Skip to content
POLARA

Thesis

AI, energy, robotics, and space

These four fit together more tightly than most people realize. Here's the full picture.

The 5 layers of AI

Jensen Huang (Nvidia's CEO) popularized describing AI as a five-layer cake, where each layer depends on the one below it:

  1. Energy. AI runs on electricity. A single large data center can use as much power as a small city. This is why AI companies are signing deals for nuclear plants, gas turbines, and solar farms.
  2. Chips. The processors that do the math: Nvidia GPUs, Google TPUs, and custom chips (like Etched's). Chips are made almost entirely by TSMC in Taiwan, which is one of the biggest bottlenecks in the world.
  3. Infrastructure. Data centers, cooling, networking, and cloud platforms that turn chips into usable computing power (Microsoft, Amazon, Google, Oracle, CoreWeave, xAI's Colossus).
  4. Models. The AI systems themselves, like GPT, Claude, Gemini, and Grok. They're trained on huge amounts of data using everything below them.
  5. Applications. The products people actually use: coding tools, AI doctors, legal assistants, robots, agents. This is where most of the value eventually reaches everyday life.

Why it matters for investing: money and value move between layers over time. Right now chips and energy capture a lot of the value because they're scarce. Over time, more value usually shifts toward applications. Knowing which layer is the bottleneck tells you where the pricing power is.

Four limiting factors

The same logic applies to civilization as a whole. Progress happens one bottleneck at a time. Elon Musk describes it in terms of limiting factors: find the constraint holding everything back, remove it, then find the next one. At the scale of civilization there are four: energy, intelligence, labor, and space. Remove one, and the others get easier.

Energy

Energy is the foundation. Data centers, robots, and rockets all come down to the same question: how much power can you get, and at what price?

  • Demand is surging. The International Energy Agency expects data center electricity use to more than double by 2030, with AI as the main driver.
  • Power now sets the pace. Microsoft signed a deal to restart a reactor at Three Mile Island, and Google and Amazon are backing small modular reactors. Increasingly, electricity, not chips, decides how fast data centers get built.
  • Supply is getting cheaper and more inventive. Solar and batteries have each fallen roughly 90% in cost since 2010. Companies like Aalo Atomics, building compact nuclear reactors, and Zeno Power, building nuclear power systems for space and remote places, are pushing further.

The binding constraint is the grid. In the US, more new generation is waiting to connect than every power plant running today. Whoever solves generation, storage, and transmission holds pricing power across everything built on top.

Intelligence

AI turns energy and chips into intelligence, the most general tool there is. It already writes code and drafts contracts. More importantly, it is starting to speed up research itself: designing chips, proposing new drugs, and running experiments. That is how it reaches medicine, through faster drug discovery and earlier diagnosis, and how it lets us work on problems beyond our own understanding.

Labor

We need more work done than there are people to do it. Robots are the answer, and for the first time they are becoming general purpose. Three things changed:

  • AI gave robots a brain. The same models behind ChatGPT let a robot see, follow an instruction like “clean up this table,” and plan its own movements.
  • Simulation made training fast. Robots practice millions of times in virtual worlds before touching anything real.
  • Hardware got cheap. Phones and electric cars made motors, batteries, and sensors affordable.

That is why Tesla (Optimus), Figure, and others are building humanoid robots. If AI is the brain, robotics is how it does physical work: building, farming, manufacturing, and care.

Space

Space is the next step for humanity. We think it was always our purpose to explore the stars.

Space gives us room to grow: resources, energy, and places to build beyond Earth. It is finally practical because launch got cheap.

  • The Space Shuttle cost roughly $54,000 per kilogram to reach orbit.
  • SpaceX's reusable Falcon 9 brought that to roughly $3,000 or less.
  • Starship aims for a few hundred dollars per kilogram.

Reusability did it: rockets now land and fly again instead of being thrown away. And space finally pays for itself, with Starlink earning billions from internet service. Companies like Axiom Space, building the first commercial space station, are turning orbit into a place to work and build.

Why it compounds

Every tool helps build the next one. For decades, chips doubled in power about every two years (Moore's Law), and costs fall by a steady percentage each time production doubles (Wright's Law), which is exactly what happened to solar, batteries, and rockets. Now AI is speeding up research itself, and when the tool that improves technology is itself improving, progress accelerates.

Individual technologies still follow S-curves: fast growth, then a slowdown at physical limits. What looks like one long exponential is many curves stacked. The growth continues, but the leaders change, which is what matters most for investing.

How it fits together

The four factors form a loop. Energy powers AI. AI gives robots a brain. Robots make space workable, since people are expensive and fragile there. And space returns the favor: sunlight in orbit is constant, and the cold helps cool computers, which is why data centers in orbit are being seriously explored. Each turn of the loop makes the next one cheaper.

Elon Musk's businesses map onto the four constraints almost exactly: Tesla's batteries and solar on energy, xAI on intelligence, Optimus on labor, SpaceX on space. Backing them early would have been extraordinary. Polara looks for the companies removing the bottlenecks we face as a civilization right now.

The flywheelEnergy powers AI, AI gives robots intelligence, robots make space workable, and space provides energy and room for more AI. Together the loop compounds toward abundance.EnergyAIRoboticsSpaceAbundancethe compounding loop
  1. Abundant energy powers compute.
  2. Intelligence gives machines a brain.
  3. Robots make space workable.
  4. Orbit offers constant sunlight and room to grow.

Where it leads

Energy

Clean power so cheap it stops being a constraint.

Medicine

AI-discovered drugs and longer, healthier lives.

Work

Robots that build, farm, manufacture, and care for people.

Frontier

A permanent human presence on the Moon and Mars.

Discovery

Science that compounds because AI helps do the research.