Conceptualizing the Business Model of Machine Learning as a Service Platform
J.L. Tan (TU Delft - Technology, Policy and Management)
G.A. de Reuver – Graduation committee member (TU Delft - Technology, Policy and Management)
H. Khodaei – Graduation committee member (TU Delft - Technology, Policy and Management)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Despite the growing industrial impact of Machine Learning as a Service (MLaaS) platforms in lowering the barriers for SMEs and large organisations to adopt artificial intelligence, there is a limited understanding of their business models. This thesis addresses three key gaps in existing research: conflated definitions of MLaaS platforms, the absence of business model analysis, and the lack of insights from recurring archetypes. So this research develops a business model taxonomy of MLaaS platforms and identifies common archetypes based on their dimensions and characteristics.
The study adopts a hierarchical conceptual approach, integrating Teece’s (2010) value creation, delivery, and capture mechanisms with Osterwalder and Pigneur’s (2010) Business Model Canvas components. A literature review was conducted to establish working definitions and extract key dimensions. Following Nickerson et al.’s (2013) iterative taxonomy development method, 24 MLaaS platform cases were analysed in conceptual-to-empirical and empirical-to-conceptual cycles. The completed taxonomy was evaluated using Szopinski et al.’s (2020) structural validity criteria and Miles and Huberman’s (1994) practical usability guidelines. The resulting business model taxonomy systematically captures MLaaS platform business models across value creation, delivery, and capture mechanisms, revealing distinct configurations shaped by cloud infrastructure dependencies and machine learning workflow orchestration. Four archetypes (cloud orchestrator, data orchestrator, aggregator, and niche specialist) emerged, with each archetype reflecting specific priorities in scalability and specialisation.
This research presents the first classification framework for analysing the business models of MLaaS platforms, addressing a significant gap in existing literature. By identifying and distinguishing key dimensions and archetypes, this research extends platform theory to reflect the variety of business models beyond those of large technology firms, and it challenges existing views that see Big Tech dominance as inevitable. Alongside its theoretical contribution, the taxonomy provides practical value. For startups and new market entrants, it offers a structured approach to determine their strategic position before developing their business model. For established MLaaS platforms, it serves as a tool for benchmarking, supporting decisions on innovating current models or transitioning to others in line with market changes and technological developments.