Welcome to the new Energy Central — same great community, now with a smoother experience. To login, use your Energy Central email and reset your password.

Jim Baird
Jim Baird
Top Contributor

AI, Fusion and the Limits of Growth

Understanding the future trajectory of energy growth is crucial, as governments and organizations worldwide base their assumptions on the expectation that this growth trend will continue for centuries.

Professor Thomas Murphy, a physicist at the University of California, San Diego, suggests that the assumption regarding energy consumption may be reasonable. Based on data from the U.S. Energy Information Agency, he points out that since 1650, U.S. energy consumption has shown a remarkably steady growth trajectory, with an annual growth rate of 2.9% per the following graphic.

This represents a tenfold increase every century.

Since the X axis of this graphic is logarithmic, the increase appears linear. However, when it is converted to whole numbers, the exponential nature of this growth becomes more apparent.

It now looks more like a chart of the early growth rate of the COVID-19 virus.

Since the current annual consumption is 20 terawatts (TW), at the current rate, we would consume 200 TW in 2125, 2000 TW in 2225, and, in 400 years, the annual consumption would exceed the total solar power incident on the Earth, at which point the Earth’s surface would reach the boiling point of water.

Skeptics claim this is unrealistic because fossil fuels, which have enabled most of this growth over the last 200 years, are expected to be depleted or abandoned long before this limit can be reached.

But beyond emissions, waste heat is increasingly becoming a factor in warming. Although it is currently much smaller than the warming caused by greenhouse gas emissions, it nevertheless traps heat in the atmosphere and the ocean.

Mark Flanner of the National Center for Atmospheric Research has calculated that waste heat could directly warm industrialized parts of the world by between 0.4 °C and 0.9 °C by 2100. And Harvard astrophysicist E. J. Chaisson, in a paper Long-Term Global Heating From Energy Usage says, “within 1000 years our technological society could find itself up against a fundamental limit to growth: an unavoidable global heating of roughly 3ºC dictated solely by the second law of thermodynamics, a bio geophysical effect often ignored when estimating future planetary warming scenarios.”

“The total energy budget of society on Earth will likely continue growing for three reasons. First, world population is projected to increase until at least the late 21st century, when it might level off at approximately 9 billion people. Second, developing countries will mature economically, perhaps for the next several centuries, until equity is achieved among the world community of nations. And third, the per capita energy rate will probably continue rising for as long the human species culturally evolves, including conditioning our living spaces, relocating cities swamped by rising seas, and sequestering increased greenhouse gases—which implies that even if the first two reasons for growth end, the third will continue increasing society’s total energy budget, however slowly,” Chaisson noted.

One current hope is that artificial intelligence can be a panacea for the world’s most pressing environmental challenges. However, according to an increasing number of studies, there is also a negative side to the rapid expansion of AI and its related infrastructure. Short-term (2023–2030),     20%–30% compound annual growth rate is commonly cited in industry reports like the IEA, McKinsey, and semiconductor forecasts. And in the long term (2030–2050), AI could be responsible for 5%–15% of global electricity demand by 2040 if current trends continue, which would be twice to ten times higher than the historical energy growth rate.

Although fossil fuels are expected to be depleted or abandoned long before the thermodynamic limit is reached, most of their replacements are fraught.

Wind and solar are both intermittent, and as Bryan Leyland has pointed out on these pages, the notion that some affordable way to store surplus electricity at scale is a pipe dream. As is his, and others claims, that “nuclear power could provide all the low- cost, low emissions electricity the world needs for hundreds or thousands of years.”

Chaisson points out, “nuclear methods (either fission or fusion) that emit no greenhouse gases, these energy sources would still spawn additional heat above what the Sun’s rays create naturally at the Earth’s surface.”

“If world population plateaus at 9 billion inhabitants by 2100, developed (Organisation for Economic Cooperation and Development, or OECD) countries increase non-renewable energy use at 1% annually, and developing (non-OECD) countries do so at roughly 5% annually until east-west energy equity is achieved in the mid-22nd century, after which they too will continue generating more energy at 1% annually, then a 3ºC rise will occur in about 320 years (or 10ºC in ~450 years), even if carbon dioxide emissions end,” he warns.

COVID-19 is a global warming metaphor. It wasn’t just a pandemic but also a systems failure, and AI, fission, and fusion are following the same arc. They are falsely promising salvation, overhyped, and underregulated, and they have thermodynamic impacts, including urban heat islands, coral bleaching, oceanic stratification, and disrupted weather patterns from heat plumes.

"Bending the curve" was the call in the early days of the COVID-19 pandemic and refers to measures taken to slow down the spread of the virus to prevent a rapid surge in cases that could overwhelm healthcare systems per the following graphic.

Bending the energy curve is what is needed today to prevent runaway warming. The heat accumulating in the oceans needs to be converted into 13 tranches of work spread out over 3000 years to prevent the Earth’s energy system from being overwhelmed, even as this work replaces fossil fuels without spawning additional heat above what the Sun’s rays create naturally at the Earth’s surface.

4