The Essential Ingredient Binding the “A.I. Five Layer Cake”

Underappreciated Policy Issues and Stakeholder Blind Spots for the “Largest Infrastructure Buildout in Human History” 

NVIDIA CEO Jensen Huang describes the artificial intelligence ecosystem as a “five-layer cake”: energy, chips and computing infrastructure, cloud infrastructure and data centers, models, and applications and software. Each layer enables the next, with energy as the foundation. As Huang argues, AI is on track to become the largest infrastructure buildout in human history, requiring massive capital investment and coordinated execution. But disruptions—economic, reputational, or political—at any layer can cascade across the system.  

In this analogy, policymaking is the binding ingredient. However, each layer is governed by siloed public policy institutions, with authority further siloed between executive, legislative, and quasi-judicial bodies at multiple levels of government. 

BOTTOM LINE UP FRONT: Across all five layers, underappreciated policy issues and stakeholder blind spots are shaping AI’s trajectory. Anticipating those challenges early offers an opportunity to educate key audiences and influence how the AI ecosystem takes shape.  

ENERGY

• Underappreciated Policy Issue - Power Transmission: Recent headlines highlight how AI’s energy demands could increase utility bills for consumers. However, congestion on regional electric grids has quietly become a bottleneck for AI development. Permitting processes for new transmission lines are increasingly limiting other aspects of AI deployment. Even when power generation is adequate, it often cannot be delivered to where it is needed. 

• Stakeholder Blind Spot - State and Local Regulators: Energy transmission is mostly overseen by public utility commissions, regional transmission operators, and local permitting agencies. These lesser-known stakeholders control the pace of energy transmission expansion, and larger private-sector players are only now beginning to realize the importance of engaging them.  

CHIPS AND COMPUTING INFRASTRUCTURE

• Underappreciated Policy Issue – Supply Chain Integrity: Semiconductor manufacturers are competing fiercely to deliver advanced computing power to customers, but fragile global supply chains complicate chip production at scale. Imports of raw materials and unfinished components are highly sensitive to fluctuations in trade and geopolitical events.  

• Stakeholder Blind Spot – Shared Consumers: Regulators overseeing trade, tariffs, and investment policies have received plenty of input from individual companies over the past year about specific products. However, companies throughout the AI supply chain are often connected by their need for the same raw materials. Better coordination among companies with shared dependencies would improve their ability to communicate the critical stakes of AI supply chain security.  

CLOUD INFRASTRUCTURE AND DATA CENTERS

• Underappreciated Policy Issue – Water Use: Data center companies have been facing high-profile battles over electricity costs and local land-use decisions. However, data centers also depend on water for cooling and electricity, which is causing a rising conflict in many communities. This issue is expected to intensify as local jurisdictions consider water rights, resource management, and long-term supplies, especially in drought-prone areas like California and the broader Southwest. 

• Stakeholder Blind Spot – Environmental and Emergency Officials: In many states and localities, control over water use can fall outside the purview of traditional utility commissions. Environmental regulators and emergency management officials have considerable influence over water policy and often enjoy high public trust. Early engagement and validation from these stakeholders can help foster trust for data center developments. 

AI MODELS

• Underappreciated Policy Issue – Antitrust: Policymakers are increasingly vocal about restricting the use of AI models. At the same time, rapid model development is fueling mergers and acquisitions as developers seek competitive advantages. “AI safety” dominates much of the policy discussion. However, it is evolving without widely accepted standards, creating uncertainty about how competition and antitrust rules will ultimately be enforced. 

• Stakeholder Blind Spot – Academics and Economists: AI remains relatively new for U.S. Supreme Court justices, state attorneys general, and other competition regulators. In antitrust cases, decision-makers often turn to academic work to understand specific situations and emerging market dynamics. Engaging with economists and researchers could help developers build the analytical foundation needed to continue fostering innovation and growth.

APPLICATIONS AND SOFTWARE

• Underappreciated Policy Issue – Workers in Regulated Industries: There is widespread public concern about the impact of AI on the workforce. The fact that this concern persists, even as Americans express general distrust of AI and its impact on human jobs, demonstrates that the potential disruption driven by AI technologies has not been fully considered by policyholders or communicated effectively by employers. Indeed, certain sectors of the economy are already highly regulated – like education, financial services, and healthcare – and are siloed into their own purgatories for dealing with technological disruption. 

• Stakeholder Blind Spot – Non-Tech Regulators and Labor Unions: Labor groups are already organizing around technology-driven changes to jobs, compensation, and workplace expectations, as illustrated by the Writers Guild of America. Worker engagement and policy debates will continue to shape AI adoption long before applications reach widespread use. From local school boards influencing curriculum decisions to federal health agencies like the Department of Health and Human Services, developers could find themselves quickly needing to engage industry-specific regulators and labor unions simultaneously.  

INFLUENCE FOLLOWS THE PROCESS: The private sector’s instinct is to find “the most powerful person in the room” and make the substantive case. But in a fragmented decision-making environment, power is distributed—and the decisive leverage often sits with the people who set the agenda, control timelines, define standards, and run permitting and enforcement. No layer of the AI cake is insulated from market, reputational, and political shockwaves that start elsewhere. The advantage goes to the companies that map the process early, engage the right stakeholders before conflict hardens, and turn blind spots into durable partnerships that shape America’s AI ecosystem. 

Matt Moon is a Managing Director at Narrative Strategies, serving as the Head of the West Palm Beach, Florida office. Matt was previously the Deputy Executive Director at the National Republican Senatorial Committee, leading outreach to the business community and policy communications with Senators on crucial floor votes and committee action. Want to continue the conversation? Email me at mmoon@narrativestrategies.com