Publication Type

PhD Dissertation

Version

publishedVersion

Publication Date

5-2026

Abstract

This dissertation comprises three essays which examine the impact of AI investments from an investor’s perspective, and whether they have generated abnormal returns.

Chapter 1. Do Investments in AI Companies by Listed Companies Create Value for Shareholders[1]

We find that public acquirors of Artificial Intelligence (AI) companies have generated economically lower cumulative abnormal returns (CARs) of 2.48% versus acquirors of non-AI companies. Acquirors in the US (2.71% lower) and Asia (3.31% lower) have witnessed significantly lower CARs as compared to acquirors of non-AI companies. Acquirors in utility, energy, business equipment and finance industries underperform, while only acquirors in consumer non-durables outperform non-AI acquirors. Companies with higher returns on asset mitigate these negative effects alluding that acquiror shareholders trust efficient acquirors to benefit from AI acquisitions identifying a potential mechanism for AI acquisitions to create value. Findings reveal shareholders of bidders of AI companies are highly price sensitive and penalise highly valued transactions. They prefer for their investee companies to acquire smaller stakes in AI targets deemed to be a better channel for fostering cooperation between the acquiror and target. We also find evidence that the release of ChatGPT has attenuated the lower CARs observed for acquirors of AI companies, likely indicating a change in investor sentiment with the onset of the Generative AI era. Our results indicate that managers of public acquirors may be exhibiting herding behaviour in pursuing AI acquisitions, potentially driven by fear of missing out on a transformative technology, while their shareholders appear sceptical of transactions that assign rich valuations and secure large ownership in the AI targets. The results persist even after addressing identification concerns including instrumental variable approach using the Global Innovation Index (GII) to address endogeneity concerns and propensity score matching (PSM) methods. These findings are robust to winsorisation of outliers, alternative event windows, exclusion of extreme CARs, and controls for potential omitted variables. Our findings provide key insights for practitioners to consider for their AI strategy and further academic inquiry into value creation in mergers and acquisitions.

Chapter 2. Do Investments in IPOs of AI Companies Create Value for Investors

Research reveals contrasting investment returns for IPO investors in AI-related firms, with the performance being significantly moderated by the launch of ChatGPT. Using buy-and-hold abnormal returns (BHARs) for AI and non-AI IPOs listed on NYSE and NASDAQ from January 2011 to October 2025, the results show AI IPOs launched pre-ChatGPT generated over an 18-month post-IPO horizon, an annual return of 18.1% lower compared to non-AI IPOs, indicating that investors were likely initially sceptical of the long-term potential of AI companies. This phenomenon reverses completely and significantly following the launch of ChatGPT on 30th November 2022, with the net effect of investment performance for AI IPOs turning positive by 35.2% over the same horizon, suggesting a structural shift in investor sentiment as generative AI validated the commercial viability of AI business models – the “post-ChatGPT AI IPO premium”. Results also show that the broader IPO market experienced a significant downturn in the post-ChatGPT period, coinciding with aggressive monetary tightening by the US Federal Reserve. Somewhat counter-intuitively, higher free cash flow is associated with weaker long-run abnormal performance for AI IPOs, suggesting that investors reward AI companies which invest cash in their businesses to fund growth – the “Cash Burn Puzzle.” These findings are robust to alternative event horizons, propensity score matching, IPO market volume controls (hot or cold markets), SIC-based industry fixed effects and robust regression downweighting outliers. These results provide the first empirical evidence on post-IPO investment performance of AI companies and offer guidance for practitioners, such as asset managers, evaluating AI IPO investments especially in light of the upcoming large IPOs of AI companies such as SpaceX, OpenAI and Anthropic.

Chapter 3. Do Private Market AI Focused Funds Create Value for Investors

Research reveals that venture capital funds with an AI investment focus have generated statistically significant and economically meaningful higher abnormal returns relative to non-AI focused funds. Using a comprehensive dataset of 840 private equity and venture capital funds with detailed cash flow data and industry vertical classifications from Preqin, I employ the Kaplan-Schoar public market equivalent (PME) methodology supplemented by total value to paid-in capital (TVPI) to investigate both relative and absolute fund investment performance. AI-focused venture capital funds have generated significantly higher returns than their non-AI counterparts, outperforming by approximately 30% on TVPI, 14% on PME relative to the S&P 500, and 19% on PME relative to the Russell 2000. In contrast, there is no evidence that private equity (or buyout) funds with an AI focus have generated positive abnormal returns, suggesting that value creation from AI investments in private markets is concentrated in the venture capital space. Larger funds have experienced lower investment returns, consistent with diseconomies of scale, while funds managed by larger GPs with greater total assets under management have experienced higher returns, possibly due to superior resources and deal sourcing capabilities. Geographically, AI-focused funds in Australasia and the Middle East show outperformance across performance metrics, however the fund sample sizes in these two regions are small, restricting the generalisability of the findings at a regional level, while North America, Europe and Asia show no significant differential effect. These findings are robust to propensity score matching, winsorisation of outliers, and restriction to mature funds with vintage years of 2020 or earlier to address concerns regarding unrealised valuations in younger funds. These findings provide the first empirical evidence on investment performance of AI-focused private market funds using fund-level cash flow data and offer valuable guidance for limited partners evaluating AI-focused fund commitments.

[1] Co-authored paper with Professor Aurobindo Ghosh titled: To Buy or Not To Buy: Do Investments in AI Companies Create Value for Shareholders

Keywords

Artificial Intelligence, Mergers & Acquisitions, Initial Public Offerings, Venture Capital, Private Equity, Shareholder Value Creation, Event Study, Cumulative Abnormal Returns (CAR), Buy-and-hold Abnormal Returns (BHAR), Public Market Equivalent (PME)

Degree Awarded

PhD in Business (General Management)

Discipline

Corporate Finance

Supervisor(s)

GHOSH, Aurobindo

First Page

1

Last Page

181

Publisher

Singapore Management University

City or Country

Singapore

Copyright Owner and License

Author

Download

Available for download on Monday, July 12, 2027

Share

COinS