Table Of Links
1. Introduction and Motivation
- Relevant literature
- Limitations
2. Understanding the AI Supply Chain
- Background history
- Inputs necessary for development of frontier AI models
- Steps of the supply chain
3. Overview of the integration landscape
- Working definitions
- Integration in the AI supply chain
4. Antitrust in the AI supply chain
- Lithography and semiconductors
- Cloud and AI
- Policy: sanctions, tensions, and subsidies
- Synergies
- Strategically harden competition
- Governmental action or industry reaction
- Other reasons
6. Closing remarks and open questions
- Selected Research Questions
6. Closing remarks and open questions
We have the challenge of understanding the dynamics of the quickly changing AI industry. As with the rest of the hardware industry and software industries (e.g., Tirole, 2023), there are significant barriers to entry associated with relatively low marginal costs in multiple steps of the AI supply chain which may give companies substantial market power.
Additionally, we may observe network effects in the industry as well as interoperability issues. The relevance of each of these aspects for each of the steps of the AI supply remains an open question to be sure to what extent the economic theoretical framework needs to be readapted — or reinvented — to fully understand the dynamics of these new industries, similarly to what happened with the rise of the digital economy. It is clear, however, that the effectiveness of AI regulatory proposals is likely to be heavily impacted by the structure of the AI industry.
This final section aims to lay down open questions that can be useful to best understand these implications, paying special attention to vertical relationships. This complements already proposed research agendas such as Siegmann (2023) and Chapter 4 of Winter et al (2021), which respectively pose questions for economists and lawyers regarding AI governance and safety. Many of the regulatory implications of having more vertical integration in these industries are related to possible trade-offs between competition and safety that can appear in different contexts in the industry.
For instance, if we accept the argument that we should decelerate AI development to allow time to understand and address its associated risks, then employing antitrust policies to increase industry competition might be counterproductive. As this would typically fall outside the mandate of antitrust authorities, this may suggest a demand for structural remedies by regulators, besides behavior remedies. In this sense, there may be tension in looking to enhance short-term consumer welfare and economic efficiency with the mitigation of risks that may arise from frontier AI systems.
This trade-off may be less salient if we think about long-term effects on the well-being of consumers. There are, additionally, national security concerns that may play a role in this. Foster & Arnold (2020) already elaborated how there may be tension between breaking up big tech companies because of their market power and national security. As O'Keefe (2020) pointed out, this attention between focusing on national security and economic efficiency has already been a central part of significant litigations such as AT&T.
These concerns also played a role in the block of the merger of Qualcomm-Broadcom. Noticeably, the regulatory implications of having more horizontal integration can be substantially different from that of having more vertical integration. Hua & Belfield (2020) and O’Keeffe (2021) already explored these horizontal antitrust considerations.
However, considerations for vertical integration can significantly differ. Antitrust authorities typically exhibit more leniency toward vertical integration than horizontal, recognizing its potential to boost welfare through efficiency gains. Vertically integrated companies often benefit from shared capabilities across similar economic activities, gain efficiencies from economies of scale, and arguably tend to invest more in research and development as well as in safety measures. It is challenging to assess if some implications are overall welfare enhancing, especially when we consider that AI is a dual-use technology that creates externalities.
The amount and kind of integration in the AI supply chain may be decisive in how effective different regulatory proposals for frontier AI models are. Antitrust policy will probably impact or complement regulation. Towards effective antitrust and regulatory intervention, there are major questions about the dynamics of the industry and its industry that need to be tackled.
6.1 Selected Research Questions
6.1.1 How might the prevailing market structure shape the trajectory of AI industry advancements? Balancing safety and competition in the AI industry is a complex policy challenge, as these goals can sometimes be at odds. The effects of vertical integration on competition and R&D in the AI supply chain are unclear, requiring specific empirical research. While there are theoretical studies on arms races in the horizontal development of AI (e.g., Armstram, Bostrom, and Shalman, 2013), specific research on vertical relationships and empirical studies in either domain is lacking.
While increased R&D investment and faster technological progress are generally positive, the implications are ambiguous in the AI sector due to potential risks associated with advanced AI technologies. Vertical integration could encourage more investment in safety, especially if dominant firms prioritize it (Jensen, Emery-Xu, and Tragery, 2023). This could be beneficial for managing race dynamics among leading AI firms and fostering a safety-focused industry. However, less competition might lead to higher prices, reduced innovation, and fewer consumer choices, highlighting the need to carefully weigh these trade-offs.
6.2 How current market structure within the AI supply chain affect regulatory proposals? Increased vertical integration in companies can make operational data less transparent, as integrated companies can closely control the flow of information. In the AI supply chain, this can include the number of AI accelerators purchased and the size of training runs. This opacity may complicate the task for external stakeholders like regulators and consumers in accurately monitoring and assessing the firm's activities that may be necessary to track for effective compute oversight.
