The race takes off in the next big arenas of competition

This McKinsey Global Institute virtual event discusses a new industrial environment characterized by the emergence of "arenas" – industries exhibiting hyper-growth, hyper-investment, and hyper-profitability. A report two years prior identified that companies representing 10% of global revenue accounted for over half of value creation in these arenas, which were termed "wizards" compared to "muggle" industries. These arenas were characterized by large end-use markets, significant technological shifts, and a competitive dynamic of "escalation" or "races," where investment leads to capability expansion rather than just increased capacity. The initial report identified 18 such arenas, projected to generate up to $50 trillion in revenue and account for one-third of future economic growth. Two years later, a follow-up analysis confirmed that these arenas accounted for half of all global value creation, with a 29% compound annual growth rate in market capitalization. Notably, if these arenas were excluded, the rest of the corporate data set showed almost no growth, indicating that these 18 arenas were responsible for nearly all net growth. The identified arenas are at various stages of growth, from early-stage shared autonomous vehicles to more mainstream AI software and services, and established but still rapidly growing e-commerce. Over three years, these arenas added $18 trillion in market capitalization. A significant portion of this growth is in AI foundations (semiconductors, cloud services, and AI software), but other platforms like digitization (digital advertising, e-commerce), electrification (EVs, batteries), hard tech (physical technology applications), and new bio frontiers (e.g., GLP-1 drugs) are also seeing substantial advancement. Kevin Russell further elaborates on the AI foundation aspect, highlighting three key points: extreme escalation, the importance of ecosystems, and the rise of "omni-scalers." Investment in AI foundations (semiconductors, cloud, AI software/services) is projected to reach $750 billion by 2025, double the Apollo program's cost in similar dollars over a shorter period. This massive investment is shifting overall arena investment from 23% to 28% of the economic landscape and influencing foreign direct investment globally. The AI ecosystem involves a complex web of interconnected players, from semiconductor manufacturers like TSMC and NVIDIA, to hyperscalers providing compute and cloud services, and AI labs developing software and services. These interconnections mean each player has a stake in the entire ecosystem's success. The report raises questions about potential constraints in this system, particularly concerning the expected continued growth in AI application demand. A novel finding is the emergence of "omni-scalers"—nine companies (six US, two Chinese, one Korean) that invest at least $20 billion annually in CapEx and R&D and operate in three or more of these arenas. Examples include Alphabet, Meta, Amazon, Tesla, and Samsung. These companies have expanded significantly from their core businesses in 2010 to encompass multiple arenas today, particularly in AI foundations and digital services. Some, like Alphabet, Amazon, Tesla, and Samsung, also play in hard tech and electrification. While omni-scalers are not taking over all arenas, they demonstrate large operating cash flows that enable significant investments (31% of revenue goes into CapEx and R&D, three times more than non-arena companies). They are not growing significantly faster than other arena companies yet, but their investment patterns are transformative. Regionally, the US and China dominate in terms of arena market capitalization, with the US having the largest share. In revenue terms, the US leads in digital and AI foundation arenas, but China shows strong presence in electrification and hard tech segments. Europe and Japan/Korea also have niche capabilities in specific arenas like robotics and bio-frontiers. For companies, the report suggests that "we're all in the arena in some way," meaning even traditional non-arena industries will be disrupted or find opportunities from these trends. For example, a traditional retailer could see transformations in supply chains, operations (robotics, cyber, cloud), and demand shifts due to AI software and services. During the panel discussion, Naveen Sastry addresses whether omni-scalers are merely modern conglomerates. He argues they are different due to their nimbleness, intertwined business units with synergies (talent, IT sharing, self-consumption), and founder-led or founder-influenced control, which allows for long-term investment horizons. He notes that while NVIDIA and frontier labs like OpenAI are not omni-scalers, they exemplify the "recipe" of huge product-market fit and free cash flow that allows for future investments. Gayatri Shani discusses the physical constraints underlying the AI boom, particularly the massive compute infrastructure required. She emphasizes that the limiting factors are not just capital, but physical resources like power, land, cooling, and water. Building gigawatt-scale data centers (equivalent to powering a million homes) requires rapid mobilization, permitting, and grid access, which are constrained by geography, policy, and community acceptance. She argues that while we are not at the ceiling, these factors must be addressed quickly for continued growth. The geopolitical dimension of AI is also highlighted. While US firms hold a significant market cap share, China is rapidly gaining ground. Data sovereignty is fragmenting the market, forcing hyperscalers to adapt with local zones and government cloud offerings. This makes IT data architecture a dynamic environment where agility is crucial. Brendan Gaffey adds that the AI ecosystem has struggled to communicate its benefits to the public, leading to distrust around data centers concerning water, energy prices, and job displacement. He stresses the need for the technology sector to articulate how AI improves lives. Brendan also addresses the impact of AI on "normal" companies. He disagrees that AI is not yet showing up in P&Ls, citing adjacent companies innovating business models (e.g., Keysight's AI platform for data center optimization, Lumen's fiber expansion). He acknowledges that the average big company might take longer to see productivity gains, drawing parallels to the Solo paradox with computers and electricity's slow initial impact. However, he points to AI-native companies scaling at $10 million per FTE and radical productivity changes in coding and call centers (50-60% automation with voice agents) as evidence of immediate impact. He notes that transforming enterprises requires reimagining workflows and investing in data, technology, talent, and adoption. Naveen Sastry further discusses the robustness of omni-scalers and the AI investment wave. He highlights that CapEx productivity and usage are high, with demand being met rapidly. Real revenues, not just intercompany transfers, are driving this growth, often in a supply-constrained market. Most financing comes from operating cash flows, not debt or VC, indicating financial stability. He dismisses fragility concerns, especially as AI offerings become more applicable to average and large companies. The panel also discusses value migration. While NVIDIA currently captures significant profit, historical trends suggest value moves up the chain from hardware to software, platforms, and applications. Future disruption and value capture are likely for those who own proprietary data, can leverage it to drive value, and integrate into enterprise workflows, owning the workflow end-to-end. Regarding global competition, the panel acknowledges the US and China's dominance but suggests opportunities for "middle power" countries like India or European nations. These countries can focus on specific arenas like robotics, which bridge AI foundations with the physical world, and consider their role in the value chain (manufacturing, logistics, healthcare, infrastructure management). Policies that attract and retain talent, reduce regulatory burdens, and facilitate new company formation are crucial. Building sufficient compute infrastructure is also a fundamental starting point for any national policy. In closing, Brendan Gaffey emphasizes that scale is not destiny; focus and learning curves are paramount, applicable to both startups and omni-scalers. He encourages traditional companies to learn from these extremes to carve out their own "niche omni-scaler" strategies. Gayatri Shani reiterates that this is an intellectually exciting time, with value chains shifting and winners determined by strategic focus, data, reimagined workflows, and talent. Naveen Sastry concludes by stating that omni-scalers have provided a "time machine," offering a glimpse into the future potential for every company to reinvent itself for greater productivity and a better future.