Ai architecture: A resilient Hawaiʻi through data infrastructure & clean energy sovereignty

Abstract

The Intelligence Age marks the evolution of the Information Age, ushering in a new paradigm defined by the rapid advancement of Artificial Intelligence—from narrow and generative models to the anticipated rise of Artificial General Intelligence (AGI). These technologies are not theoretical abstractions; they are rapidly integrating into the core operations of governments, industries, and societies worldwide. Unbeknownst to many, AI’s existence is entirely dependent on physical systems and cannot function without the robust computational, operational, connective, and energy infrastructure that supports it. This Doctor of Architecture thesis argues that AI data infrastructure—specifically, data centers, supercomputing systems, and clean energy production—must be treated as critical infrastructure priorities. These systems must be inherently interconnected, particularly in Hawai‘i, where a statewide mandate requires a transition to 100% renewable energy by 2045 for Oʻahu and by 2035 for the neighbor island counties. Energy and data centers are so inseparable that data centers themselves are planned, scaled, and valued in units of power (megawatts). However, in Hawai‘i—where land and electricity costs rank among the highest in the nation—this interdependency has traditionally posed a barrier to AI infrastructure and, by extension, progress itself. This thesis flips the traditional model. Whereas conventional data centers rely on the grid as their primary energy source—placing strain on aging infrastructure—this proposal prioritizes on-site clean and renewable generation, relegating the grid to a secondary role. This reversal reflects a broader pursuit of energy sovereignty, especially in a state still heavily reliant on imported fossil fuels. In parallel, Hawai‘i also outsources most of its computational capacity, resulting in diminished data sovereignty, higher latency, and vulnerability to external control. Establishing locally rooted, sovereign computing infrastructure offers a pathway to reclaim control over both data and energy futures. Together, these two imperatives—energy sovereignty and data sovereignty—form the backbone of a holistic infrastructure strategy. By designing energy and compute systems in tandem, this thesis contends that Hawai‘i can transform constraints into opportunities. This approach gives rise to what this thesis defines as AI Architecture—a new framework for sustainable, secure, and resilient AI data infrastructure. Allowing Hawai‘i the opportunity not only to overcome these structural barriers, but to set a precedent for others to follow. This thesis proposes the architectural development of a 5–15 MW data center with high-performance computing (HPC) capabilities, guided by strategic site selection across three geographically and energetically distinct locations in Hawai‘i: Project 1: Ocean Energy-Powered Data Center — utilizing Seawater Air Conditioning (SWAC), Ocean Thermal Energy Conversion (OTEC), and Tidal Energy at Hawai‘i Pacific University’s Oceanic Institute (Waimānalo, O‘ahu Island)Project 2: Nuclear Energy-Powered Data Center — incorporating a Small Modular Reactor (SMR) at Joint Base Pearl Harbor-Hickam (Waipi‘o Peninsula, O‘ahu Island) Project 3: Geothermal Energy-Powered Data Center — leveraging geothermal energy at the Puna Geothermal Venture site (Puna, Hawai‘i Island) Utilizing a Multi-Criteria Decision Analysis (MCDA) framework, the project evaluates each proposal across seven dimensions: Economic Viability, Regulatory Feasibility, Energy Reliability, Technical Maturity, Sustainability, Scalability, and Adaptability—identifying the true challenges, opportunities, and feasibility of each. Structured in three parts, the thesis proceeds as follows: Part I: Framing the Intelligence Age: The Critical Need for AI InfrastructurePart II: Strategic Siting & Development — A Multi-Criteria Analysis of Energy and Site Suitability for AI Infrastructure Part III: Designing AI Infrastructure: Conceptual Proposals This thesis operates as both research and design, advancing a transdisciplinary, resilience-focused framework that reimagines how sovereign AI, energy, and data infrastructure can be architecturally developed. Its aim is not simply for Hawai‘i to participate—but to lead—in the Intelligence Age. Consequently, if Hawai‘i is to not only remain relevant but lead in the Intelligence Age, it must begin to build and control its own AI infrastructure. Hawai‘i currently imports the majority of its compute capacity, resulting in diminished data sovereignty, increased latency, and heightened vulnerability due to external control. To counter, this thesis advocates for a place-based approach to AI infrastructure—one that integrates clean energy, resilient development, and technological sovereignty—a reality that gives rise to what this thesis defines as, AI Architecture.