Investing in AI

Introduction 

Since publication of the first edition of Reed Smith’s Entertainment & Media AI Guide, investment in the AI sector has accelerated at an unprecedented pace. By some estimates, annual global investment in AI almost tripled from 2023 to 2025, reaching close to $600 billion of investment activity in 2025. The explosive growth shows no signs of abating, with 2026 venture investments to date already eclipsing the annual total for 2025 – even before factoring in the expected blockbuster initial public share offerings from companies such as OpenAI and Anthropic.

The Evolving AI Investment Landscape

The massive investments in AI reflect not just the growing interest in frontier AI companies and related technology developers, but also the rapid increase in infrastructure costs. Average investment transaction size has more than doubled over the last two years – reaching over $60 million in 2025 – with more than two dozen $1 billion-plus financing transactions in 2025 alone. At the same time, the ecosystem of companies receiving financial backing has also broadened, with well over 2,000 AI-focused enterprises closing funding rounds in 2025.     

In the 2023 Guide, we outlined key areas that investors in AI-related businesses should consider when evaluating the risks, opportunities, and overall viability of an AI investment. Most of these key areas remain highly relevant – including ownership of IP and technology, data quality and access, team experience, business models, and exit strategies.

At the same time, a number of other key issues have emerged as highly relevant to many AI-related investment prospects:

• Technical defensibility and model dependency. Many new AI sector businesses focus on offering platforms and tools that leverage third-party foundation AI models. Key questions for these vendors include whether they own significant and defensible proprietary technology and whether reliance on third-party models presents significant risks, for example, if access to these models ceases to be available.
• Compute unit economics. The economics of a data-intensive AI business will depend heavily on compute capacity requirements and associated costs. These costs – both current and projected – should be evaluated carefully in assessing the viability of the enterprise.
• Regulatory compliance risks. The last few years have seen a proliferation of new laws and regulations in most major jurisdictions addressing matters such as liability for AI-generated errors or losses, privacy, misinformation, and copyright infringement. Investors will need to evaluate the impact and potential risks presented by current and future changes in the legal environment.
• Competition for talent. Attracting experienced AI talent is increasingly competitive and expensive. The quality of an AI business’s current team and its prospects for expansion are more important than ever, as is evaluating incentives and strategies in place for retaining key talent.
• Addressable market and competition. As AI-based start-up businesses have proliferated, many new ventures focus on increasing narrow enterprise and consumer niches. Particularly in light of rapidly accelerating costs, it may be important to evaluate whether the market opportunity addressed by the venture is sufficiently large and compelling, and whether the level of competition (including from potential market entry by established vendors such as OpenAI, Anthropic, Google, and Microsoft) within the sector is manageable.