The Infrastructure Behind AI: Why the Grid Alone Cannot Keep Up
Why the Grid Alone Cannot Keep Up The Infrastructure Behind AI: Why the Grid Alone Cannot Keep Up Solar DC Power is planning to develop agrivoltaic-powered rural data centers and…
Why the Grid Alone Cannot Keep Up The Infrastructure Behind AI: Why the Grid Alone Cannot Keep Up Solar DC Power is planning to develop agrivoltaic-powered rural data centers and…
Feeding People and Powering Machines: AI, Farmland, and Food Security Post 4 in our series on AI, energy, and the future of rural America. The last post argued that the…
AI and the Climate AI and the Climate: The Question Is Not Whether, but How Post 3 in our series on AI, energy, and the future of rural America. The…
What AI Can Do for Climate The potential of AI to address climate instability is real and already partially realized. Here is what the evidence shows: Energy systems optimization. AI…
The Hidden Cost of Waiting: How Engineering Experience Saves Years on Data Center Permitting How 42 years of civil engineering experience translates into faster permitting and fewer costly surprises. The…
What a Fish Farm in Spain Taught Me About Energy There is a fish farm in southern Spain that does not feed its fish. The farm is called Veta La…
The Grid Is the Vulnerability: Architecture Eliminates the Risk Most data centers are only as secure as the grid they depend on. A cyberattack on a regional substation, a severe…
Reversing Desertification - and What It Means for Agrivoltaics The relationship between grasslands and the animals that graze them is older than agriculture itself. When megafauna graze grass down, the…
This is the first post in a six-part series drawn from Solar DC Power's Technical Policy Brief on artificial intelligence. The series covers what AI is, how it must be powered, and what it means for climate, food security, jobs, and governance.
Why This Brief Exists
Solar DC Power is planning to develop the physical infrastructure that makes artificial intelligence possible: solar-powered, agrivoltaic data centers on working farmland in the American Southeast. We have a direct stake in how AI develops, how it is powered, and what it does in the world.
This series is written for journalists, policymakers, community leaders, and business partners who want to understand AI not as abstraction, but as the most consequential technology since electricity, with all of the opportunity and responsibility that comparison implies.
What Artificial Intelligence Actually Is
Artificial intelligence is, at its foundation, a set of computational techniques that allow machines to identify patterns in data and use those patterns to make predictions, generate content, or take actions, without being explicitly programmed for each task.
The version of AI now reshaping the global economy is built on a family of architectures called large language models and foundation models, trained on vast datasets using a technique called deep learning. The result is systems capable of reasoning across domains, generating text and images, writing and debugging code, interpreting scientific data, and engaging in nuanced conversation.
This is not the robotic AI of science fiction, nor the narrow rule-based systems of earlier decades. Modern AI is general-purpose in ways no prior technology has been. A single model trained on human knowledge across medicine, law, agriculture, engineering, and literature can assist a doctor, help a farmer, explain a contract, or simulate a molecule, depending on what it is asked.
The Acceleration Curve
What makes the current moment historically distinct is the rate of capability improvement. AI capabilities doubled approximately every two years between 2012 and 2020. Since 2020, the doubling rate has accelerated to roughly every six to twelve months on benchmark measures of reasoning, coding, and scientific problem-solving. This is not a linear progression. It is exponential, and it is compressing timelines in ways that are challenging for policy, education, and industry to absorb.
A few milestones that mark how fast this has moved:
In 2012, AI began outperforming humans on narrow visual tasks. By 2017, the transformer architecture that underlies all modern AI was published. In 2020, GPT-3 demonstrated broad language capability across domains with no task-specific training. In 2022, ChatGPT reached 100 million users in 60 days, the fastest product adoption in history. By 2023 and 2024, AI was passing bar exams, medical licensing boards, and coding benchmarks at expert human levels. Today, in 2025 and 2026, AI agents are beginning to take multi-step autonomous actions: booking, coding, researching, and designing independently.
What Comes Next
AI does not run on ambition alone. It runs on electricity, cooling, and physical infrastructure, and the choices being made right now about how to build that infrastructure will shape what kind of AI future gets built in the American Southeast and beyond.
Post 2 in this series addresses that directly: the infrastructure behind AI, why the grid alone cannot support it, and what Solar DC Power is building instead.
