How has the thematics landscape changed in recent years?
Thematic indexing has continued to evolve alongside the broader financial markets. However, its core objective remains unchanged: providing rules-based exposure to structural trends and technologies that are not cleanly captured through traditional sector and industry classifications. Recent advances in artificial intelligence (AI) have accelerated both the development of new themes and the tools used to research them. The emergence of large language models (LLMs) and the buildout of AI-related infrastructure have become important drivers of economic activity, creating new opportunities to develop thematic indices.
The growing breadth and pace of the thematic landscape have also changed how thematic opportunities in indexing are researched and monitored. Last year, S&P Dow Jones Indices (S&P DJI) engaged with Onthos (formerly Theia Insights) to enhance its thematic research and indexing capabilities. More recently, S&P DJI launched a monthly S&P Thematics Dashboard covering more than 200 themes across global markets, to provide a tool helping to monitor the evolving thematic landscape.
The interaction between AI and thematic indexing can broadly be viewed through two lenses: AI-enabled thematic indexing and thematic indexing for AI exposure. The first relates to the growing use of AI-powered tools that assist with theme discovery, classification and company identification: as these tools have become more sophisticated, their adoption across the index construction process has increased. The second focuses on thematic indexing for AI exposure, examining how indices can systematically capture companies involved in the development, deployment and adoption of AI technologies.
While final decision making and governance in S&P DJI’s index construction and calculation processes remain human-led, S&P DJI has been implementing the AI and machine learning (ML) tools of S&P Global’s Kensho framework since 2018 to support theme identification and company classification. Recent advances in ML algorithms and LLMs have augmented these capabilities, supporting both the thematic classification and the breadth of information that can be incorporated into the analysis.
A practical benefit of these tools is an ability to connect information across a broader set of data sources, conceptually similar to the mosaic approach used by industry analysts, where multiple sources of information are brought together to assess a company’s exposure to a theme.
This capability can be particularly useful when mapping companies within emerging themes, thus supplementing traditional classification frameworks to more fully capture the thematic ecosystem. Humanoid Robotics is one such example, with relevant companies spanning software, hardware, automation and component suppliers across multiple sectors and industries. Using knowledge graph frameworks built from company disclosures, research reports and other information sources, AI tools can help to identify these connections and assess each company’s relevance to the broader thematic value chain. This allows companies to be identified even when they do not directly produce humanoid robots but play a critical role in enabling the ecosystem.
