AI Supply Chains, Nuclear Power, and Data Dominate Global Competition

Global competition is intensifying as frontier AI labs, nuclear-powered data centres, proprietary data control, and talent poaching reshape digital markets, while firms balance open-source flexibility with governance and navigate energy and regulatory challenges for resilient, future-ready operations.

Key Points

• Market concentration limits consumer choice and entry and shapes innovation, the dynamic frontier model market and vertical integration strengthen control, AI Ph.D. graduates entering industry rose from 21% in 2004 to 73% in 2022, low-income countries have gaps in compute, talent, and data, AI data centres require high energy and water leading Google to use small nuclear reactors, proprietary data increases product value, firms face differing US-EU regulations and captured only $400 billion of a $6 trillion social value increase, and firms balance risk and resilience through sustainable infrastructure, open-source safeguards, data accumulation, talent competition, and managing resource gaps.
  ​​
  
• Recommender systems primarily use human behaviour to keep users engaged but are moving towards aligning with user values through middleware and conversational agents while rising AI opacity increases demand for explainability, many AI systems refine themselves silently, exploit behavioural biases, change unpredictably, and create echo chambers, and organizations use explainable AI, generative agents, diverse data signals, and AI audits to manage risks and match evolving user values.
  
  
• AI algorithms reshape information processing and decision-making, machines influence social networks, QR codes and automation shift tasks to customers, Big Tech operates transnationally, firms handle diverse regulations like the EU’s GDPR, manage opaque AI models, meet real-time data demands that can stress workers, face supply chain gaps in compute, talent, and data, and use AI to automate decentralized energy control, optimize shipping and 3D printing to cut energy use by up to 27% by 2050, improve predictions with human forecasts, and maintain accessibility by blending AI with human support.
  
  
• ​Organizations make AI pro-worker by promoting augmentation, accelerating innovation without automating creativity, prioritizing upskilling, using partnerships and sandboxes for complementary AI development, while many firms deploy “so-so” AI without productivity gains, face skills and access gaps, underinvest in AI safety with only 2% of research, low-income regions lack resources, and policies, partnerships, small language models, and continuous training help bridge these gaps and build resilience.

Download the Full Report

​Integrate key insights into a holistic strategy for your company.

te250616_-_ai_supply_chains_nuclear_power_and_data_dominate_global_competition.pdf
File Size:	776 kb
File Type:	pdf

Download File

---

Market consolidation, energy-intensive infrastructure, and proprietary data strategies are redefining the foundations of digital competitiveness and global AI leadership

Market concentration and talent flows drive AI competition

Market concentration raises policy concerns by limiting consumer choice and restricting entry by smaller and newer firms. It shapes innovation away from socially desirable outcomes.
 
The frontier foundation model market is dynamic, fluid, and marked by intense competition among dozens of AI labs. Open-source models contribute both flexibility and risk.
 
Concentration occurs through vertical integration, where a few firms consolidate upstream activities such as data and chips and downstream distribution to consumers. Low-income and many middle-income countries face significant gaps in compute, talent, and data.
 
The proportion of AI Ph.D. graduates entering industry increased from 21 percent in 2004 to 73 percent in 2022.
 
High energy demands and regulatory diversity challenge digital firms

AI data centres have voracious appetites for energy and water. Google is turning to small nuclear reactors to power them.
 
Digital platforms and social media firms have accumulated massive amounts of proprietary data, amplifying product value as more people use it.
 
The United States has emphasized innovation and light regulation. The European Union has prioritized individual protections and potential social harms, requiring multinationals to navigate differing compliance regimes.
 
New digital firms appropriated only $400 billion of the $6 trillion increase in social value created by digital firms.
 
Sustainable energy investment and data control support resilience and growth

Google is turning to small nuclear reactors to power AI data centres. This demonstrates investment in sustainable infrastructure to address high energy demands.
 
The market for frontier foundation models is dynamic, fluid, and characterized by intense competition among dozens of AI labs. Open-source models contribute flexibility but require safeguards.
 
Digital platforms and social media firms have accumulated massive amounts of proprietary data, amplifying product value as more people use it.
 
The proportion of AI Ph.D. graduates entering industry rose from 21 percent in 2004 to 73 percent in 2022, showing competition for talent.
 
Low-income and many middle-income countries face yawning gaps in compute, talent, and data inputs.
 

Shifting from opaque, behaviour-driven AI to transparent, value-aligned systems reflects growing demands for explainability, trust, and ethical personalization in consumer engagement

Recommender systems evolve from behaviour-based engagement to values alignment

Recommender systems derive recommendations primarily from human behaviour. They often aim simply to keep users engaged on the platform.
 
There is discussion about moving beyond relying on online behaviour to match who people are and what they value and have reason to value.
 
