Generative AI has become perhaps the most important technological development since the Internet, attracting hundreds of billions of dollars in capital investment and filling untold newspaper columns.
While some view it as a revolutionary breakthrough, others worry it’s another overhyped technology destined for disappointment. We believe that the truth, as is often the case, lies somewhere in between. For companies outside the technology sector, generative AI offers genuine opportunities, but only if pursued with a pragmatic focus on concrete applications and measurable benefits.
To date, the most significant beneficiaries of generative AI have been semiconductor manufacturers, driven by hyperscalers’ vast infrastructure spending. Over the past year, Alphabet, Amazon, Meta, and Microsoft collectively invested $199 billion in capital expenditure, with year-on-year growth averaging 55%[1]. Although this pace is slowing, Morgan Stanley projects capex will reach $300 billion by 2025, nearly double the levels of 2023. For this investment to remain sustainable, it must lead to new business models and value-added services for which end clients and consumers will pay handsomely.
Next stage drivers
This depends on two critical factors. Proprietary data is vital for creating AI solutions that are unique and hard to replicate, while a deep understanding of customer workflows ensures these tools solve real-world problems. Without both, there is a significant risk of producing tools that fail to deliver value or gain client adoption. Too many such missteps could leave companies unable to justify continued investment.
The companies best positioned to succeed with AI are those already skilled at leveraging proprietary data to address customer needs. Informa, a global leader in events and exhibitions, exemplifies this through its IIRIS platform. IIRIS collects and analyses detailed data from thousands of live B2B events worldwide, capturing interactions between attendees and exhibitors. This unique dataset reflects customer behaviours, preferences, and networking patterns that are difficult for competitors to replicate.
Informa also has deep insight into how events deliver value to participants. Exhibitors rely on events to generate leads and showcase their offerings, while attendees look for relevant content and meaningful connections. With AI, Informa can enhance these experiences. Automated agenda creation, personalised follow-up recommendations, and smarter lead assessments could improve engagement and productivity. While these are incremental rather than radical changes, they meaningfully enhance the event’s value, ensuring it remains a key date in participants’ calendars.
A second example is Verisk, a leader in property and casualty insurance data and analytics, which has built its success on the ISO Statistical Database, a contributory dataset containing over 34 billion premium and loss records as of 2023.
Access to this leading dataset is contingent on insurers sharing their own data, ensuring its continued growth in line with the volume of insurance premiums written in the US. Verisk’s strength lies not only in the scale of its data but also in its deep understanding of how insurers utilise it, buttressed by Verisk’s powerful position in designing the standard policies (‘forms and rules’) used by insurers. This enables the company to support critical decisions across benchmarking, ratemaking, product development, and strategic planning.
The introduction of AI allows Verisk to analyse its dataset more effectively, answer complex queries with greater precision, and streamline claims processing. To give a concrete example, its Discovery Navigator module can improve the productivity of claims managers by efficiently and reliably abstracting lengthy and unstructured medical documents. This is highly attractive to insurers given the traditionally costly and manpower-intensive nature of claims adjustment. However, this progress remains incremental rather than disruptive, enhancing rather than overhauling its proven approach.
While generative AI offers significant potential, its implementation presents companies with several challenges. Data privacy remains a critical concern, as the vast datasets powering AI often contain sensitive information that requires robust safeguards. Generative AI systems remain at risk of ‘hallucination’, the provision of erroneous data that may prove serious (and costly) to enterprise clients. There are significant regulatory complexities, as companies must navigate evolving rules and compliance standards governing AI usage. Finally, financial viability is critical. AI-driven services can only create lasting value if the additional revenue they generate exceeds the costs of delivering them. This includes not only development and infrastructure investment but also the ongoing expenses of maintenance and support. Without disciplined execution and careful cost management, these initiatives risk eroding profitability and undermining their long-term success.
Generative AI may yet prove to be a boom or bust, but for companies like Verisk and Informa, the technology is an opportunity rather than a necessity. Both are well positioned to meet the challenges of AI implementation, with proprietary data assets which reduce ‘hallucination’ risk, and a deep understanding of client workflows that enable them to integrate AI in ways that create tangible, immediate value to clients. Their long-term prospects are undoubtedly enhanced by AI, but they are not dependent on it, ensuring their business models remain resilient regardless of how the technology evolves.
[1] Brian Nowark, Morgan Stanley, November 2025.
By Chris Elliott, portfolio manager of the Evenlode Global Equity fund at Evenlode Investment
