is costing sellers millions.
The first mistake is treating AI as a feature to mention rather than as an intangible asset to document. A founder might say: "We use AI to personalise recommendations." A buyer hears: "You use a commercial recommendation engine like every other company in your sector."
The second mistake is failing to quantify the AI impact with before/after metrics. A seller says: "Our AI improves customer retention." A buyer asks: "What was retention before? What is it now? How much of the improvement is attributable to AI vs. other factors (price change, marketing, competition)?" The seller often cannot answer.
The third mistake is waiting until due diligence to discuss AI capability. If the AI story is strong, it should be in the investment committee paper, in the process document, and in preliminary conversations. If it is mentioned for the first time in due diligence, it reads as an afterthought or a claim the seller is not confident about.
Here is the approach I recommend to founders 12-18 months from an anticipated exit.
Step 1: Document the AI capability stack (Months 1-3)
What AI systems does the company actually operate? For each system, document:
This documentation serves two purposes: it makes the capability real to a buyer, and it identifies what is actually working vs. what is aspirational.
Step 2: Establish baseline metrics and measure impact (Months 3-6)
For each AI system, establish a clear baseline: What was the metric before AI (customer churn, support cost, forecast accuracy, etc.)? What is it now? How much of the improvement is causally attributable to AI?
This requires discipline. A company might see churn improvement from 8% to 6% after deploying an AI churn prediction model. But churn might have also improved because:
A rigorous approach is to set up a randomised control trial or propensity-match comparison. If that is not feasible, at minimum separate out the factors you can control or observe. Then attribute the residual to AI.
A SaaS company deployed an AI-driven customer success system that proactively identifies churn risk. They measured: total churn improved from 8% to 6% (2% absolute improvement). But a new product launch accounted for 1% of improvement, and macro factors accounted for 0.5%. The AI system was genuinely responsible for 0.5% improvement. That 0.5% improvement, applied to their 10,000 customer base and £200 average annual value per customer, is worth £1 million annually. Using a 4x multiple (PE buyer assumption for customer-centric improvements), that is a £4 million value attribution to AI. That is a number you can defend.
Step 3: Build the intangible asset inventory (Months 6-9)
Catalogue all AI-driven intangible assets:
For each asset, describe:
Step 4: Apply consistent valuation methodologies (Months 9-12)
Use the same frameworks a PE buyer would use:
The goal is not to produce a precise number (valuation always has error bars). The goal is to show that you have thought about the problem rigorously and can articulate the basis for your AI value claims.
Step 5: Prepare the data room (Months 12-15)
Create a dedicated section in your data room for AI capability. Include:
This should be a coherent narrative that a buy-side team can consume in 2-3 days and understand the AI value argument completely.
Step 6: Rehearse buyer conversations (Months 15-18)
Anticipate the questions:
Prepare specific, evidence-based answers. Do not bluff.
The gap between buyer and seller perspectives on AI value is real. But it is bridgeable through a shared framework.
When a buyer asks "How much value does AI create?" and a seller responds "A lot," the conversation stalls. When a seller says "Our AI reduces customer acquisition cost from £800 to £600, creating £1.2 million in annual incremental value, which we value at £4.8 million using a conservative 4x multiple, supported by documented before-after metrics," the buyer can engage seriously.
The most successful AI valuations in recent M&A transactions have been those where buyer and seller shared the same measurement discipline.
Use this checklist before entering the process:
Capability Documentation
Impact Measurement
Risk Mitigation
Valuation Readiness
| Checklist Item | Seller Impact | Buyer Impact |
|---|---|---|
| Missing capability documentation | Red flag; reduces valuation 20-30% | Acquisition risk; requires more integration work |
| No impact measurement | Buyer discount for unproven claims | Cannot assess real vs. claimed value |
| Talent retention not locked | Key risk if team leaves post-acquisition | Liability if capability is lost |
| Undiscovered vendor lock-in | Discovered in due diligence; negotiated down | Must model switching cost into returns |
| Valuation framework not prepared | Buyer imposes their framework; seller gets shortchanged | Due diligence extends; deal timeline slips |
When a buyer and seller sit down to discuss AI capability, one of three outcomes occurs:
They agree on both the value of AI and the risk profile. A £5 million enterprise value is mutually acceptable. The deal closes at expectations.
The seller believes AI is worth more than the buyer. The seller believes AI creates £2 million in incremental value and should add a £6 million premium to the base valuation. The buyer believes it creates £1 million in incremental value and is only willing to add £3 million. The gap is negotiated.
The buyer questions whether AI value is real. The seller makes claims that the buyer cannot verify. The buyer applies a heavy discount or walks away. The seller is forced to accept a valuation that does not reflect their AI investment.
Which outcome you experience depends entirely on the quality of your preparation. Companies that have documented, measured, and de-risked their AI capability achieve narrow gaps between their valuation expectations and buyer offers. Companies that have treated AI as marketing ephemera face significant discounting.
For PE-backed portfolio companies with 12-18 months before an anticipated exit, the single highest-return activity is rigorous AI capability documentation and valuation preparation. The difference between a buyer who can clearly see your AI value and one who cannot is often £5-10 million in enterprise value.
That preparation is the responsibility of both the seller (to build and articulate the narrative) and the buyer (to assess it rigorously). When both sides do their work, the price of AI capability becomes defensible and fair.
Tony Hillier is Co-Founder of Opagio. He holds an MA from Balliol College, Oxford and an MBA with distinction. His career includes executive board roles at NM Rothschild & Sons and GEC Finance, and a non-executive directorship at Financial Security Assurance in New York, where he specialised in structured finance and asset-backed securities across global capital markets.
Mark Hillier is Co-Founder and CCO of Opagio. He has spent 30+ years advising businesses through growth and PE exit, with institutional clients including Legal & General, AEW UK Investment Management, and Salmon Harvester.
Estimate the value of your intangible assets using industry-standard methods like Relief from Royalty, MPEEM, and With & Without.
AI-enabled companies trade at 25-40% higher multiples than peers, but not all AI capability is created equal. This article examines what drives the AI valuation premium, how to assess whether it is justified, and what happens when the premium unravels.
Read more →
Mergers and acquisitions involving artificial intelligence companies fail at a 62% rate in terms of value creation. The root cause is not market misjudgment — it is due diligence frameworks designed for physical assets, applied to intangible-asset-heavy technology companies. Here is the anatomy of that failure, and a framework that fixes it.
Read more →
How the IFRS 13 and ASC 820 fair value hierarchy applies to intangible assets. Covers the three input levels, why most intangibles are Level 3, and the disclosure obligations that follow from the classification.
Read more →Get the latest insights on intangible asset growth and productivity delivered to your inbox.
Book a free consultation to see how the Opagio Growth Platform can help your business.