For example, information about Google's Tensor Processing Units (TPUs) is more restricted compared to Nvidia's Graphics Processing Units (GPUs). Increased vertical integration could, however, also mean that companies are more readily able to adhere to strict rules on data privacy and cybersecurity. Having a generally more concentrated AI supply chain may also help to make coordinated efforts, such as, e.g., the MLCommons and Partnership for AI. Here, vertical integration could potentially provide integrated firms with the ability to establish and influence industry standards more easily.
By controlling multiple stages of the supply chain, these companies can align their practices and technologies to create standards, which may be positive for AI governance concerns regarding standard-setting. In the context of compute oversight, this may be valuable to quickly adopt safety standards for AI accelerators, such as in-hardware monitors and shutdown mechanisms. Conversely, it might increase the risk of collusive behavior, as observed in the DRAM market. This could stir the antitrust regulator’s concerns and this apprehension could inhibit potentially advantageous agreements between AI labs.
Finally, more concentration could also mean a facilitated path for regulatory capture (see, e.g., Moshary and Slattery, 2023). Drawing from the experiences of regulation development in industries such as electricity, civil aviation, and the banking sector can provide useful insights for shaping AI regulation. Lessons learned from these industries can help identify effective regulatory approaches, understand potential challenges, and inform the development of appropriate frameworks for the AI sector. Koesller and Schuett (2023) offer an overview of how risk assessments in other industries may impact this.
Insights from the literature regarding third-party reporting in supply chains can inform the development of AI regulations. This notion that vertically integrated firms are more capable of dodging taxation seems relatively established in the literature (see, e.g, De Paula and Scheinkman, 2010, Poremanz, 2015, Singh, 2020), but less so in regulation that demands the disclosure of key information. Lessons learned from public finance literature can contribute to the formulation of effective reporting and auditing mechanisms for AI development and deployment.
6.3 Will structural remedies be necessary to make effective regulatory frameworks in the AI industry? As we discussed, the trade-offs between different forms of market integration — horizontal, vertical, and conglomerate — vary significantly. A key area for future research is identifying the optimal market structure for the AI supply chain that aligns with various policy objectives. For example, having a horizontally integrated market could potentially slow down AI development and reduce race dynamics at the potential expense of power concentration and less product diversity, while a less vertically integrated market might facilitate regulatory oversight.
Another issue to consider is whether unbundling principles, similar to those applied in the electrical and railway sectors, should be implemented in the frontier AI industry. The impact of such policies on consumer welfare is mixed and requires further examination. This is particularly relevant for enhancing third-party reporting mechanisms. In terms of regulatory oversight, it would be beneficial to assess which types of integration hinder or support compute oversight.
For instance, identifying clear bottlenecks where vertical integration should be avoided could enable more effective monitoring by regulators. Comparing incentive-compatible self-reports versus mandatory external audits can reveal which approach is more effective, depending on the level of integration in the industry. As Athey said “hopefully we will learn that we need to learn more quickly this time than we have in the past” (Centre for Economic Policy Research, 2023)
References
Acemoglu, D. (2021). Harms of AI. Social Science Research Network. https://doi.org/10.2139/ssrn.3922521
Acemoglu, D., Griffith, R., Aghion, P., & Zilibotti, F. (2010). Vertical integration and technology: Theory and evidence. Journal of the European Economic Association, 8(5), 989–1033.
Acemoglu, D., & Lensman, T. (2023). Regulating transformative technologies (Working Paper No. 31461). National Bureau of Economic Research. https://doi.org/10.3386/w31461
Allied Market Research. (2023). Cloud machine learning market by product type (Private clouds, Public clouds and Hybrid cloud), organization size (Small & Medium-sized Enterprises (SMEs), and Large Enterprises), and industry vertical (BFSI, Life Sciences & Healthcare, Retail, Telecommunication, Government & Defense, Manufacturing, Energy & Utilities, and Others): Global opportunity analysis and industry forecast, 2021–2031. https://www.alliedmarketresearch.com/cloud-machine-learning-market-A09569
Allen, G. C. (2023, May 3). China’s new strategy for waging the microchip tech war [Report]. Center for Strategic and International Studies. https://www.csis.org/analysis/chinas-new-strategy-waging-microchip-tech-war
Amazon. (2023, October 16). What you need to know about the AWS AI chips powering Amazon's partnership with Anthropic. https://www.aboutamazon.com/news/aws/what-you-need-to-know-about-the-aws-ai-chips-powering-amazons-partnership-with-anthropic
Amazon Press Center. (2023, September 25). Amazon and Anthropic announce strategic collaboration to advance generative AI. https://press.aboutamazon.com/2023/9/amazon-and-anthropic-announce-strategic-collaboration-to-advance-generative-ai
Amazon Web Services. (n.d.). Inovação com silício da AWS. Retrieved June 1, 2024, from https://aws.amazon.com/pt/silicon-innovation/
AMD. (2009). Intel antitrust rulings. Retrieved June 1, 2024, from https://www.amd.com/en/legal/notices/antitrust-ruling.html
Anthropic. (2023, February 3). Anthropic partners with Google Cloud. https://www.anthropic.com/news/anthropic-partners-with-google-cloud
Armour, H. O., & Teece, D. J. (1980). Vertical integration and technological innovation. The Review of Economics and Statistics, 62(3), 470–474.
