Market forces have proven to increase productivity and decrease prices over time. If these forces are allowed to act indefinitely, with proper guardrails, there should be no end to how low prices can go. There is no limit on the productivity of humanity.

The legos of modern industry

All goods and services can be derived from this (vastly oversimplified) list of inputs:

1. Natural resources (land, water, ore, oil, wood, topsoil, wind, etc)
2. Energy
3. Transportation
4. Human labor (both physical and intellectual)
5. Machines (automating human labor)
6. Computers (automating human intellectual labor)

Together these inputs create everything humans could ever want or need: food, clean water, clothing, houses, cars, furniture, etc. 

The nesting doll nature of prices

Take a pair of shoes for example. At the time of assembly, the shoe company buys leather, shoe laces, material for the soles, etc and then uses a combination of machines and human labor to assemble them into a shoe. The shoe company then sells the shoe and pockets the difference between the sale price and the cost to produce it.

Before that someone had to produce the leather and all the other inputs. Each with their own inputs until you get back to raw materials in or on the Earth somewhere. At each step the producer had input costs and pocketed the difference between the sale and the costs.

The price of every good or service can be thought of as one of those nesting dolls (matryoshkas). As you open the doll, you reveal the next input and its price which comprises the larger doll. You can do so recursively until you get to the smallest doll inside. 

One of the fundamental tenets of capitalism (which has on balance turned out to be true) is that if someone can produce something for lower cost and has competition, the price will go down over time.

The cumulative effect of technological advances leading to increased industrial capacity leading to lower prices is the best way to explain the explosion in prosperity over the last few hundred years.

Step changes in prices

If capitalism is so good at reducing prices, then why isn’t everything free? Prices have been collapsing for most goods over the course of human history, but there’s always been a bottleneck holding back further price collapse.

Throughout humanity’s history of expanding industrial capacity the large step changes in progress were enabled by fixing bottlenecks in the core inputs of industry. 

To use an example from Part 1, let’s talk about clothing. The mechanical loom was invented in the late 1700s and was an innovation in the field of machines. It wasn’t a fundamental advance in energy production or using novel materials; it was using existing known technologies to create an innovation.

The loom amplified each human’s work and required less skill to operate than previous methods of creating fabric. At that time, the bottleneck for clothing was not raw materials or energy; it was human labor. This invention unbottlenecked the textile industry.

Until it hit the next bottleneck.

Enabling what wasn’t possible before

Smelting aluminum is incredibly energy intensive: 13kWhs of electricity per kg of aluminum. That’s about ⅓ of what an American household consumes daily. Just to smelt one kilogram of aluminum. Aluminum was incredibly expensive and rare until electricity became cheap enough.

Aluminum is so useful that (in 2021) humanity consumed ~3% of all electricity generated in the world just to smelt aluminum. (total smelting, electricity per ton, total global energy production). Without aluminum, we wouldn’t have many products, including modern aircraft. In fact, it became so cheap that aluminum’s primary use is in packaging (soda cans).

Dropping the cost of electricity made new industries possible—aluminum is just one example. In turn, cheap aluminum made more new industries possible (modern air travel, food packaging, etc.).

As the prices of goods and services drop, more goods and services become economically viable.

Requirements for broad price collapse

Some areas of innovation are more fundamental than aluminum smelting and have a more profound, sweeping effect on the world, like labor. Labor is an input at almost every stage in the production of nearly everything.

If we can drastically increase production with the same amount of human labor, we can drastically decrease the price of almost everything. The problem is finding more general-purpose productivity improvements. The loom automated and amplified a specific kind of labor, so there was a drastic drop in price in that specific industry, but it didn’t increase the productivity of iron ore refining, for example.

Machines that are more general-purpose and can automate or amplify human labor broadly drop the price of everything to some degree. The internal combustion engine is an example. That innovation enabled the adoption of motorized vehicles, which dropped the price of transportation so much that it affected the price of everything. Today, the cost of transportation is a significant component of the cost of just about everything.

If advances in the most fundamental inputs to industry can be applied broadly across the economy, we could see an accelerating loop where each advance reduces prices for the rest of the economy in an endless cycle.

(Potentially) the last bottlenecks to price collapse

As a species, our technology base is sophisticated enough to enable general-purpose automation that wasn’t possible when the pinnacle of technology was stone tools. 

