Vivek Ghosal reviews the data, economics, and market conditions of the growing artificial intelligence market and finds that it is quite dynamic in terms of evolving partnerships and firms, and is relatively competitive. Thus, Big Tech investments into AI startups do not warrant investigation by the government at this time.
This article is part of a symposium which asks experts to evaluate the anticompetitive harms of Big Tech investments in AI startups in light of recent investigations from antitrust agencies on both sides of the Atlantic. See here to read Matt Perault’s contribution and return later this week for contributions from John Kirkwood and Stacey Dogan.
In January 2024, the Federal Trade Commission issued orders under Section 6(b) of the FTC Act to Alphabet, Amazon, Anthropic, Microsoft, and OpenAI, requiring them to provide information on investments and partnerships involving generative artificial intelligence and cloud services. Microsoft has invested billions into OpenAI, the creator of ChatGPT, and now owns 49% of the startup. Alphabet and Amazon have likewise invested billions into Anthropic, whose language learning model is Claude. In June 2024, the Justice Department and the FTC agreed on an arrangement to investigate potential anticompetitive conduct in AI markets. The DOJ will take the lead in investigating the behavior of Nvidia, the biggest maker of AI chips. The FTC will take the lead in investigating AI markets. Below we focus on FTC-related investigations and ask if these firms’ partnerships constitute anticompetitive behavior.
The key alliances that are on FTC’s radar are Microsoft-OpenAI, Amazon-Anthropic, and Google-Anthropic. Microsoft is heavily incorporating AI into its products, particularly Word, PowerPoint, and Bing. In addition, it owns 49% percent of OpenAI and has invested about $13 billion in the company. Microsoft had an observer seat on OpenAI’s board, but recently stepped down. OpenAI’s other investors include Infosys, and venture capital funds Tiger Global, Sequoia Capital, Founders Fund, Thrive, K2 Global, and Andreessen Horowitz.
Google has invested about $2 billion in Anthropic. The Google-Anthropic collaboration involves incorporating Claude 3, which is Anthropic’s frontline technology, into Google’s products. Further, Google has developed its own AI, Gemini, which is a cluster of AI models designed to compete with OpenAI. Amazon has invested about $4 billion in Anthropic. AWS will become Anthropic’s primary cloud provider. AWS consumer and business data will help Anthropic develop its next generation technologies. Further, Anthropic will use the AWS Trainium and Inferentia AI chips to develop future foundation models. Anthropic’s other investors include Zoom, Spark Capital, Salesforce, HOF Capital, Menlo Ventures, Sound Ventures, Pioneer Fund, among others.
FTC Investigations and Rationale
At its core, the FTC investigations are designed to evaluate the competitive impacts of these investments and partnerships. As is common in such investigations and market analysis, the FTC required the firms to provide data and information on products and services related to generative AI, market shares, competitors, future expansion into product or geographic markets, and competition for AI inputs, among other aspects.
FTC Chair Lina Khan noted that “History shows that new technologies can create new markets and healthy competition. As companies race to develop and monetize AI, we must guard against tactics that foreclose this opportunity,” and that the study “…will shed light on whether investments and partnerships pursued by dominant companies risk distorting innovation and undermining fair competition.”
In the June 2023 FTC report “Generative AI Raises Competition Concerns,” it was noted that “Incumbents that control key inputs or adjacent markets, including the cloud computing market, may be able to use unfair methods of competition to entrench their current power or use that power to gain control over a new generative AI market.”
The FTC noted that generative AI is a rapidly evolving technology that will transform many sectors. Of particular concern are network and platform effects. In network effects, a firm, by being a first mover, could claim an advantage over its rivals if its AI models, with the benefit of longer duration in the market and larger customer base, generate more useful content and data than rivals’ products. As the generative AI models require a lot of customer data and are dependent on feedback loops, the network effects have the potential to impart market power. The platform effect is of concern as firms can become highly dependent on a particular platform for their generative AI needs in general, and cloud services in particular. As platform switching costs can be high, firms and customers can get locked-in. This can exacerbate market power, with consequent effects on prices and profitability. Thus, the FTC’s concern is that Microsoft, Amazon, and Google use these partnerships with OpenAI and Anthropic to create AI platforms dependent on Big Tech’s current products, particularly cloud service. This would force firms that want to use the new AI products to lock into Big Tech’s other offerings.
