# AI Too Cheap to Meter - (Part 1)

> AI is turning intelligence into a utility. How abundance arrives, where scarcity moves when thinking gets cheap, and what happens to work and firms.

Source: https://www.rohandey.com/writings/ai-too-cheap-to-meter-part-1  
Published: 2026-09-01  
Updated: 2026-10-02  
Category: AI

In 1954, **Lewis Strauss**, the chairman of the US Atomic Energy Commission predicted that nuclear power would make electricity "too cheap to meter." That never happened with energy, but it is now happening with something just as foundational: intelligence.

For all of recorded history, skilled human thinking has been the bottleneck behind every product, service and institution. A diagnosis needed a doctor, a contract needed a lawyer, and a software product needed a team of engineers. AI breaks that bottleneck. The same work now costs a fraction of what it did, runs around the clock and scales instantly.

Economics is the study of allocating scarce resources. So what happens when one of the scarcest resources stops being scarce? This is the first of two parts. It covers how AI ushers in abundance, where scarcity moves once intelligence is cheap, and what that means for work and firms.

## How abundance arrives

Abundance won't arrive as one big event. It comes through five mechanisms that compound on each other.

1. **The Collapse of Cognitive Cost**:
   - Tasks that took an expert hours (drafting, coding, analysis, translation) now take minutes and cost cents
   - Inference prices keep falling as models get more efficient and hardware improves
   - Once something costs close to nothing at the margin, it gets used everywhere, much as electricity did a century ago
2. **Agents as a New Labor Supply**:
   - AI agents add what is effectively a new, elastic workforce on top of the human one
   - A firm's output is no longer capped by headcount; it is capped by compute, capital and coordination
   - That is a supply shock comparable to adding millions of workers to the economy, without the demographic lag
3. **Accelerated Science and R&D**:
   - The deepest abundance comes from AI speeding up discovery: drug design, materials, energy storage and chip design
   - Faster R&D lowers the cost of physical goods, not only digital ones
   - This is the mechanism that turns cheap intelligence into cheap energy, cheap medicine and cheap materials
4. **Deflation in Human-Bound Services**:
   - Healthcare, education and legal services have grown more expensive for decades because they could not scale
   - AI tutors, diagnostic assistants and legal agents let these services reach billions at a near-zero marginal cost
   - For the first time, quality expertise can be both cheap and personalised
5. **Compounding Through Software**:
   - AI writes software, software deploys AI, and each cycle makes the next one cheaper
   - Digital goods were already non-rival; AI removes the cost of producing them in the first place
   - The result is a flood of tools, products and services aimed at markets too small to serve before

## Where scarcity moves

Abundance doesn't abolish scarcity. It relocates it. Economics will spend the next decade working out where value moves.

1. **Physical Inputs Become the Bottleneck**:
   - **Energy**: Data centres turn electricity into intelligence, so grid capacity becomes a strategic asset
   - **Compute**: Advanced chips depend on a handful of fabs and a fragile supply chain
   - **Land and Materials**: Cooling, water and critical minerals become new points of constraint
2. **The Jevons Paradox, Applied to Intelligence**:
   - When steam engines used coal more efficiently, total coal demand rose rather than fell
   - Cheaper intelligence will be consumed in far greater quantities, so demand for inputs keeps climbing
   - Abundant outputs can coexist with tight, expensive inputs
3. **Human Scarcities Gain Value**:
   - **Trust and Verification**: When anything can be generated, provenance and authenticity become premium
   - **Attention**: Infinite content competes for a fixed 24 hours
   - **Presence**: Care, craftsmanship and in-person experience carry a growing relative premium
4. **Baumol in Reverse**:
   - Baumol showed that labor-heavy services get relatively pricier as other sectors grow more productive
   - AI flips this for cognitive services, which become cheap
   - It intensifies the effect for physical, human-only work like plumbing, nursing and elder care

## What happens to work

One of the biggest relocations is in work. Labor economics assumed that human skill was the main way to create value, and that assumption is weakening.

1. **Tasks Get Unbundled Before Jobs Disappear**:
   - A job is a bundle of tasks; AI automates the routine cognitive ones first
   - The analyst becomes the reviewer of ten AI analysts, and the developer becomes an architect of agents
   - The quiet risk is the loss of entry-level work, the rung where people traditionally built skills
2. **Wages Polarise**:
   - People who direct, verify and take responsibility for AI output capture large productivity gains
   - Routine knowledge work faces the same wage pressure that factory automation brought to manufacturing
   - Physical and relational roles hold or gain relative value
3. **Comparative Advantage Survives, With a Catch**:
   - Humans will specialise where their relative edge is greatest, even if AI is better at everything
   - But the market wage for that edge can fall below a living standard
   - Markets determine where people work; policy determines whether that work pays enough
4. **The Theory of the Firm Gets Rewritten**:
   - Coase argued that firms exist because coordinating through markets is costly
   - Agents slash coordination costs, so firms can be far smaller for the same output
   - Expect the rise of the one-person, ten-agent company and an economy of solo builders and micro-firms

## Conclusion

Abundance isn't a single breakthrough. It's a set of compounding forces: cognition gets cheap, agents expand the workforce, science speeds up, services deflate and software builds on itself. Together they push the cost of intelligence toward zero.

But cheap intelligence doesn't end scarcity; it moves it. Energy, compute, trust, attention and human presence become the new constraints, and whoever controls them gains pricing power. Work isn't disappearing so much as being reorganised: tasks get automated before jobs do, wages split between people who direct AI and those it replaces, and a single founder can run what once needed a full company.

In [Part 2](https://rohandey.com/writings/ai-too-cheap-to-meter-part-2), we zoom out to the whole economy: what abundance does to prices and money, who owns the gains, and what could stop it from arriving at all.