If we can advance in the following areas (only possible now and in the near future), there is no theoretical limit to what humanity can produce.

Energy

Energy is an input at every point in industrial processes. Drastically dropping the price of energy will directly reduce the cost of everything from food to furniture.

On top of that, many technologies are known to humanity that are only economically non-viable because the cost of electricity is too high.

Water abundance and desalination

Water is a critical resource for humans to survive. In some regions, natural freshwater is incredibly scarce (and therefore expensive). Desalination is a well-established technology; it just takes a lot of electricity. In some regions (Saudi Arabia, Israel, etc.), desalinating water is the primary water source because it’s cheaper than drawing water from other sources.

In other places, it’s not economically viable because natural water sources are cheaper to draw from than desalination—at least until the rivers and aquifers run dry.

If electricity becomes cheap enough, water wouldn’t be scarce anywhere on Earth.

Carbon-neutral fossil fuels

Fossil fuels get a bad rep because pulling them out of the ground and burning them adds carbon to the atmosphere, which causes climate change. However, it’s also true that humanity wouldn’t be where it is without fossil fuels. If they suddenly disappeared, the amount of humans that could survive on Earth would instantly drop dramatically.

In addition, there are some applications where fossil fuels are the only energy source that makes something feasible: aviation, rockets, and heavy industry.

Fossil fuels are just carbon, hydrogen, and sometimes oxygen. That’s all available in our atmosphere, such as water, the oxygen we breathe, and CO2 (from, at least, previous fossil fuel burning). What if we created fossil fuels from what’s in our atmosphere?

Terraform Industries is working on just that. When electricity is cheap enough, their fossil fuel generation plants will produce natural gas at prices lower than drawing from the underground. Because solar power costs are dropping rapidly, they’re well on their way to this.

If energy costs drop, we’ll be able to enjoy the benefits of fossil fuels without the downsides.

Trends in electricity cost

The cost of electricity from solar energy, even if you include batteries, has become cost competitive with fossil fuels (source). That’s just how it is today. The cost of solar and batteries are on a long rapid decline. They’re roughly dropping 10-20% with every doubling in production, which happens every 2-3 years at current pace. (solar, batteries). 

If trends continue, could we see electricity prices drop by half? 90%? 99%? It’s not clear what the next bottleneck will be or when we’ll hit it.

Recursive cost improvement

Since a large component of the cost of producing solar panels and batteries is electricity, the cost of producing more solar panels and batteries drops as electricity becomes cheaper. Therefore we can deploy even more solar panels and batteries. 

Round and round it goes.

That’s just solar

With just the projected advances in solar and batteries, we could see incredible price declines in energy. That doesn’t even mention what could be possible with the current capabilities of nuclear fission and the future possibilities if we crack fusion.

Nuclear fission (which takes no technological advances to take advantage of) is cheaper than even solar is today (source). “…the main factor leading to high plant construction costs is not the design of the reactors or various safety features that they employ, but the uniquely strict QA requirements that apply (only) for the fabrication of safety-related nuclear plant components (i.e., ‘nuclear-grade’ components).” (source), which could be on a similar cost curve to solar energy if we could produce enough nuclear plants to take advantage of Wright’s law.

Nuclear fusion was always “20 years away” until the 2020s. Now, there are companies with actual construction plans and commercial deals with buyers like Microsoft to install utility-scale fusion power plants by 2028 (source). If we crack fusion, we could see energy drop even lower than what we could get from solar panels.

Energy abundance is a positive feedback loop

Energy is a fundamental bottleneck for industrial expansion, but with today’s technology and the plans we already have underway, we’re on track for a dramatic energy cost reduction. 

This could unlock abundance for resources like water, heating/cooling, fuel, manufacturing, and computing. Abundance in these resources will further drop costs to produce even more energy and we may see energy prices collapse to near zero.

Digital Intelligence

When the mechanical loom, electricity, or the internal combustion engine were invented (discovered?), they required human-level intelligence—sometimes many human-level intelligences—to bring them about.

It’s not just invention and innovation; human-level intelligence drives the entirety of the knowledge economy (building software, project management, marketing, etc.).

Until recently, the possibility of human-level intelligence from a digital system seemed to be something for the distant future. “Since 2019, futurists have collapsed the time to AGI (Artificial General Intelligence) from 80 years to 8 years.” (source)

The effect of human-level, digital intelligence 

Even if we don’t progress from human-level digital intelligences to superhuman intelligences, the price of many goods and services will drastically reduce.

