Contents

The CMA Should Not Intervene In the Amazon-Anthropic Partnership. 1

The Amazon-Anthropic Partnership Does Not Harm Competition. 2

The Amazon-Anthropic Partnership Has Had Pro-competitive Effects 2

Conclusion. 3

Endnotes 3

On behalf of the Information Technology and Innovation Foundation (ITIF), thank you for the opportunity to provide comments to the Competition and Markets Authority (CMA) on the Amazon-Anthropic partnership.[1]

ITIF is an independent, nonprofit, nonpartisan research and educational institute that has been recognized repeatedly as the world’s leading think tank for science and technology policy. ITIF’s mission is to formulate and promote policy solutions that accelerate innovation and boost productivity to spur growth, opportunity, and progress. ITIF’s Center for Data Innovation focuses on the intersection of data, technology, and public policy.

These comments make three points. First, the Amazon-Anthropic partnership does not meet the criteria set for a “relevant merger situation” as set forth in the Enterprise Act 2002, and thus the CMA should not intervene. Second, even if the CMA believes that the partnership reaches the threshold to create a relevant merger situation, the partnership does not have a negative impact on competition. Third, any potentially negative impacts on competition would be justified by the pro-competitive impact of this partnership.

The CMA Should Not Intervene In the Amazon-Anthropic Partnership

The first question CMA is considering is “whether it is or may be the case that the Amazon/Anthropic partnership has resulted in the creation of a relevant merger situation under the merger provisions of the Enterprise Act 2002.”[2] The Enterprise Act 2002 plainly states that “a relevant merger situation has been created if…two or more enterprises have ceased to be distinct enterprises…”[3] In the Amazon-Anthropic partnership, this criterion is clearly unmet as Amazon does not have corporate control over Anthropic (or vice versa). Both organizations continue to operate independently, maintaining distinct governance and operational structures. The partnership functions as a strategic collaboration, not as an integration of business operations under common control, which stands at the heart of the concept of a merger.

The publicly reported details of this partnership fully support this conclusion. Amazon’s $4 billion investment in Anthropic has given it a minority ownership stake in the startup.[4] Amazon is not the only major investor: Google has invested $2 billion in Anthropic.[5] And Amazon does not have a director or observer role on the Anthropic board.[6] In fact, Anthropic has a unique governance structure, designed to allow the company to balance shareholder profits with its goal of building advanced AI safely for the benefit of humanity. The company is a Public Benefit Corporation (PBC) incorporated in Delaware. In addition to the board of directors for the PBC, there is the Anthropic Long-Term Benefit Trust, an independent body consisting of five trustees with the authority to elect and remove board members of the PBC. Amazon has no reported role in either the PBC or the Trust, which means it has no meaningful control of the company.[7]

The Amazon-Anthropic Partnership Does Not Harm Competition

The partnership between Amazon and Anthropic has not had a negative impact on competition in the UK (or elsewhere) in terms of reduced innovation, consumer welfare, or choice.

With respect to vertical effects, the partnership between Amazon and Anthropic has not, and will not, prevent competitors from using Anthropic’s models or Amazon’s cloud services. For example, Google recently announced the availability of Anthropic’s enterprise large language models (LLMs) in Google Cloud, including Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku.[8] Developers can choose which cloud platform best fits their needs. For example, Amazon offers features like Amazon CodeWhisperer, an AI-coding companion to help developers increase their productivity.[9] In addition, Anthropic does not have exclusive access to Amazon’s cloud services, which Amazon offers widely and seeks to fully monetize. Indeed, Amazon Bedrock, Amazon’s fully managed service for access to foundation models, offers customers access to a broad array of AI models, including from Cohere, Meta, Mistral AI, and Stability AI.

The partnership does not lessen Amazon’s incentive to enter the AI space: The Anthropic investment is complementary to—not at the expense of—Amazon’s massive investments in AI innovation. For example, Amazon also offers access to its own models, such as Amazon Titan, through Amazon Bedrock.[10] Amazon is also reportedly working on a 2-trillion parameter AI model of its own, which would rival any LLM on the market today.[11]

The Amazon-Anthropic Partnership Has Had Pro-competitive Effects

Far from reducing competition, the partnership between Amazon and Anthropic will likely prove to be pro-competitive and bring additional consumer benefits going forward. With Amazon’s investment, Anthropic can grow more quickly and expand its offerings—benefiting consumers. Amazon’s partnership with Anthropic also creates more competition between different AI models because developers can easily experiment with different models from different providers within the Bedrock platform and fine tune and deploy models using enterprise data already stored in Amazon Web Services (AWS).

The partnership also offers opportunities for more innovation and competition around the chips used for AI training and AI inference. Amazon offers compute instances using both NVIDIA chips and its own custom chips, AWS Trainium and AWS Inferentia.[12] Closer collaboration between Amazon and Anthropic will allow Amazon to optimize chips for its customers’ AI workloads and Anthropic to optimize its AI models for these chips. The synergies from this collaboration could result in faster, more efficient AI models—efficiency which could be passed on to customers and creates a more competitive market. Neither company could achieve this type of advancement on its own, which is why these types of partnerships are useful for fostering competition.

Conclusion

Given that the Amazon-Anthropic partnership does not create a relevant merger situation, does not harm competition, and has a pro-competitive impact on the market, the CMA should not intervene. Intervention would have a deleterious impact on digital innovation not in the AI sector. Moreover, given that the partnership involves two U.S.-based companies, the CMA should be cautious when considering extraterritorial enforcement of its authority, especially in hotly contested emerging markets like AI, because intervening could create the impression that the UK is attempting to stifle foreign competition by sabotaging successful technological partnerships. Instead, in the absence of evidence of harm and any clear impact on the UK market, the CMA should defer to U.S. regulators to consider whether to investigate any antitrust concerns arising from this partnership.

Thank you for taking this analysis into account in your review.

Endnotes