Markets, Technology, and Competition
Artificial Intelligence
However, there are several reasons to not be concerned that these partnerships may entrench Big Tech market power in either the AI or cloud computing markets. First, the AI market.
The number and diversity of the privately held AI firms’ is highlighted in Forbes’ sixth annual AI 50. The list of firms encompasses those that are AI model and AI business and consumer developers and those that provide AI-enabled general purpose search, AI app development tools, AI model development tools, AI-enabled coding autocompletion, and enterprise generative AI software, among other services. Underlying the growth of the broader AI market is a wide range of firms that are helping to develop and implement the technologies. Based on the data, Forbes notes that the popularity of apps like ChatGPT has led to a surge in new firms in the market developing and implementing advances in generative AI. This red-hot technology sector has also captured the attention of investors with a significant amount of fundraising.
Based on the data reported by Forbes, below we highlight a few of the firms that appear particularly important for the evolution of the technologies are (firm name, technology, funding, year founded are noted in parentheses): Adept (AI model developer; $415 million; 2022); Anthropic (AI model developer; $7.7 billion; 2020); Character AI (Consumer chatbot app; $193 million; 2021);; Figure AI (Autonomous humanoid robots; $754 million; 2022); Glean (Enterprise search engine; $360 million; 2019); Lang Chain (AI app development tools; $35 million; 2023); OpenAI (AI model developer; $11.3 billion; 2015); Perplexity (General purpose search app; $102 million; 2022); Pinecone (Database software; $138 million; 2019); Replicate (AI app deployment software; $60 million; 2019); Scale AI (Data labeling and software; $600 million; 2016); Together AI (AI model development tools; $229 million; 2022); and Unstructured (AI app development tools; $65 million; 2022). As one can tell, there are a fair number of firms competing in the relevant AI markets.
While Anthropic and OpenAI have the most funding, the list of existing firms reveals a dynamic market. Just because at the current time some firms may have lower levels of funding, it does not imply that they cannot get significant additional funding in coming periods. Further, one firm having better technology today does not imply that other firm(s) cannot provide leading edge technologies to the market. Forbes notes the rapid growth of AI firms in their list. Finally, tech firms like IBM and Salesforce have recently developed AI partnerships with Mistral, Slalom and Workday. This signals that the market offers many firms with competing technologies.
Cloud Computing
The FTC’s other concern is that the Big Tech firms use these AI partnerships to capture or protect market share for their cloud services, upon which most of AI products rely. The top three providers of cloud services are Amazon Web Services (AWS), Microsoft (Azure), and Google Cloud Platform (GCP), coincidentally the three Big Tech firms most involved in these partnerships. AWS will become Anthropic’s primary cloud provider. AWS consumer and business data will help Anthropic develop its next generation technologies. Further, Anthropic will use the AWS Trainium and Inferentia chips to develop future foundation models. While Azure has been the primary cloud provider for OpenAI, increasingly we are seeing multi-cloud partnerships. For example, Oracle’s partnership with OpenAI and Google. The market is undergoing a dynamic change in partnerships that are likely to fundamentally alter innovation in AI markets. Given the size and financial strengths of the bigger Tech firms, it is inevitable that the evolving AI technologies from different firms and these evolving partnerships will equilibrate the playing field in cloud markets.
Estimates show the following approximate market shares for cloud computing: Amazon (31%), Microsoft (25%), and Google (11%). By no means do we have a situation resembling a monopoly in this market. Google, with the lowest market share of the big three, is a formidable current competitor with deep pockets and has the ability to gain market share in the future. In addition to the above three, smaller competitors such as Salesforce, IBM and Oracle account for market shares of approximately 2-4% each. While these firms have low current market shares, they are among the fastest growing cloud vendors. Further, each of the smaller three are big technology firms with significant amounts of capital to invest in AI if they so desire (similar to the investments and partnerships by Microsoft, Amazon and Google). Among the firms with lower market shares such as IBM and Salesforce, there is evidence that they are forming new AI partnerships to increase their market presence. These include, for example, IBM’s partnership with Mistral, and Salesforce’s partnerships with Slalom and Workday. In conclusion, a fairly competitive market for cloud services means that these Big Tech firms are unlikely to be able to use these partnerships to capture greater market share.