If the same software engineer can produce 10X more in the same amount of time, the cost to produce software products will plummet. 

If the same marketer can supervise a few digital intelligences that operate different aspects of marketing, the costs of the products being marketed can be reduced. The list goes on. 

Even with human-level intelligence, we can augment scientists and researchers of all kinds. If the same scientist can work on more projects, we will make more discoveries. If engineers and researchers can try more iterations on their inventions, we’ll get more innovations—and faster.

All of this feeds back into the technology and industrial expansion cycle.

Autonomous transportation

A human-level intelligence can drive a car or a truck. Roughly 43% of the cost of shipping via truck is the driver (source). Roughly 38% of your average ridesharing trip fare goes to the driver (after the platform fee and the cost to operate the car).

The upshot is that human-level digital intelligences can be used to drop the cost of transportation by nearly half.

The bottleneck to widespread deployment of digital intelligence

Currently, the main drivers of the cost to create and operate digital intelligences are chips and electricity.

The costs (in chips) to train the same level of intelligence is dropping rapidly (~70% per year) due to learned efficiencies from researchers. Just using existing resources more efficiently (source).

As of 2024, we’re also seeing massive expansion in chip manufacturing to meet the insatiable demand. So that may not be a bottleneck for long.

Soon the limitation will be how much electricity there is to power those chips. For example: “If every Google search became an [Large Language Model] interaction, the electricity needed to power this could amount to the same as a country [the size of] Ireland at 29.3 TWh per year.” (source) 

It’s not far-fetched to imagine everyone dropping Google search for a ChatGPT-like alternative with a Large Language Model (LLM) in the coming years, not to mention every other task we want to offload to LLMs and agents powered by LLMs.

So advances in the cost and abundance of electricity will directly affect our ability to deploy digital intelligences widely enough to see the widespread productivity benefits.

General-purpose machine labor

Humanity extensively uses machines in manufacturing and agriculture and has for many decades. They are used when the quantity of output for a given good is so large that the cost of specialized machines is less than it will produce.

Designing and building a machine to solve a specific problem is still fairly expensive, so their usage is reserved for applications with high throughput.

If there were general-purpose machines that were roughly as capable as a human but far cheaper to hire (operate), then the number of applications where automation is profitable will increase dramatically. 

This isn’t groundbreaking reasoning, but it’s important to state WHY we haven’t seen machines automate everything yet. It’s because machines haven't achieved human-level dexterity and adaptability yet.

More generally capable robots

In Amazon warehouses, swarms of robots with different capabilities automate tasks that used to require humans (moving racks, moving and packing boxes, etc.). These are more generally capable than previous generations of industrial robots. 

Recently, though, we’ve seen companies like Figure, Tesla, and Boston Dynamics showcase robots that seem to be approaching human-level capability. Whether robots that are human-level capable are two years or ten years away, it doesn’t seem like they’re 20 years away anymore.

When these become able to do 20%, 50%, or even 90% of the physical labor humans can do, humanity’s capacity to build in the physical world can grow far beyond what humanity can now.

The bottleneck to widespread robotics usage

These robots will contain very complex hardware and be powered by electricity. 

At least for a time, humans will be more efficient laborers on a unit-by-unit basis. So, to automate all the human labor that can be automated, it will likely take billions of units. The only hardware we’ve ever built billions of units for is smartphones, and those units are simple and small compared to humanoid robots.

Add in the new demand for metals, batteries, plastics, chips, etc, we’ll need to build a large manufacturing capability to fully utilize the potential automation of general-purpose robots.

That’s not even to mention the amount of electricity needed to operate all this new machinery and manufacturing.

Spot, Boston Dynamics’ dog-like robot, runs at ~0.4kWs for about 90 minutes at a time (source). There are some rough numbers here, but if a humanoid robot consumes twice that (being roughly twice the size) and operates 12 hours a day, you’d only need about three humanoid robots to consume more electricity per day than your typical American residential home.

This amount of electricity consumption could quickly add up. Improvements in energy costs and capacity will greatly accelerate our ability to take advantage of the automation of human labor through robotics.

Space-based industry

This set of breakthroughs isn’t necessary in the short and medium term for humanity to achieve high levels of abundance. Still, at some point, land will become the bottleneck for humanity's growth.

