Twenty-eight percent of global M&A activity in 2025 was AI-related, according to PitchBook data. Private equity firms are acquiring AI-enabled companies at record rates — and at record premiums. But the due diligence playbook for AI targets is fundamentally different from traditional technology acquisitions, and most PE firms are still applying industrial-era frameworks to assess intelligence-era assets.
The gap is expensive. A recent study from Bain & Company found that 62% of technology acquisitions fail to meet their financial targets, with poor technical due diligence cited as the primary cause. For AI-enabled targets, the failure rate is likely higher because AI introduces risks that traditional due diligence does not address: model degradation, training data provenance, algorithmic bias liability, and the AI washing problem.
Traditional technology due diligence focuses on code quality, architecture scalability, security posture, and technical debt. These remain important for AI targets, but they miss the dimensions that determine whether AI capability is genuinely valuable.
An AI system is not just code — it is code, data, models, training processes, deployment infrastructure, and the organisational capability to maintain all of these. Examining the code alone is like valuing a factory by inspecting the machinery while ignoring the raw materials, workforce skills, and supply chain.
AI due diligence requires four additional dimensions beyond traditional technology assessment: data asset evaluation, model quality and sustainability, AI talent depth and retention risk, and regulatory compliance posture. Missing any one of these creates material acquisition risk.
Data is the raw material of AI. Without adequate, clean, relevant data, AI models cannot be trained, maintained, or improved. Data due diligence must assess:
Data provenance and rights. Where does the training data come from? Does the company have clear legal rights to use it for AI training? Post-GDPR and under the EU AI Act, data provenance is a material compliance risk. Training on scraped web data, user-generated content, or third-party data without proper licences creates liability.
Data quality and coverage. Is the data representative of the target population? Are there systematic biases in collection, labelling, or sampling? Poor data quality produces poor model performance — and this is a structural problem, not a fixable bug.
Data refresh rate. How frequently is new data acquired? Static datasets produce models that degrade over time as the world changes. Dynamic data pipelines that continuously refresh training data are far more valuable.
| Data assessment area | Key questions | Red flags |
|---|---|---|
| Provenance | Where does training data come from? | No documented data lineage |
| Rights | Legal basis for AI training use? | Reliance on scraped or unlicensed data |
| Quality | Error rates, bias, completeness? | No data quality monitoring |
| Volume | Sufficient for model requirements? | Model performance gaps in sparse segments |
| Refresh | How often is data updated? | Static dataset with no update pipeline |
| Exclusivity | Can competitors access the same data? | Training on publicly available datasets only |
A PE firm acquired an AI-powered fraud detection company at a 4x revenue premium based on the target's claimed 98% detection accuracy. Post-acquisition, the firm discovered that the model had been evaluated on a curated test set that did not represent real-world fraud patterns. Production accuracy was 71%. The model had not been retrained in 14 months. The acquisition premium was destroyed within the first year.
Model due diligence should examine:
AI capability is embodied in people as much as in systems. Key-person risk in AI teams is acute because the talent market is extremely competitive and institutional knowledge is difficult to transfer.
Assess the depth of the AI function by examining:
If the target's AI capability depends on one or two key individuals, the acquisition risk is extreme. These individuals may leave post-acquisition — particularly if they are founders with earnout structures that have been fully vested. Structure retention agreements before closing.
The regulatory landscape for AI is evolving rapidly. The EU AI Act (effective August 2026) introduces tiered compliance obligations based on risk level. SEC enforcement around AI washing continues to accelerate. Industry-specific regulators (FCA, FDA, FAA) are developing AI-specific guidance.
Catalogue every AI/ML system in production: purpose, architecture, training data source, team responsible, measurable business impact, and last retraining date. This is the foundation of all subsequent analysis.
Have an independent ML engineer evaluate model performance on production data — not the target's curated test sets. Compare against baseline alternatives (including non-AI approaches) to validate that AI adds genuine value.
Review data provenance, legal rights for AI training use, data quality metrics, and the refresh pipeline. Data assets are often the most valuable — and most legally complex — component of an AI acquisition.
Map the AI team, assess retention risk, benchmark compensation, and identify key-person dependencies. Structure retention agreements for critical AI personnel before closing.
Assess the target's exposure to AI-specific regulation (EU AI Act, SEC enforcement, industry regulators). Identify compliance gaps and estimate remediation costs as part of the acquisition model.
AI due diligence does not end at closing. Post-acquisition integration of AI systems presents unique challenges:
Model portability: Can the target's models be migrated to the acquirer's infrastructure, or are they tightly coupled to specific cloud services, data pipelines, or deployment environments?
Data continuity: Will the data sources that feed the target's AI models remain available post-acquisition? Contractual data sources, partner APIs, and user-generated data flows may be disrupted by ownership changes.
Talent retention: Post-acquisition attrition in AI teams averages 30-40% within the first year (LinkedIn Talent Insights, 2025). Plan for this in the integration model.
The Opagio Growth Platform provides structured intangible asset assessment for PE firms evaluating and monitoring AI-enabled portfolio companies. The questionnaire systematically evaluates technology capability, data assets, and human capital across the AI function.
AI due diligence is not an extension of traditional technology assessment — it is a distinct discipline requiring data evaluation, model quality analysis, talent assessment, and regulatory compliance review. PE firms that apply industrial-era due diligence to AI-era acquisitions will continue to overpay for capability that does not exist or cannot be sustained. The four-dimension framework provides a systematic approach to separating genuine AI value from AI marketing.
Ivan Gowan is Founder and CEO of Opagio. He spent 15 years as a senior technology leader at IG Group (LSE: IGG), where he built and evaluated technology teams, systems, and acquisition targets. Learn more about the Opagio team.
A comprehensive 40-point AI due diligence checklist covering technology verification, data asset assessment, talent evaluation, regulatory compliance, and commercial validation. Designed for investors and PE firms evaluating AI-enabled acquisition targets and portfolio companies.
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 →AI washing inflates valuations and distorts capital allocation. This guide provides a structured due diligence methodology for detecting exaggerated or fabricated AI claims during investment evaluation, with specific technical and commercial red flags.
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.