New models like middleware and large language model recommender agents aim to scaffold human agency in interactions with online content.
 
The increasing opacity of AI systems makes it difficult to know why complex AI systems choose what they choose and whether their choices reflect the choices people would make.
 
Self-adapting AI and behavioural bias traps hinder consumer trust

Many AI systems are not simply trained once but are instead refined with data and experience. They evolve dynamically and silently.
 
Current systems exploit system-1 thinking, which refers to behavioural biases that digital platforms exploit for engagement.
 
Even well-characterized AI can change over time, perhaps suddenly and silently, rendering understanding and interventions obsolete.
 
Engagement-optimizing algorithms risk showing content aligned with pre-existing beliefs and creating echo chambers.
 
Explainable AI and audits build trust and align recommendations with user values

Investing in explainable AI or explainable machine learning supports human oversight. It helps users understand why specific recommendations are made.
 
Generative AI agents can learn about what matters to the user by engaging conversationally. They refine content to match evolving user values.
 
Moving beyond purely behavioral data toward signals about user preferences, motivations, and context can bridge the gap between what customers do and what they actually want.
 
Organizations should adopt AI audits to characterize how an AI system functions, its risks, biases, and other relevant factors.
 

​AI algorithms, regulatory complexity, and automated systems are transforming operational models, introducing both efficiency gains and structural strain across global markets

Algorithms and automation reshape social interactions and customer service

AI-powered algorithms reflect a fundamental change in how information is processed, how decisions are made, and how people live their lives.
 
By inviting machines into social networks, the choices they make begin to impact people through these networks in similar ways that people impact each other.
 
The rise of QR code ordering, self-checkouts, and automated check-ins shows that substantial service tasks are passed on to customers without reducing prices.
 
Big Tech operates transnationally, moving data and services across borders.
 
Complex regulation, opaque models and supply gaps strain operations

Firms face diverse regulatory landscapes like the EU’s GDPR and the 'Brussels Effect,' creating compliance complexity across markets.
 
AI models often operate beyond the effective control of the people who design and deploy them, creating governance challenges.
 
Technologies like automated guided vehicles, sensors, and pick-to-light systems increase real-time data demands.
 
These tighten just-in-time dynamics but also cause worker stress and repeated errors if processes break down.
 
A simplified AI supply chain hinges on computing power, talent, and data, with gaps in each input for many countries.
 
AI-driven energy grids, logistics, and human-AI forecasts boost efficiency

AI can automate the control of distributed, decentralized, and micro-grid energy systems to improve grid resilience and energy access.
 
AI-driven optimization of shipping routes and three-dimensional printing can reduce surplus inventory and cut energy use by up to 27% by 2050.
 
Mixing forecasts from AI models with those from human experts yields better predictions than those of human experts alone.
 
Firms should blend AI tools with human support to counteract customer frustration with full automation and ensure accessibility for diverse consumer groups.
 

Organizations are evolving through AI-augmented structures, targeted policy support, and workforce development to close innovation gaps and build resilient, inclusive ecosystems

Organizations embed AI augmentation and partnerships to amplify human capability

Organizations are rethinking structures to make AI pro-worker through institutions and policies that empower workers to use AI to augment what they do while limiting AI curbs on worker agency.
 
Firms are embedding practices to harness AI’s potential to accelerate science and technological innovation by augmenting, not automating, creative processes.
 
Organizations are prioritizing workforce upskilling so AI’s flexibility and adaptability can be leveraged to personalize education and healthcare.
 
Firms and governments are increasingly adopting public–private partnerships, regulatory sandboxes, and impact-based funding to steer AI R&D toward technologies that complement human skills.
 
Mediocre automation, skill gaps and poor infrastructure weaken innovation potential

Many organizations still deploy AI that merely mimics what people do, automating work but resulting in job losses without delivering broader productivity gains.
 
Access and relevant skills are limiting factors for using technology more broadly, and the persistent digital divide is a major barrier.
 
Only about 2% of AI research focuses on safety, while commercial applications like computer vision dominate.
 
Low-income regions struggle with insufficient funding, weak digital infrastructure, and a shortage of skilled professionals, limiting equitable collaboration and internal innovation culture.
 
Policies, small models and continuous training build resilient AI culture

Policies and incentives should tilt the balance towards augmentation, promoting fiscal policies that encourage augmentation so that AI enhances jobs instead of replacing them.
 
Cross-sector partnerships and inclusive AI ecosystems ensure diverse perspectives, bridge infrastructure gaps, and align global AI safety frameworks with local governance norms.
 
Developing small language models for specific use cases increases data privacy, reduces latency, and ensures continuous operation.
 
Continuous investment in education, digital literacy, and on-the-job training helps workers perform higher-value tasks, reinforcing a resilient and innovative internal culture.