IP Value: Assessing Innovation Defensibility vs. Integration Expertise

Proprietary IP companies exit at 2–3× the multiples of integration-only companies in comparable markets. That spread has been consistent across European tech M&A for a decade. It's also the spread that gets obscured most frequently in pitch decks, where integration expertise is routinely described as proprietary technology.

The distinction matters for pricing. If you're underwriting a company as an IP business and it's actually an integration business, you're paying the wrong multiple for the wrong risk profile.

What actually constitutes a moat

Four things constitute genuine defensible IP. A novel algorithm or methodology that achieves outcomes competitors can't replicate without rebuilding from scratch. Proprietary training data accumulated over time that isn't available for purchase. A regulatory position, certification, approval, or compliance infrastructure, that creates structural barriers to new entrants. And accumulated domain expertise embedded in the product through years of iteration that can't be hired or bought quickly.

Integration expertise, being very good at connecting existing systems, excellent implementations of third-party APIs, bespoke customer workflow automation, is a real business. It generates real revenue. But it's not a moat. A well-resourced competitor can replicate it. That limits both the multiple and the exit optionality.

The AI complication

Foundation models have made a significant portion of what was previously 'proprietary AI' obsolete. Companies that spent 2019–2022 building custom NLP models, classification systems, and generative pipelines now have those capabilities available off-the-shelf from OpenAI, Anthropic, and Google at a fraction of the cost.

This means the IP assessment for AI companies requires an additional question: has the proprietary capability been commoditised by foundation models? If the answer is yes, you're likely looking at an integration business regardless of how it was positioned two years ago.

Three diligence questions

• What would a well-funded competitor need to replicate your core technical capability, not your brand, not your customer relationships, but the actual technical output? If the answer is 'six months and a good engineering team,' that's integration. If the answer involves years of data accumulation or novel research, that's IP.
• Has any of your core IP been made redundant by foundation model advances in the last 24 months? A team that can answer this honestly, and describe what they've done in response, is a team that understands its own competitive position.
• What's the ownership structure of the IP? Patents, trade secrets, and data ownership all carry different legal and commercial implications. IP that lives in a founder's head and hasn't been formalised is a significant concentration and succession risk.

The 2–3× multiple premium for genuine IP is real. So is the gap between what companies claim to have and what they actually have. Closing that gap before you commit is the job.