Armstrong, S., Bostrom, N., & Shulman, C. (2016). Racing to the precipice: A model of artificial intelligence development. AI & Society, 31(2), 201–206.
ASML. (2012, October 17). ASML to acquire Cymer to accelerate development of EUV technology [Press release]. Retrieved from https://web.archive.org/web/20140308034237/http://www.asml.com/asml/show.do?ctx=5869&rid=47397
ASML. (2016, November 3). ZEISS and ASML strengthen partnership for next generation of EUV lithography [Press release]. https://www.asml.com/en/news/press-releases/2016/zeiss-and-asml-strengthen-partnership-for-next-generation-of-euv-lithography
ASML. (2019, January 23). Nikon, ASML and Carl Zeiss sign agreement to settle all litigation [Press release]. https://www.asml.com/en/news/press-releases/2019/nikon-asml-and-carl-zeiss-sign-agreement-to-settle-all-litigation
ASML. (n.d.). About ASML: HMI. Retrieved June 1, 2024, from https://www.asml.com/en/company/about-asml/hmi
Associated Press. (n.d.). Dutch government to restrict sales of processor chip tech. Retrieved from https://apnews.com/article/dutch-government-restricts-chip-tech-sales-2023
Barrera, C., & Waldman, M. (2019). Vertical integration, secrecy, and high tech industries. Unpublished manuscript.
Belfield, H., & Hua, S.-S. (2022, August 19). Compute and antitrust: Regulatory implications of the AI hardware supply chain, from chip design to cloud APIs. Verfassungsblog. https://verfassungsblog.de/compute-and-antitrust/
Bengio, Y., Hinton, G., Yao, A., Song, D., Abbeel, P., Darrell, T., Harari, Y. N., Zhang, Y.-Q., Xue, L., Shalev-Shwartz, S., Hadfield, G., Clune, J., Maharaj, T., Hutter, F., Baydin, A. G., McIlraith, S., Gao, Q., Acharya, A., Krueger, D., Dragan, A., Torr, P., Russell, S., Kahneman, D., Brauner, J., & Mindermann, S. (2024). Managing extreme AI risks amid rapid progress. Science, 384(6698), 842–845. https://doi.org/10.1126/science.adn0117
Bloomberg. (2023, February 3). Google invests almost $400 million in ChatGPT rival Anthropic. Retrieved from https://www.bloomberg.com/news/articles/2023-02-03/google-invests-almost-400-million-in-ai-startup-anthropic
Boa, I., Elliott, M., & Foster, D. (2023). A capability approach to merger review. Unpublished manuscript.
Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx, S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., Brynjolfsson, E., Buch, S., Card, D., Castellon, R., Chatterji, N. S., Chen, A. S., Creel, K. A., Davis, J., Demszky, D., Donahue, C., Doumbouya, M., Durmus, E., Ermon, S., Etchemendy, J., Ethayarajh, K., Fei-Fei, L., Finn, C., Gale, T., Gillespie, L. E., Goel, K., Goodman, N. D., Grossman, S., Guha, N., Hashimoto, T., Henderson, P., Hewitt, J., Ho, D. E., Hong, J., Hsu, K., Huang, J., Icard, T. F., Jain, S., Jurafsky, D., Kalluri, P., Karamcheti, S., Keeling, G., Khani, F., Khattab, O., Koh, P. W., Krass, M. S., Krishna, R., Kuditipudi, R., Kumar, A., Ladhak, F., Lee, M., Lee, T., Leskovec, J., Levent, I., Li, X. L., Li, X., Ma, T., Malik, A., Manning, C. D., Mirchandani, S. P., Mitchell, E., Munyikwa, Z., Nair, S., Narayan, A., Narayanan, D., Newman, B., Nie, A., Niebles, J. C., Nilforoshan, H., Nyarko, J. F., Ogut, G., Orr, L., Papadimitriou, I., Park, J. S., Piech, C., Portelance, E., Potts, C., Raghunathan, A., Reich, R., Ren, H., Rong, F., Roohani, Y. H., Ruiz, C., Ryan, J., Ré, C., Sadigh, D., Sagawa, S., Santhanam, K., Shih, A., Srinivasan, K. P., Tamkin, A., Taori, R., Thomas, A. W., Tramèr, F., Wang, R. E., Wang, W., Wu, B., Wu, J., Wu, Y., Xie, S. M., Yasunaga, M., You, J., Zaharia, M. A., Zhang, M., Zhang, T., Zhang, X., Zhang, Y., Zheng, L., Zhou, K., & Liang, P. (2021). On the opportunities and risks of foundation models. ArXiv. https://crfm.stanford.edu/assets/report.pdf