Role of Potential Competition
To evaluate the competitiveness of the burgeoning AI market, we also need to consider the barriers to entry and growth prospects for startups: the high risk and funding required. It is well known that failure rates for new firms are quite high. This implies that, ex ante, an entrant must carefully assess both the likelihood of success and the fraction of entry cost that may be non-recoverable, or sunk costs. Examples of sunk costs include investments in R&D and related expenditures, and advertising and promotional expenditures needed to establish a foothold in the market. The common thread for these expenditures is that they are essential for entry into specific types of markets and subsequent success, and there is no meaningful resale value. We distinguish between fixed (or overhead) costs versus sunk costs. It is important to note that high fixed costs do not necessarily imply high sunk costs as many types of investments may have fair resale value.
While AI investment expenditures are substantial for firms like OpenAI, Anthropic, among others noted above, it is far from clear what percentage of these expenditures are truly sunk. AI firms develop proprietary code. In the event a firm wants to exit the market, it is fair to assume that the proprietary software has considerable market (resale) value – have low sunk costs. The fact that the technology is proprietary and likely has fair resale value (low sunk costs) implies that the market is meaningfully contestable. This facilitates the market dynamics involving entry of AI startups and new investments in these firms. The data presented above on the different types of firms currently in the market and the significant investments in them clearly point to the evolving market being relatively contestable.
In addition to the contestability argument noted above, Richard Posner and William Landes outlined in 1980 a framework for measuring potential competition. Their central thesis was that under certain conditions, foreign firms may be able to divert output to domestic markets relatively quickly, imposing a competitive discipline. While their focus was on imports coming into domestic markets, the spirit of their analysis is general and can be applied to a wide range of antitrust investigations and the entry, exit and growth dynamics of firms into specific markets. In our context, the essence of the Posner and Landes approach would be to evaluate the response of actual and potential firms if the economic conditions in that market change–such as rapid growth of the market, greater adoption of the technology, and increase in future profitability. The data we presented above on the different types of firms currently in the market, the year they were founded, and the implied overall market dynamics clearly point to the evolving AI market to be currently and potentially competitive.
The combination of the sunk costs and contestability effects, potential competition, and the data presented in Forbes paints a picture of the AI market and technologies that is quite dynamic.
Concluding remarks
The United States has a long history of landmark antitrust interventions. These include, for example, Standard Oil (1911), IBM (1969), AT&T (1982), and Microsoft (2001). While the markets and the specifics of anticompetitive conduct varied, the antitrust interventions fundamentally changed the future path of markets and technologies.
FTC Chair Khan noted recently regarding the AI investigations that “It is absolutely part of an effort to make sure we’re asking questions in a timely way. And being able to issue-spot potential problems at the inception rather than years and years and years later, when problems are deeply baked in and much more difficult to rectify.”
The FTC’s investigative actions in the generative AI markets raise important questions. First, there are at least three firms that have significant market share and deep pockets that are competing: Microsoft, Google and Amazon. Each has extensive resources to invest, form new partnerships, acquire startups and compete. In addition to these three, IBM, Salesforce and Oracle also compete. If there were a single firm with a dominant market share, the analysis and concerns would be different. Second, aside from OpenAI and Anthropic, there appears to be a meaningful number of players showing growth and potential for investments. This is different from situations where the markets show a lack of dynamism. Third, multiple firms, big and small, are competing in cloud services. This competition will improve prices and service offerings. Fourth, strong antitrust actions by FTC at this stage could potentially reduce investments in AI technologies and slow the growth of the innovative sector. In conclusion, there is little reason to believe that these Big Tech-AI startup partnerships are currently producing anticompetitive effects, and the government needs to be careful not to damage growth and prospects in this frontier technology.
Author Disclosure: the author reports no conflicts of interest. You can read our disclosure policy here.
Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.
