Concentrating Intelligence: Scaling and Market Structure in Artificial Intelligence

Concentrating Intelligence:
Scaling and Market Structure in Artificial
Intelligence
Anton Korinek and Jai Vipra1
Working Paper No. 228
October 2nd, 2024
ABSTRACT
This paper examines the evolving structure and competition dynamics of the rapidly growing
market for foundation models, focusing on large language models (LLMs). We describe the
technological characteristics that shape the industry and have given rise to fierce competition
among the leading players. The paper analyzes the cost structure of foundation models,
1 University of Virginia and Centre for the Governance of AI.
This is a revised version of a paper presented at the 79th Economic Policy panel meeting on Apr 4/5, 2024,
in Brussels. We thank Susan Athey, Emma Bluemke, Emilio Calvano, Giacomo Calzolari, Claire Dennis,
Avi Goldfarb, Doh-Shin Jeon, Aidan Kane, Pia Malaney, Sarah Myers West, Sanjay Patnaik, Nicholas
Ritter, Max Schnidman, Eli Schrag, Rob Seamans and Joseph Stiglitz as well as two anonymous reviewers
for thoughtful comments and conversations. Any remaining errors are our own. Korinek gratefully
acknowledges financial support from the Center for Innovation, Growth and Society of the Institute for
New Economic Thinking (INET-CIGS, grant no INO21-00004). Vipra received financial support as a
Winter Fellow from the Centre for the Governance of AI. An earlier version of this paper was circulated
under the title “Market Concentration Implications of Foundation Models: The Invisible Hand of
ChatGPT.”
Ineteconomics.org | facebook.com/ineteconomics 	60 E 42nd St, Ste 450, New York, NY 10165

-- 1 of 32 --

2
emphasizing the importance of key inputs such as computational resources, data, and talent, and
identifies significant economies of scale and scope that may create a tendency towards greater
market concentration in the future. We explore two concerns for competition, the risk of market
tipping and the implications of vertical integration, and use our analysis to inform policy remedies
to maintain a competitive landscape.
https://doi.org/10.36687/inetwp228
JEL codes: D43, O33, L86, L40, L41, K21
Keywords: Artificial Intelligence, economic concentration, vertical integration, AI regulation.

-- 2 of 32 --

3
1. Introduction
In a post titled “Moore’s Law for Everything”, OpenAI CEO Sam Altman predicted that within
the next few decades, AI technology would “do almost everything, including making new
scientific discoveries that will expand our concept of ‘everything’” (Altman 2021). A paper co-
authored by researchers at OpenAI estimated that most occupations are exposed to the deployment
of large language models (LLMs) like ChatGPT, and that once complementary investments are
made, up to 49 percent of workers could have half or more of their tasks exposed to LLMs
(Eloundou et al. 2024). If AI models will indeed play such an important role in our economy, then
the structure of the market in which they are offered will have first-order implications for social
welfare. Policymakers,

... (truncated, 80 KB total)