Brundage, M., Avin, S., Clark, J., Toner, H., Eckersley, P., Garfinkel, B., Dafoe, A., Scharre, P., Zeitzoff, T., Filar, B., Anderson, H., Roff, H., Allen, G. C., Steinhardt, J., Flynn, C., Ó hÉigeartaigh, S., Beard, S., Belfield, H., Farquhar, S., Lyle, C., Crootof, R., Evans, O., Page, M., Bryson, J., Yampolskiy, R., & Amodei, D. (2018). The malicious use of artificial intelligence: Forecasting, prevention, and mitigation. ArXiv. https://arxiv.org/abs/1802.07228
Buchanan, B. (2020). The AI triad and what it means for national security strategy. Center for Security and Emerging Technology. https://cset.georgetown.edu/research/the-ai-triad-and-what-it-means-for-national-security-strategy
Campbell, J., Goldfarb, A., & Tucker, C. (2015). Privacy regulation and market structure. Journal of Economics & Management Strategy, 24(1), 47–73. https://doi.org/10.1111/jems.12079
Carugati, C. (2023). The antitrust privacy dilemma. European Competition Journal, 19(2), 167–190.
Centre for Economic Policy Research. (2023, July 1–2). GEN AI & market power: What role for antitrust regulators? [Webinar]. CEPR Competition Policy RPN. https://cepr.org/system/files/2023-07/Competition%20Policy%20RPN%20Webinar_Gen%20AI%20%26%20Market%20Power_12%20July%202023_Transcript.pdf
Chellapilla, K., Puri, S., & Simard, P. (2006, October). High performance convolutional neural networks for document processing. In Tenth International Workshop on Frontiers in Handwriting Recognition. Suvisoft.
Chen, J., Elliott, M., & Koh, A. (2023). Capability accumulation and conglomeratization in the information age. Journal of Economic Theory, 210, 105647.
Chiang, M. H. (2023). Taiwan Semiconductor Manufacturing Company: A key chip in the global political economy. East Asian Policy, 15(1), 36–46.
Chiang, W. L., Zheng, L., Sheng, Y., Angelopoulos, A. N., Li, T., Li, D., ... & Stoica, I. (2024). Chatbot arena: An open platform for evaluating LLMs by human preference. arXiv preprint arXiv:2403.04132.
Chowdhry, A. (2014, January 27). Google to acquire artificial intelligence company DeepMind. Forbes. https://www.forbes.com/sites/amitchowdhry/2014/01/27/google-to-acquire-artificial-intelligence-company-deepmind/
CNBC. (2023, April 8). Microsoft’s $13 billion bet on OpenAI carries huge potential along with plenty of uncertainty. https://www.cnbc.com/2023/04/08/microsofts-complex-bet-on-openai-brings-potential-and-uncertainty.html
CNET. (2005, March 8). Japan antitrust watchdog bites Intel: Chipmaker mulls a response to ruling, which accuses the company of offering rebates to PC makers that snubbed its rivals. https://www.cnet.com/tech/tech-industry/japan-antitrust-watchdog-bites-intel/
Computerworld. (2022, April 14). Samsung beats Intel as top semiconductor seller as overall demand rises. https://www.computerworld.com/article/1620848/samsung-beats-intel-as-top-semiconductor-seller-as-overall-demand-rises.html
Cunningham, C., Ederer, F., & Ma, S. (2021). Killer acquisitions. Journal of Political Economy, 129(3), 649–702.
Davies, P. (2024, March 18). Apple is reportedly in talks with Google’s Gemini to use its AI in iPhones. Euronews. https://www.euronews.com/next/2024/03/18/apple-is-reportedly-in-talks-with-googles-gemini-to-use-its-ai-in-iphones
de Paula, Á., & Scheinkman, J. A. (2010). Value-added taxes, chain effects, and informality. American Economic Journal: Macroeconomics, 2(4), 195–221.
