[The USA] uses 500 trillion kcal of energy to produce our food currently…consumes 25 quadrillion kcal per year, working out to about 77 million kcal per person or 210 thousand kcal per day
- Herman Ponzer (emphasis my own)
Last year, my brother, Ben, and I read the book “Burn: The Misunderstood Science of Metabolism” by Herman Pontzer. That means we spend 50 times more energy growing, processing, and distributing food than we ultimately eat. How does the US put in so many inputs to produce so little food? Globally, we produce about 9 quadrillion Kilocalories (Kcal) of food with about 3.4 quadrillion Kcal of inputs. That’s a pretty great return with 0.37 Kcal generating 1 Kcal of consumable food.
So how is it we Americans need eight times the amount of energy to produce food?
Our food system has traded energy efficiency for labor productivity that has had a massive impact on what and how we eat today. Let’s walk through the accounting and see what the deposits and withdrawals are, where they come from, and how they’ve changed over time.
Inputs: how much does it cost to feed us?
Unfortunately, this is going to be a graph heavy post. Fortunately, I’ll break it down. Let’s first look at the energy we put into growing, distributing and preparing food in the US over three timestamps: 1900, 1970, and 2022. To understand how energy shapes our food system, let’s look at the three biggest drivers of agricultural energy use: fertilizer, labor, and fuel. Each tells a story about what we’ve chosen to prioritize and what we’ve left behind.
Fertilizer
The first aspect that popped out to me right away, and is likely pretty obvious is how much fertilizer we need to produce food. In 1900, most fertilizer was produced from manure, bone meal and guano (or bird poo). Each of these are natural sources of nitrogen to help grow plants, which Jean-Baptiste Boussingault confirmed in 1836 based on the previous work of Justus von Liebig. This isn’t to say that before these dates, no one inherently knew this, since Native American tribes like the Haudenosaunee grew the Three Sisters of beans, squash, and corn. Legumes like beans, fix nitrogen and deposit into the soil to aid all three to grow. They would also allow fish to decay at the base of plants for more nitrogen deposits.
What this did enable was a mad dash for the three aforementioned products to get as much nitrogen into the ground as possible. One of the most interesting stories from this time period is Guano Island and the wars between Spain and their former colonies for this rich deposit of nitrogen. All the more interesting when the birds that would drop feces on this island stopped coming due to El Niño cycles.
Anyway, nitrogen was a premium to keep societies growing and stave off famines. In 1909, that all changed when Fritz Haber demonstrated the condensation of nitrogen from the atmosphere, of which 78% is nitrogen. Six years later, Carl Bosch industrialized the process to make seemingly endless nitrogen in the form of ammonia, which could then be made into synthetic fertilizers. A miracle! However, this process alone now accounts for upwards of 2% of global energy consumption and 3% of greenhouse emissions due to the amount of heat and pressure required to make it happen.
Regardless, the ability to grow vast quantities of food with cheap fertilizer was too good of an option for us to ignore and the globe went all in. You can see both by 1970, it made up almost a third of all the energy needed to produce food, and that same rate continues today. Ultimately, it’s one of the only ways we could grow a global population of over 8 billion, but still needs a way to become greener.
Now look at where we stand compared to the rest of the world: we use over 10% more energy from fertilizers alone. Is that because our plants require that much more fertilizer? Is it because the US produces the most fertilizer in the world and the market is awash with cheap nitrogen? Given the quantity of genetically modified crops in the United States and the fact that they are designed to use nitrogen more efficiently makes this all the more puzzling. We used to rely more on biology and ecology, but have leaned into chemistry and combustion to continue increasing our food supply.
Despite our technological edge, genetically modified crops, and abundant farmland, the U.S. still uses significantly more energy to grow food than many other countries. That’s not because we have to, but rather because we’ve optimized our system for productivity, not efficiency.
Human Labor
The next portion of energy accounting that pops out is the amount of people within agriculture. Almost a quarter of the energy in agriculture came from the blood and sweat of agricultural workers in 1900, when 40% of the country’s population farmed. Today, less than 2% farm and account for barely half a percent of the energy needed to produce food. Where did all the farmers go and why did they get out of farming? Earl “Rusty” Butz told them to get big or get out.
You’ll notice that the rest of the world still heavily relies on people to carry out much of the labor sowing, tending to, and harvesting crops. Almost a fifth of the global population works in agriculture, a number the US hasn’t seen since the Dust Bowl. We’ll dig into this more in a bit.
