As artificial intelligence increasingly becomes a strategic imperative across industries, the merger and acquisition (M&A) landscape is witnessing a surge in activity centered around AI companies. Investors and acquirors, keen on tapping into this potential, face unique challenges that go beyond traditional due diligence. They must adapt their evaluation processes to address the intricacies of AI technology, data governance, regulatory landscapes, and intellectual property. A recent discussion with industry experts highlights how to properly assess risk and determine the true value of AI companies in these transactions.
Understanding the Unique Value of AI Companies
Unlike traditional software firms, AI companies often derive their value not just from software code, but from sophisticated machine learning models and invaluable datasets—their “data moat.” Industry veteran Edoardo de Martin, CEO and Founder of Industrio AI, points out that the real winners in this space will be those firms that build robust tooling infrastructure to enable broader industry adoption of AI solutions. This mirrors the 19th-century lesson: when everyone is digging for gold, being in the pick-and-shovel business is advantageous.
Key considerations include:
- Infrastructure and Security: AI companies require advanced hardware, high-performance servers, and cutting-edge networking solutions. Decisions around on-premises versus cloud-based systems, and data storage, critically affect their scalability and customer attractiveness.
- Intellectual Property Nuances: The core IP of AI firms often lies in exclusive datasets and proprietary algorithms, distinguishing them from traditional software companies that mainly rely on unique code. Clear ownership of these assets is paramount to value.
Tailoring Due Diligence for AI Transactions
Due diligence in AI-focused M&A must evolve to cover facets that are specific to the technology and its ecosystem. Prospective investors and acquirors should structure their inquiries around several detailed categories:
Intellectual Property
- Registered and Non-Registered IP: Request comprehensive lists of all intellectual property, including patents, trademarks, trade secrets, and proprietary algorithms. Understand the geographical scope and registration status.
- Data Collection and Ownership: Investigate how training data is sourced, licensed, and managed. Ensure that the firm holds rights to use the data to train and refine their models and that no data rights issues jeopardize future operations.
- Third-Party Agreements: Assess contracts tied to third-party AI inputs, outputs, escrow arrangements, and beneficiary rights that may influence model ownership and usage rights.
Technology and Data Architecture
- AI Technology Overview: Obtain a detailed description of the types of AI technology in use—such as machine learning or deep learning—and how these technologies are integrated into products and services.
- Development Processes: Examine the company’s software development lifecycle, AI model training, data pipeline, and quality control measures. Review how bias detection and remediation strategies are built into the development process.
- Infrastructure Readiness: Evaluate the robustness of the company’s technology stack, deployment architecture, and data security, ensuring that it can support scaling and mitigate security vulnerabilities.
Privacy, Data Protection, and Compliance
- Privacy Policies and Data Types: Review internal and external privacy policies, and understand the types of personal data collected, processed, and stored.
- Data Safeguards: Assess data security measures, IT system management, encryption practices, data retention, and destruction protocols.
- Regulatory Compliance: Scrutinize evidence of compliance with data protection laws, any known breaches, and steps taken to remediate them. Given the evolving regulatory environment—including proposals like Canada’s AIDA, the EU’s Artificial Intelligence Act, and efforts in the U.S.—ensure the target company is prepared or adaptable to upcoming legal changes.
Regulatory and Licensing Review
- Approvals and Licenses: Request copies of significant regulatory approvals and licenses. Understand how these govern the use of AI technologies, particularly distinguishing between internal use and client-facing outputs.
- Regulatory Risks: Identify any areas where the company may face compliance challenges with emerging AI-specific laws or where its business model might be impacted by new legislative measures.
Assessing the Role and Governance of AI in the Business
A key step is to understand how integral AI is to the company’s operations:
- Centrality of AI: Determine whether AI underpins the company’s products and services or plays a peripheral role. The deeper the integration, the greater the need for detailed scrutiny.
- AI Governance Framework: Evaluate how the company governs the AI lifecycle. This includes assessing ethical AI practices, mechanisms for mitigating bias, processes for ensuring transparency and accountability, and strategies for upholding data integrity.
Mitigating Common Risks in AI M&A
Potential risk areas specific to AI must be identified and mitigated:
- Ownership Uncertainty: Verify that the target has clear and undisputed ownership of its AI models and related IP.
- Data Rights and Authority: Confirm the company possesses the necessary rights to all training and operational data used in model development.
- Reliance on External Models: Understand dependencies on foundational or open-source models which might impose restrictions or lead to licensing complications.
- Regulatory Compliance and Future-Proofing: Ensure the company’s practices are compliant with current regulations and adaptable to future legal requirements.
Strategic Steps for Effective AI Due Diligence
Given the inherent complexity of AI firms:
- Build an Expert Team: Engage legal, technical, and industry experts who understand both the AI landscape and legal intricacies.
- Stay Informed on Regulatory Developments: Keep abreast of evolving laws and international standards (e.g., ISO/IEC 42001) that will influence AI governance.
- Scenario Planning: Anticipate changes in regulation, technology, and market dynamics that might affect the target company’s operations and valuation.
- Comprehensive Evaluation: Expand due diligence beyond traditional software metrics to include an analysis of data sourcing, model robustness, ethical considerations, and AI personnel retention strategies.
Conclusion
The dynamic nature of AI technology and its regulatory environment necessitates a nuanced and rigorous approach to M&A due diligence. Investors and acquirors must move beyond conventional software evaluation, incorporating thorough assessments of intellectual property, data practices, AI governance, and regulatory compliance. By doing so, they can uncover hidden risks, accurately value potential acquisitions, and make informed, strategic investment decisions in the rapidly evolving AI sector.