Design Reuse. (2020, June 9). Arteris IP advances onto list of top 15 semiconductor IP vendors. Design & Reuse. https://www.design-reuse.com/news/48119/arteris-ip-top-15-semiconductor-ip-vendors.html
Dey, V. (2023, April 20). Google consolidates AI research labs into Google DeepMind to compete with OpenAI. VentureBeat. https://venturebeat.com/ai/google-consolidates-ai-research-labs-into-google-deepmind-to-compete-with-openai/
Erdil, E., & Besiroglu, T. (2023). Algorithmic progress in computer vision. ArXiv. http://arxiv.org/abs/2212.05153
Elliott, M., & Golub, B. (2022). Networks and economic fragility. Annual Review of Economics, 14, 665–696.
Eloundou, T., Manning, S., Mishkin, P., & Rock, D. (2023). GPTs are GPTs: An early look at the labor market impact potential of large language models. ArXiv. https://arxiv.org/abs/2303.10130
Erdil, E., & Besiroglu, T. (2023). Explosive growth from AI automation: A review of the arguments. ArXiv. https://arxiv.org/abs/2309.11690
European Commission. (2010, May 19). Antitrust: Commission fines DRAM producers € 331 million for price cartel; reaches first settlement in a cartel case [Press release]. https://ec.europa.eu/commission/presscorner/detail/en/ip_10_586
European Commission. (2023). European Chips Act. A Europe fit for the digital age. https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/europe-fit-digital-age/european-chips-act_en
Federal Trade Commission. (2010, August 4). FTC settles charges of anticompetitive conduct against Intel [Press release]. https://www.ftc.gov/news-events/news/press-releases/2010/08/ftc-settles-charges-anticompetitive-conduct-against-intel
Federal Trade Commission. (2022, February 14). Nvidia/Arm, in the matter of. https://www.ftc.gov/legal-library/browse/cases-proceedings/2110015-nvidiaarm-matter
Fernández Llorca, D., Charisi, V., Hamon, R., Sánchez, I., & Gómez, E. (2023, March). Liability regimes in the age of AI: A use-case driven analysis of the burden of proof. Journal of Artificial Intelligence Research, 76, 613–644. https://doi.org/10.1613/jair.1.14565
Forbes. (2023, September 28). Nvidia’s French offices raided over antitrust concerns, report says. https://www.forbes.com/sites/antoniopequenoiv/2023/09/28/nvidias-french-offices-raided-over-antitrust-concerns-report-says/
Foster, D., & Arnold, Z. (2020, May). Antitrust and artificial intelligence: How breaking up big tech could affect the Pentagon’s access to AI. Center for Security and Emerging Technology. https://cset.georgetown.edu/publication/antitrust-and-artificial-intelligence-how-breaking-up-big-tech-could-affect-pentagons-access-to-ai/
Gehlhaus, D., Ryseff, J., & Corrigan, J. (2023, August). The race for U.S. technical talent: Can the DOD and DIB compete? Center for Security and Emerging Technology. https://cset.georgetown.edu/publication/the-race-for-us-technical-talent/
Gibbs, S. (2014, January 27). Google buys UK artificial intelligence startup DeepMind for £400m. The Guardian. https://www.theguardian.com/technology/2014/jan/27/google-acquires-uk-artificial-intelligence-startup-deepmind
Goldfarb, A., Taska, B., & Teodoridis, F. (2023). Could machine learning be a general-purpose technology? A comparison of emerging technologies using data from online job postings. Research Policy, 52(1), 104653. https://doi.org/10.1016/j.respol.2022.104653
Google Cloud. (n.d.). Cloud Tensor Processing Units (TPUs). Retrieved June 1, 2024, from https://cloud.google.com/tpu
Greene, T. (2023, April 18). Microsoft is developing its own AI chip to power ChatGPT: Report. Cointelegraph. https://cointelegraph.com/news/microsoft-is-developing-its-own-ai-chip-to-power-chatgpt-report
Hadfield, G. K., & Clark, J. (2023). Regulatory markets: The future of AI governance. ArXiv. https://arxiv.org/abs/2304.04914
Higgins, B. W. (2024). Legal elements of an AI regulatory permit program. In J. B. Bullock et al. (Eds.), The Oxford Handbook of AI Governance. Oxford University Press. https://doi.org/10.1093/oxfordhb/9780197579329.013.18
Hoffmann, J., Borgeaud, S., Mensch, A., Buchatskaya, E., Cai, T., Rutherford, E., ... & Sifre, L. (2022). Training compute-optimal large language models. arXiv preprint arXiv:2203.15556.
Hua, S. S., & Belfield, H. (2020). AI & antitrust: Reconciling tensions between competition law and cooperative AI development. Yale Journal of Law & Technology, 23, 415.
