M&A in the AI Era: What Buyers Can Do to Confirm and Protect Value

Key Points

• As more transactions involve AI, buyers face challenges in validating and protecting the value of their acquisitions.
• Legal structures such as earnouts can help to bridge valuation gaps with sellers and ensure that the ultimate price aligns with actual performance.
• Alternative structures may be necessary when talent is the primary asset.
• Buyers may also want enhanced representations, covenants and indemnities with long durations to cover risks such as those involving data rights, model performance and regulatory compliance.

__________

As artificial intelligence (AI) continues to become a core differentiator across industries, dealmakers are seeing more transactions where a key driver of value is not just a traditional product line, customer base or physical asset, but the AI itself.

Whether the target is an AI-native startup, a conventional business with a high-performing internal AI platform or a company whose competitive advantage is tied to proprietary data and model performance, the ability to correctly attribute, validate and protect AI-based value is now essential to the successful execution of M&A involving the technology.

However, AI-driven value presents unique challenges. That value can evaporate if the underlying data is noncompliant, key technical experts depart or the models underperform outside of carefully controlled demos.

In addition, buyers often struggle to price AI assets correctly, as the difference between perceived value and validated value can be substantial. As such, AI-focused M&A transactions increasingly require deeper legal and technical due diligence, tighter valuation frameworks and stronger contractual protections for buyers.

Validating What You Are Buying

When a seller markets its AI capabilities, that label can refer to anything from a handful of Python scripts to a robust, scalable, multimodal platform deployed across large enterprises. Buyers should therefore identify the true source of value and conduct robust due diligence to validate it.

In practice, this often means asking targeted questions:

• What proprietary datasets does the target own or have rights to, and how permissioned and traceable is that data?
• How were the models trained, and how does their performance hold up under red-teaming (i.e., adversarial stress-testing to probe for vulnerabilities), edge-case testing (i.e., evaluating performance in rare or extreme scenarios) and scaling?
• Are compute costs sustainable at commercial volumes? 
• Is the target’s core know-how concentrated in only a few key individuals?

Without a clear basis for attributing value, buyers risk overpaying for AI assets. Because these assets are highly technical in nature, they often require tailored diligence, including by specialized third-party diligence firms, especially where such assets are a key driver of the perceived value of the target.

Deal Architecture to Bridge Valuation Gaps

AI value is often uncertain, and model performance can change significantly in a short span of time. As a result, buyers are using specific deal mechanisms to bridge valuation gaps and align the purchase price with validated capabilities.

Common structuring tools include earnouts tied to AI-related metrics, with additional consideration payable only if the target achieves defined performance benchmarks, deployment milestones, revenue thresholds and/or compute-efficiency goals.

Buyers may also hold back a portion of the purchase price through escrows, to mitigate the risk of technical underperformance or data rights issues that surface post-closing.

In some cases, buyers may turn to alternative structures — such as joint ventures, “acquihires” or strategic hires with significant compensation packages — where the primary value lies in securing key talent rather than acquiring the technology itself.

These structures allow buyers to access critical AI expertise while avoiding the valuation uncertainty associated with the target’s underlying technology.

Representations and Covenants to Protect Value

In circumstances where AI is a critical value driver, buyers may request that specific AI-related representations be categorized as “fundamental” (or another enhanced category) with longer survival periods and higher indemnity caps than those applicable to “general” representations. Examples include representations regarding:

• Rights to data used for training.
• Absence of material data protection or intellectual property violations.
• Accuracy of disclosures about model architecture and performance, model safety and explainability.
• Absence of undisclosed third-party dependencies or restrictions.
• Compliance with AI-specific regulations.

In addition, given AI assets evolve rapidly and talent retention is critical, buyers often need stronger covenants between signing and closing to prevent deterioration of AI value during the interim period. These covenants may require the target to:

• Refrain from materially changing model architecture or datasets.
• Ensure lawful use of training data.
• Not change terms of use or privacy policies.
• Retain key AI engineers.
• Maintain sufficient graphics processing unit (GPU) or cloud capacity.

(For a discussion of key M&A deal terms in the AI sector, see our June 2024 article “M&A in the AI Era: Key Deal Terms to Watch.”)

Recourse to Address AI-Specific Risks

When AI is the primary value driver, securing meaningful recourse becomes even more critical. Buyers may need to negotiate tailored indemnities in private deals to address the unique risks inherent in AI assets, including protections against:

• Misrepresentations regarding data provenance or licensing.
• Unauthorized or noncompliant training practices.
• Defects or nonperformance of key AI functionality.
• Violations of data protection or emerging model risk management requirements.

Indemnification in private deals may also be used to address breaches of the preclosing covenants discussed above. In doing so, buyers should ensure that survival periods, indemnity caps and baskets reflect the magnitude of potential AI-related exposure.

Buyers may also turn to representations and warranties insurance (RWI) to help manage AI-specific risks. However, as these risks become larger and more common, RWI insurers are taking a closer look at AI-specific issues, which could lead to policy exclusions for data provenance, model performance or other AI-related representations.

Meanwhile, evolving antitrust and national security policies add additional complexity. (See our June 2025 article “M&A in the AI Era: Key Antitrust and National Security Considerations.”) Transactions involving AI, sensitive data and/or critical compute infrastructure increasingly face heightened regulatory scrutiny and extended review timelines.

As a result, sellers may seek stronger regulatory covenants or a reverse termination fee if the deal collapses due to regulatory hurdles beyond their control. Buyers must balance this commercial expectation with the need for recourse on AI-specific risks, calibrating the overall remedy package to reflect both the strategic value of AI assets and the regulatory uncertainty surrounding AI-focused deals.

See the full 2026 Insights publication

This memorandum is provided by Skadden, Arps, Slate, Meagher & Flom LLP and its affiliates for educational and informational purposes only and is not intended and should not be construed as legal advice. This memorandum is considered advertising under applicable state laws.