All this industrial expansion has and will take a toll on Earth’s natural ecosystems. There may be enough raw resources on Earth to feed the industrial machine, but it will come at a cost that we may not want to pay in the long term.

Those resources, and the space to do heavy industry without hurting natural ecosystems is out there. We just need to get off of Earth.

Raw materials

The solar system is rich in raw materials. Even if you only constrain yourself to asteroids, many are in orbits close enough to Earth to be more easily reachable than the Moon or Mars. In these asteroids are all the raw materials humanity would need for our industrial needs.

It may not make economic sense to mine asteroids and ship the material back to Earth’s surface for use in our industry. But it may make a lot of sense to mine and use them in space-based industries to relocate industry off-Earth. This could reduce the ecological footprint of humanity’s activities and increase the ceiling on humanity’s industrial output.

Living space

There simply isn’t enough land on Earth for 5-10 billion humans to have even an American standard of living without carpeting the earth in suburban sprawl. At some point, we’ll need to build up and/or make more land.

O'Neill cylinders, a conceptual space habitat developed in the 1970s, could greatly expand humanity’s available living space off Earth. They are gigantic metal cylinders in space (usually at least 10 miles in diameter) that rotate to create artificial gravity on the inner surface. 

These habitats could have solar panels and other power generation to sustain its inhabitants’ needs, while housing, schools, shops, and all the systems needed to support comfortable human living could be built inside without needing a planet.

As long as we had a space-based industry, and the political will, we could build these habitats with existing technology. No exotic materials, or advances in physics needed.

In sufficient quantities, these habitats could house millions or billions of humans in high-levels of comfort, providing whole new frontiers for humans to explore and live.

Further energy abundance

Very little of the Sun’s energy that hits Earth is utilized by humanity (far less than 1%) and Earth only receives a miniscule percent of the energy output by the Sun at any given time (< 0.00003%). 

With a space-based industry, we could manufacture immense amounts of solar panels (mounted on O’Neill cylinders or not) that could orbit the Sun to make energy even more abundant than we could do with what’s just hitting Earth.

This concept is called a “Dyson Swarm” and if humanity ever has a trillion members, this is likely how we’d arrange ourselves to sustain our numbers without leaving the Solar System. The amount of energy we could harness if we built to utilize even 1% of the Sun’s output would be unimaginable for modern humans.

The raw materials are out there, a space-based industry is all we’d need to get started.

Advances in space travel and space-based industry

SpaceX’s Starship plans to bring the cost of launching payloads into orbit by several orders of magnitude. Projects that would previously be ridiculously unprofitable could suddenly find themselves within reach.

SpaceX plans to use this price drop to seed a self-sustaining city on Mars. Along the way, they may be the launch partner of choice for companies that wish to start building space-based industries to take advantage of the opportunities discussed earlier.

Then there are companies like Varda that are building space-based manufacturing capabilities for products uniquely able to take advantage of conditions in space (low gravity specifically) that wouldn’t be possible on Earth. Their efforts may seed the space-based industry that incentivizes more economic activity off-Earth.

As folks work live and work in space, there will be more demand for everything humans currently enjoy on Earth, but in space. Schools, gyms, tourism, casinos, whatever. As more folks live in space or work in these places, we’ll see a positive feedback loop that leads to more humans in space and, therefore, more space-based industries to support them.

We’re not far away from this sci-fi future being a reality.

The techno-abundance loop

You can see from each of the above sections that each advance will not only improve themselves recursively, but also accelerate the improvements of other industries. 

There’s no reason this can’t go on until the price of all human necessities approaches zero. 

I call all these forces working together to reduce the cost of goods and services the “techno-abundance loop.” It’s the pattern of technology unlocking new industries which generate capital and abundance which further feed the development of new technology and so on.

This can be our future

This future of abundance isn’t inevitable, but it is much closer and higher probability than most people currently imagine. We must be intentional about building toward this future to ensure every person has equal access to the opportunities of such an abundant future.

In Part 3, we’ll explore the clear roadblocks that could slow down or stop humanity’s advance towards abundance.

Part 3: Roadblocks on the path to abundance

Preview:

1. Concentration of power
2. Inefficient allocation of human labor
3. Sustainability
4. Skipping the hard part via Communism
5. NIMBYs
6. Lack of imagination