Fuel
Another third of the energy needed in 1900 was for fuel to transport products to market and to use in the brand-new invention: the tractor. Tractors became bigger and continually more efficient, but still made up more than a fifth of the energy used in the 1970s, and 17% today. While transport costs have stayed more or less the same in these time periods, it’s worth revisiting the fertilizer discussion earlier. The Haber-Bosch process runs on fossil fuels, so when you look at how much fuel we used in the 1970s and today, the real fuel accounting is 62% and 61%, respectively, of all the energy growing and transporting food. Oh, and the cooking in those time periods? Unless you have an electric range that’s powered by solar panels, you can tack on that, bringing our reliance on fossil fuels to almost three-quarters of getting food from farm to table. Compare the amount of fuel the entire world uses to the US today at 54%. A full third less on transport, fertilizers, and cooking.
you are what you eat
I’m not advocating that we go back to a mainly agrarian society, yet it does beg the question of how we expect to continue filling the stomachs of our populace given the high energy costs that go into its production. There are less farmers today than the population of New York City, and their farms have ballooned in size, requiring more mechanization. More machinery means more fuel and more dialed in commodity seeds that are consistent and vetted.
With the aid of these seeds and machines, we now produce more than three times more than we did in 1948. There is no question this is a feat of science and engineering in agriculture, the likes that haven’t been seen since the Agricultural Revolution itself. But what did we optimize for? All of these industrial inputs push for calorie-dense, low-labor foods that make their way into processed, shelf-stable foods. It makes eating a diverse range of foods an opt in rather than opt out. And given the energy usage, waste becomes a feature, not a bug.
Now, seeing as birth rates globally, and locally in the US, are less than half what they were sixty years ago, we see more drivers for technology in agriculture. More technology means more energy and our track record shows that means more fossil fuels. What kind of system becomes sustainable with less people and more costly energy? I’m advocating for more technology here - but scaleable and renewable energy. More technology in how to grow more diverse foods that suit the ever drier, ever saltier fields. We need to put our money where our mouth is, and right now it’s covered in oil.
This dynamic of trading energy for labor and scale for nuance doesn’t stop at agriculture. In fact, we’re doubling down on it in a new frontier: artificial intelligence.
farm to model
Just as we continue pouring fossil energy into machines to replace human muscle on the farm, we’re now doing the same with silicon and servers to replace, or augment, human cognition. The rise of artificial intelligence is the latest frontier in this long-running trade: energy for labor, scale for nuance, optimization for diversity.
The Green Revolution scaled food production by using more synthetic fertilizers, machines, and fossil fuels. It created incredible abundance but came with huge costs: monocultures, soil degradation, water pollution, and enormous greenhouse emissions. Now, as large language models and generative AI systems explode in complexity and power, we're watching energy use skyrocket in the name of productivity, creativity, and efficiency.
Training a state-of-the-art AI model today can consume as much energy as thousands of households do in a year. And deploying those models where every query, every generated paragraph, every optimized supply chain adds to that footprint. Much like food, AI promises more: more yield, more content, more insight. But we haven’t yet grappled with what it costs, or whether we’re investing that energy in the right places.
The question isn't just “How do we feed more people?” or “How do we think faster?” It’s “What are we using all this power for?” Are we creating resilient, nourishing systems? Or just chasing scale at any cost?
Both our agricultural system and our emerging digital systems show the same core truth: energy enables abstraction. It gives us leverage over nature, over time, over human limitations. But that leverage can lead to fragility if it outpaces our sense of stewardship.
energy keeps the score
Energy is the invisible currency of civilization. How we spend it reveals what we value.
For centuries, we spent it to feed bodies to till soil, grind grain, transport goods, and preserve perishables. That energy transformed agriculture from subsistence into surplus, from local into global. But it also distanced us from the land, hollowed out regional food cultures, and tied nourishment to fossil-fueled infrastructure.
Now, we’re spending that same energy to feed machines that mimic minds. We train models that simulate creativity, decision-making, and conversation, the very processes we once thought were uniquely human. The knowledge economy is becoming an energy economy too, where thinking at scale carries a carbon cost.
In both domains, farming and computing, we’ve sought to abstract away labor, to maximize yield, to grow more with less. But in doing so, we've often forgotten to ask: more of what? For whom? And at what cost?
Energy doesn't care what it's used for. It can sustain life or accelerate its destruction. It can grow food or fuel distractions. The challenge isn’t just technological, it’s moral. What do we want all this power for?
Because in the end, every calorie and every kilowatt is a choice. And behind every choice is a vision of the future. Are we building systems that nourish, or just systems that consume?
super interesting, thank you!