Hugging Face. (2021, March 23). The partnership: Amazon SageMaker and Hugging Face. https://huggingface.co/blog/the-partnership-amazon-sagemaker-and-hugging-face
Inflection AI. (2023, June 23). Inflection-1. https://inflection.ai/assets/Inflection-1.pdf
Intel. (n.d.). AI hardware solutions. https://www.intel.com/content/www/us/en/artificial-intelligence/hardware.html
Jadhav, A., & Davis, J. (2023, February 21). Anthropic uses Google Cloud for potential ChatGPT rival; a prelude to M&A? EdgeIR. https://www.edgeir.com/anthropic-uses-google-cloud-for-potential-chatgpt-rival-a-prelude-to-ma-20230221
Jamison, M. (2014, April 11). Defining relevant markets in evolving industries. Public Utility Research Center, University of Florida. https://bear.warrington.ufl.edu/centers/purc/docs/papers/1317_Jamison_Defining%20Relevant%20Markets%20in%20Evolving%20Industries%20Final.pdf
Japan Fair Trade Commission. (2005, March 8). The JFTC rendered a recommendation to Intel K.K. https://www.jftc.go.jp/en/pressreleases/yearly_2005/mar/2005_mar_8_files/2005-Mar-8.pdf
Jensen, M., Emery-Xu, N., & Trager, R. (2023). Industrial policy for advanced AI: Compute pricing and the safety tax. ArXiv. https://arxiv.org/abs/2302.11436
Jon Peddie Research. (2023). GPU developments of 2023. TechWatch. https://www.jonpeddie.com/techwatch/gpu-developments-2023/
Jones, C. I. (2016). Life and growth. Journal of Political Economy, 124(2), 539–578. https://doi.org/10.1086/684750
Jones, C. I. (2023, November). The A.I. dilemma: Growth versus existential risk (Working Paper No. 31837). National Bureau of Economic Research. https://doi.org/10.3386/w31837
Kerr, W. R., & Robert-Nicoud, F. (2020). Tech clusters. Journal of Economic Perspectives, 34(3), 50–76.
Kim, J.-H., Yoo, S., & Kim, J.-Y. (2023, July 1). South Korea’s nationwide effort for AI semiconductor industry. Communications of the ACM. https://dl.acm.org/doi/10.1145/3587264
Kizuna: Linking Japan and the World. (2024). Japan’s pursuit of a game-changing technology and ecosystem for semiconductors. https://www.japan.go.jp/kizuna/2024/03/technology_for_semiconductors.html
Korinek, A., & Vipra, J. (2024). Market concentration implications of foundation models: The invisible hand of ChatGPT. Unpublished manuscript.
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 25, 1097–1105.
Lapedus, M. (2007, April 19). Samsung exec pleads guilty in DRAM scandal. EDN. https://www.edn.com/samsung-exec-pleads-guilty-in-dram-scandal-2/
LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521, 436–444. https://doi.org/10.1038/nature14539
Lee, R. S., Whinston, M. D., & Yurukoglu, A. (2021). Structural empirical analysis of contracting in vertical markets. In Handbook of Industrial Organization (Vol. 4, pp. 673–742). Elsevier.
McCarthy, J., Minsky, M. L., Rochester, N., & Shannon, C. E. (2006). A proposal for the Dartmouth summer research project on artificial intelligence, August 31, 1955. AI Magazine, 27(4), 12–12.
Möbius (Marius) Hobbhahn, L. Heim, & G. Aydos. (2023). Trends in machine learning hardware. Epoch. https://epochai.org/blog/trends-in-machine-learning-hardware
Michal S. Gal, & Aviv, O. (2020). The competitive effects of the GDPR. Journal of Competition Law & Economics, 16(3), 349–391. https://doi.org/10.1093/joclec/nhaa012
Mollick, E. (2023, July 15). How to use AI to do stuff: An opinionated guide. One Useful Thing. https://www.oneusefulthing.org/p/how-to-use-ai-to-do-stuff-an-opinionated
Mordor Intelligence. (n.d.). Semiconductor lithography equipment market size & share analysis – Growth trends & forecasts (2024–2029). https://www.mordorintelligence.com/industry-reports/semiconductor-lithography-equipment-market
Moshary, S., & Slattery, C. (2024). Consolidation and political influence in the auto retail industry. SSRN. https://ssrn.com/abstract=4832068
Motley Fool. (2023, March 22). TSMC's soaring market share: Is there room for competition like Intel and Samsung? https://www.fool.com/investing/2023/03/22/tsmcs-soaring-market-share-is-there-room-for-competition/
Mökander, J., Schuett, J., Kirk, H. R., et al. (2023). Auditing large language models: A three-layered approach. AI and Ethics. https://doi.org/10.1007/s43681-023-00289-2
Nellis, S. (2023, April 4). Google says its AI supercomputer is faster, greener than Nvidia A100 chip. Reuters. https://www.reuters.com/technology/google-says-its-ai-supercomputer-is-faster-greener-than-nvidia-2023-04-05/
Network World. (2008, January 10). Why Intel is under fire for alleged antitrust violations. Network World. https://www.networkworld.com/article/809762/data-center-why-intel-is-under-fire-for-alleged-antitrust-violations.html
NVIDIA. (2023, March 21). NVIDIA, ASML, TSMC and Synopsys set foundation for next-generation chip manufacturing. https://nvidianews.nvidia.com/news/nvidia-asml-tsmc-and-synopsys-set-foundation-for-next-generation-chip-manufacturing
O'Keefe, C. (2021, September 30). Antitrust-compliant AI industry self-regulation (LawAI Working Paper No. 2-2021). Institute for Law & AI; Centre for the Governance of AI.
