MIT Sloan Management Review Article on Who Profits the Most From Generative AI?

Unpacking what it takes to build and deploy a large language model reveals which players stand to gain the most — and where newer entrants might have the best prospects.

In the months since the public launch of ChatGPT, massive investments have been made in the form of venture capital firms plowing money into generative AI startups, and corporations ramping up spending on the technology in hopes of automating elements of their workflows. The excitement is merited. Early studies have shown that generative AI can deliver significant increases in productivity.1 Some of those increases will come from augmenting human effort, and some from substituting for it.

But the questions that remain are, who will capture the value of this exploding market, and what are the determinants of value capture? To answer these questions, we analyzed the generative AI stack — broadly categorized as computing infrastructure, data, foundation models, fine-tuned models, and applications — to identify points ripe for differentiation. While there are generative AI models for text, images, audio, and video, we use text (large language models, or LLMs) as an illustrative context for our discussion throughout.

About the Author

Kartik Hosanagar is the John C. Hower Professor of Technology and Digital Business at the Wharton School of University of Pennsylvania and a founding faculty director of Wharton’s AI center. He is the author of A Human’s Guide to Machine Intelligence: How Algorithms Are Shaping Our Lives and How We Can Stay in Control (Viking, 2019) and the newsletter Creative Intelligence. Ramayya Krishnan is the dean of the Heinz College of Information Systems and Public Policy at Carnegie Mellon University, where he is also the William W. Cooper and Ruth F. Cooper Professor of Management Science and Information Systems. He is the founding faculty director of the university’s Block Center for Technology and Society and also chairs the AI futures group of the National AI advisory committee.