AI: Acquiring Intelligence – APAC M&A Legal Playbook 2025

Executive Summary:

AI M&A appears to be a bright spot swimming against the macro headwinds in Asia-Pacific. Overall APAC deal-making has slipped as economic uncertainty, higher rates, geopolitics and election calendars weigh on sentiment. Yet within that backdrop, industry trackers report that the aggregate value of APAC AI-focused tech deals grew about 5%-10% during 2024, and early-2025 valuations are edging higher even as volumes decrease. Market observers attribute this resilience to a scarcity premium driven by the limited number of truly differentiated models, proprietary datasets and specialist teams. Sustained enthusiasm for AI keeps bidding pressure and valuations elevated.

That scarcity premium, however, does not justify weak economics or underwrite risks and liabilities of acquiring an AI business. Buyers may still cite "capability premiums", yet tighter liquidity from higher interest rates, sharper regulatory scrutiny (FDI, export-control and data-localization) and rising capex requirements and increased operating costs are driving increased scrutiny on individual opportunities. In this environment, a credible monetization pathway – cost synergies, accelerated product launches or data-driven revenue – appears increasingly indispensable to defend headline multiples.

Execution risk is also climbing. National security reviews, data-localization mandates and more assertive antitrust scrutiny mean cross-border AI deals now require a disciplined evaluation of deal certainty and post-completion regulatory risks. Value preservation is equally complex because a large share of consideration hinges on retaining human talent and expanding and monetizing the acquired capability.

This Playbook surveys APAC AI M&A as of Q1 2025. Interest from financial investors and strategics (inorganic capability lift, deeper data moats, scale and pricing power) remains high, but obstacles are becoming more significant.

Key findings of this bulletin:

• Deal landscape (Part 1). Recent APAC AI deals suggest buyers center on (1) proven model performance as a technological and financial matter, (2) domain-specific data and (3) human talent. Buyers are using bespoke structures such as minority stakes, staged buy-outs or peer mergers, perhaps to bypass FDI hurdles and valuation gaps.
• Assets in play (Part 2). AI acquisitions are not just about picking up IP. Preserving value means recognizing, and conducting diligence in respect of, multiple different asset classes including: (1) models, (2) training data, (3) compute capacity, (4) human capital, (5) IP, (6) operational and governance structures, and (7) user base with its telemetry – each of which raises its own diligence and value-preservation challenges.
• Human paradox (Part 3). Retention risk is existential: the talent that builds, maintains and monetize the assets is itself a core asset. Accordingly, retention pools, equity roll-overs, key-person conditions and bad-leaver clawbacks belong at the heart of any value-preservation strategy.
• AI-specific drafting (Part 4). In our observation, AI-deal contracts increasingly feature five recurring buy-side safeguards:
  1. Seller warranties that code, models and training data have lawful, well-documented origins;
  2. Indemnities to cover the potential cost if a regulator were to order deletion of a tainted model;
  3. Comfort that no problematic open-source code is embedded in the stack, with ring-fencing provisions where doubts persist;
  4. Clearly defined timelines, break-fees and cost-sharing to navigate export control, security, localization and antitrust clearances; and
  5. Covenants requiring the seller to comply with export control laws and regulations and assist with future compliance as AI regulations evolve (e.g., strict and comprehensive pre-completion document retention).
• Regulatory outlook (Part 5). Regulation appears to be converging on three pressure points: competition, national security and accountability. We highlight certain key frameworks, standards and regulations in this part.

Part 1 – Deal Landscape: Notable APAC Deals & Patterns in Scope, Structure and Pricing

Trends from the foregoing APAC deals:

1. Capability, not earnings, appear to be driving pricing. Buyers are willing, even eager, to secure data, models and specialist teams well before EBITDA turns positive and drives up asset valuation. For examples of this dynamic, we note (i) Canva's reported nine-figure acquisition of two-year-old Leonardo.ai and (ii) NVIDIA reportedly paying a generous premium for VinBrain.
2. Cross-border is standard and so is FDI scrutiny. Strategic and financial investors seem to be shopping across APAC to secure talent and localized data. Expect heightened FDI reviews wherever personal data, chip IP or cloud infrastructure is implicated.
3. Scale and consolidation as a way to meet capex demands. The Rebellions-Sapeon merger could be first of many regional AI firms combining to find scale in an accelerated manner and to fund runaway capex demands.
4. Sector depth commands a premium. Aurionpro-Arya deal may signal an acquisition playbook, whereby incumbents buy niche AI stacks that already meet sector governance requirements (finance, healthcare, etc.) to shorten compliance and learning timelines.
5. Deal structures stay flexible. We are observing a myriad of deal structures from cash stake-buys (Aurionpro-Arya) to full take-outs (Canva) and peer-to-peer mergers (Rebellions-Sapeon). Minority or phased structures help parties navigate valuation gaps and regulatory caps while still securing strategic control.

Part 2 – Assets in Play: What Assets Are Really Being Bought in an AI Deal?

AI transactions rarely involve "just software." In our view, a typical 2025 deal bundles seven distinct asset classes: (i) model weights; (ii) proprietary training data; (iii) compute capacity; (iv) human capital; (v) intellectual property; (vi) operational and governance workflows (certifications, audit trails, alignment reports) and (vii) user base and other customer telemetry. Each has its own distinct diligence and value-preservation challenges. The table below sets out why buyers seem willing to pay a premium for these assets and where diligence efforts typically focus.

Taken together, these assets form a self-reinforcing loop: models improve with better data and compute, while retained talent and clear governance keep that loop turning. However, value can evaporate if any strand breaks. Accordingly, purchase agreements need to lock down ownership, access and retention. In Parts 3 and 4 below, we outline the key contract levers that, in our experience, help preserve the all-important human, model, data and compute capabilities buyers are paying for.

To view the full article click here

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.