O'Keefe, C. (2022). How will national security considerations affect antitrust decisions in AI? An examination of historical precedents. Future of Humanity Institute.
Oh, C.-J., & Yoon, Y.-H. (2023, October 11). Samsung Electronics secures new order for 3nm high-performance server chips. Pulse (Maeil Business News Korea). https://pulse.mk.co.kr/news/english/10847149
OpenAI. (2019, July 22). Microsoft invests in and partners with OpenAI to support us building beneficial AGI. https://openai.com/index/microsoft-invests-in-and-partners-with-openai/
OpenAI. (2023, January 23). OpenAI and Microsoft extend partnership. https://openai.com/index/openai-and-microsoft-extend-partnership/
Oracle. (2024, March 12). Cohere and Oracle partnership brings generative AI solutions to enterprises. https://www.oracle.com/customers/cohere/
Pilz, K. (2023, July). An assessment of data center infrastructure’s role in AI governance. https://www.konstantinpilz.com/data-centers/assessment
Pilz, K., & Heim, L. (2023). Compute at scale—A broad investigation into the data center industry. arXiv preprint arXiv:2311.02651.
Pomeranz, D. (2015). No taxation without information: Deterrence and self-enforcement in the value added tax. American Economic Review, 105(8), 2539–2569.
Proia, S. (2019, April 18). From sand to wafers. SemiEngineering. https://semiengineering.com/from-sand-to-wafers/
Reuters. (2018, March 14). Timeline: Broadcom-Qualcomm saga comes to an abrupt end. Reuters. https://www.reuters.com/article/us-qualcomm-m-a-broadcom-timeline-idUSKCN1GQ22N/
Reuters. (2021, December 6). EU regulators pause investigation into Nvidia, ARM deal. Reuters. https://www.reuters.com/markets/deals/eu-regulators-pause-investigation-into-nvidia-arm-deal-2021-12-06/
Reuters. (2023, April 18). Microsoft developing its own AI chip – The Information. Reuters. https://www.reuters.com/technology/microsoft-developing-its-own-ai-chip-information-2023-04-18/
Reuther, A., Michaleas, P., Jones, M., Gadepally, V., Samsi, S., & Kepner, J. (2022, September). AI and ML accelerator survey and trends. In 2022 IEEE High Performance Extreme Computing Conference (HPEC) (pp. 1–10). IEEE.
Rikap, C. (2023). Same end by different means: Google, Amazon, Microsoft and Meta’s strategies to organize their frontier AI innovation systems (CITYPERC Working Paper No. 2023-03). City, University of London.
Rossolillo, N. (2023, January 17). Microsoft just acquired a chip design start-up. Here's what semiconductor investors need to know. The Motley Fool. https://www.fool.com/investing/2023/01/17/microsoft-just-acquired-a-chip-design-startup-what/
Samsung. (n.d.). AI Center. Samsung Research. https://research.samsung.com/aicenter
Schaller, R. R. (1997). Moore's law: Past, present and future. IEEE Spectrum, 34(6), 52–59.
Schuett, J., Dreksler, N., Anderljung, M., McCaffary, D., Heim, L., Bluemke, E., & Garfinkel, B. (2023). Towards best practices in AGI safety and governance: A survey of expert opinion. ArXiv. https://arxiv.org/abs/2305.07153
Shavit, Y. (2023). What does it take to catch a Chinchilla? Verifying rules on large-scale neural network training via compute monitoring. arXiv preprint arXiv:2303.11341.
Shavit, Y. (2023). What does it take to catch a Chinchilla? Verifying rules on large-scale neural network training via compute monitoring. arXiv preprint arXiv:2303.11341.
Shy, O. (1996). Industrial organization: Theory and applications. MIT Press.
Shy, O. (2001). The economics of network industries. Cambridge University Press.
Siegmann, C. (2023, August 2). Economics and AI risk: Research agenda and overview (Version 0.1). MIT. https://static1.squarespace.com/static/647796224461f039c379f070/t/64caecc54f52f03aee4bb67a/1691020486983/research_agenda.pdf
Silicon UK. (2023, July 3). Microsoft-backed chatbot maker Inflection AI raises $1.3bn. https://www.silicon.co.uk/e-innovation/artificial-intelligence/inflection-funding-ai-519039
Singh, D. (2018, January). Merging to dodge taxes? Unexpected consequences of VAT adoption in India. In Proceedings. Annual Conference on Taxation and Minutes of the Annual Meeting of the National Tax Association (Vol. 111, pp. 1–5). National Tax Association.
Sitaraman, G., & Narechania, T. (n.d.). Antimonopoly tools for regulating artificial intelligence. Unpublished manuscript.
Stafford, E., & Trager, R. F. (2022, July). The IAEA solution: Knowledge sharing to prevent dangerous technology races [Manuscript in preparation].
Synergy Research Group. (2022). Cloud infrastructure service map. https://web.archive.org/web/20230211114844/https://techblog.comsoc.org/wp-content/uploads/2022/10/unnamed.jpg
Tech Trends. (2023, June 21). Meta reveals new details for in-house chip designed to make AI apps and data centers more efficient. Sourceability. https://sourceability.com/post/meta-reveals-new-details-for-in-house-chip-designed-to-make-ai-apps-and-data-centers-more-efficient
TechCrunch. (2022, February 7). Nvidia calls off its efforts to acquire Arm. TechCrunch. https://techcrunch.com/2022/02/07/nvidia-calls-off-its-efforts-to-acquire-arm/
TechTarget. (2023, June 13). Oracle plans to provide generative AI services with Cohere. TechTarget. https://www.techtarget.com/searchEnterpriseAI/news/366541569/Oracle-plans-to-provide-generative-AI-services-with-Cohere
Thadani, A., & Allen, G. C. (2023). Mapping the semiconductor supply chain: The critical role of the Indo-Pacific region. Center for Strategic and International Studies.
The Information. (2014, January 26). Google beat Facebook for DeepMind, creates ethics board. The Information. https://www.theinformation.com/articles/google-beat-facebook-for-deepmind-creates-ethics-board
The White House. (2023, October 30). Executive order on the safe, secure, and trustworthy development and use of artificial intelligence. https://www.whitehouse.gov/briefing-room/presidential-actions/2023/10/30/executive-order-on-the-safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence/
Tirole, J. (1988). The theory of industrial organization. MIT Press.
Tirole, J. (2023). Competition and the industrial challenge for the digital age. Annual Review of Economics, 15, 573–605.
Tirole, J. (2023). Socially responsible agencies. Competition Law & Policy Debate, 7(4), 171–177.
Trammell, P., & Korinek, A. (2023). Economic growth under transformative AI. National Bureau of Economic Research Working Paper. https://doi.org/10.3386/w31815
TrendForce. (2024, May 16). Wave goodbye to Samsung next year? Google rumored to collaborate with TSMC in its Pixel 10 lineup, featuring 3nm SoC. https://www.trendforce.com/news/2024/05/16/news-wave-goodbye-to-samsung-next-year-google-rumored-to-collaborate-with-tsmc-in-its-pixel-10-lineup-featuring-3nm-soc/
TSMC. (n.d.). 3nm technology. Retrieved June 1, 2024, from https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_3nm
U.S. Department of Justice. (2005, October 13). Samsung agrees to plead guilty and to pay $300 million criminal fine for role in price fixing conspiracy. https://web.archive.org/web/20051112051337/http://www.justice.gov/atr/public/press_releases/2005/212002.htm
U.S. Department of Justice. (2015, April 27). Applied Materials Inc. and Tokyo Electron Ltd. abandon merger plans after Justice Department rejected their proposed remedy [Press release]. https://www.justice.gov/opa/pr/applied-materials-inc-and-tokyo-electron-ltd-abandon-merger-plans-after-justice-department
Vipra, J., & Korinek, A. (2023). Market concentration implications of foundation models. arXiv preprint arXiv:2311.01550.
Waters, R., & Shubber, K. (2023, February 3). Google invests $300mn in artificial intelligence start-up Anthropic. Financial Times. https://www.ft.com/content/583ead66-467c-4bd5-84d0-ed5df7b5bf9c
WikiChip Fuse. (2018, July 22). VLSI 2018: GlobalFoundries 12nm leading-performance, 12LP. WikiChip Fuse. https://fuse.wikichip.org/news/1497/vlsi-2018-globalfoundries-12nm-leading-performance-12lp/
Zhao, W. X., Zhou, K., Li, J., Tang, T., Wang, X., Hou, Y., ... & Wen, J. R. (2023). A survey of large language models. arXiv preprint arXiv:2303.18223.
Zimmerman, A. (2022, August 26). R&D funding breakdown: CHIPS and Science Act [PDF]. American Association for the Advancement of Science.
Author:
Tomás Aguirre
This paper is