The central lesson is that civilization-changing industries are not created by a single invention. They emerge when a whole system of technologies, infrastructure, institutions, and business models becomes economically viable at roughly the same time↳ Institutions and Social Infrastructure.
The automobile was not simply an engine attached to wheels. The computer was not simply a machine that performed logic. Each became transformational because it turned into a platform on which thousands of other products, businesses, and behaviors could be built.
First, a qualification: novelty is not the whole story
The distinction between ships and automobiles is useful, but it can be overstated.
Self-moving vehicles and artificial reasoning appeared in myths and mechanical experiments↳ What Myths? long before they became practical. The conceptual leap was not the main barrier. The barrier was making the idea reliable, affordable, manufacturable, and useful.
Nor are old categories necessarily unimportant. Modern container shipping transformed the world economy even though ships were ancient. A technology can remake society by radically lowering the cost of an old activity.
So there are at least two paths to a massive industry:
- Create a new capability, such as programmable computation.
- Reduce the cost of an existing capability by an order of magnitude, as containerization did for freight.
The best opportunities often do both.
What had to be in place for automobiles?
1. A collection of enabling technologies
No single breakthrough produced the automobile. It depended on progress in:
- precision machine tools
- metallurgy
- petroleum refining
- internal-combustion engines
- electrical ignition
- rubber tires
- interchangeable parts
- brakes, transmissions, and steering
- eventually, assembly-line production
Steam vehicles existed earlier, and electric cars competed seriously with gasoline cars around 1900. Gasoline won initially not because it was inherently destined to win, but because it developed an effective combination of energy density, range, refueling speed, cost, and industrial support.
2. A useful initial market
Early cars were expensive, unreliable luxuries. That was not a defect in the development process; it was how the industry learned.↳ Modern Parallels
Wealthy enthusiasts were willing to tolerate:
- frequent breakdowns
- difficult controls
- poor roads
- scarce fuel
- high prices
They financed experimentation before the product was ready for ordinary households. This pattern appears repeatedly: a small market with unusually high willingness to pay carries an immature technology down the cost curve.↳ Military Spending
3. Mass production
Ransom Olds pioneered early high-volume automotive production. Henry Ford’s deeper achievement was integrating standardization, specialized machinery, moving assembly, high factory throughput, and a product designed for manufacture.
The Model T was introduced in 1908. As production rose, its price fell sharply. Ford understood an important feedback loop:
Lower costs create more buyers; more buyers justify greater scale; greater scale lowers costs again.
That loop turns an invention into an industry.↳ Did Ford Forget?
4. Complementary infrastructure
Cars became valuable as society built:
- paved roads
- filling stations
- repair shops
- dealerships
- parking
- traffic laws
- driver licensing
- insurance
- consumer credit
- oil production and distribution
Some of this infrastructure followed car adoption, but it also accelerated further adoption. Once enough of it existed, the automobile became far more practical and competing transport systems became relatively less attractive.
This created path dependence. The world did not merely buy cars; it rebuilt cities and commerce around them.
5. Second-order businesses
The largest impact came from more than automobile manufacturing. Cars enabled or expanded:
- suburbs
- supermarkets
- motels
- roadside restaurants
- trucking and logistics
- tourism
- auto insurance
- consumer finance
- oil and chemicals
- road construction
- drive-through retail
A transformative technology creates valuable businesses whose founders do not think of themselves as being in the original industry.
What had to be in place for computers?
Computing followed the same pattern over a longer and more layered development.
1. The idea preceded the practical machine
Important pioneers included:
- Charles Babbage, who designed programmable mechanical computers in the 19th century
- Ada Lovelace, who understood that such machines could manipulate symbols, not merely calculate numbers
- Herman Hollerith, whose punched-card equipment industrialized data processing
- Alan Turing, who formalized the idea of general computation
- John von Neumann and others, who helped establish stored-program computer architecture
- John Mauchly and J. Presper Eckert, who developed ENIAC and later commercial machines
- John Bardeen, Walter Brattain, and William Shockley, whose transistor work made smaller, more reliable electronics possible
- Jack Kilby and Robert Noyce, who independently developed key forms of the integrated circuit
The early insight was surprisingly broad. Lovelace saw that a machine capable of manipulating symbols might operate on music or other formal systems. Turing grasped the generality of computation. Vannevar Bush anticipated aspects of personal information systems.
But even the best pioneers usually saw capabilities more clearly than markets.
2. Large institutions paid for primitive systems
The first electronic computers were too expensive for consumers. Their initial customers had unusually valuable problems:
- military ballistics
- cryptography
- scientific calculation
- census processing
- payroll and accounting
- banking and insurance records
Governments and large corporations absorbed the high costs. Like wealthy automobile enthusiasts, these customers funded improvement while the technology remained difficult to use.
3. Components improved along a steep learning curve
Vacuum tubes gave way to transistors, then integrated circuits and microprocessors. Cost, size, energy consumption, and failure rates declined while performance increased.
That combination matters. A technology becomes especially potent when it improves across several dimensions at once:
- cheaper
- smaller
- faster
- more reliable
- easier to use
Semiconductor progress repeatedly moved computers into new markets:
- governments and laboratories
- large corporations
- smaller businesses
- households
- pockets and everyday objects
- cloud-scale infrastructure and AI systems
Each drop in cost produced not merely more sales, but new uses.
4. Standard platforms allowed specialization
A general-purpose computer separates the machine from its applications. That means one company can build hardware, another an operating system, and thousands more can write software.
Important commercial pioneers—including IBM, Intel, Microsoft, Apple, and many others—built layers of a growing stack. The computer industry became enormous partly because no company needed to invent the entire system in order to participate.
This is one of the clearest signs of a foundational industry:
It shifts from selling products to supporting ecosystems.
5. Networks multiplied the computer’s value
A standalone computer automated individual tasks. Networked computers reorganized communication, media, commerce, and social life.
The internet was therefore not just another computer application. It was a complementary system that multiplied the value of every connected computer. Smartphones later combined computation, networking, cameras, sensors, location, identity, and payments in one mass-market device.
How far did the pioneers see?
The record is mixed.
Some pioneers understood the broad destination:
- Lovelace saw symbolic manipulation beyond arithmetic.
- Turing understood that one general machine could imitate many specialized machines.
- Ford believed cars could become mass-market products rather than permanent luxuries.
- Early personal-computing pioneers saw computers as tools for individuals, not just institutions.
But few anticipated the full social system that followed.
Ford did not need to foresee suburban shopping malls, interstate trucking, or drive-through restaurants. Semiconductor pioneers did not need to predict social media, cloud computing, or generative AI.
This suggests a useful distinction:
- Technological foresight identifies what a machine may eventually do.
- Market foresight identifies who will pay for it next.
- system foresight identifies the infrastructure and institutions that will form around it.
Founders generally need the first two. Investors and strategists benefit greatly from the third. No one needs a detailed prediction of the mature world.
In fact, detailed long-range predictions can be counterproductive. The better question is often not “What will the world look like in 30 years?” but:
If this capability becomes ten times cheaper and more reliable, what becomes possible next?
A general model of industry formation
A new industry has a strong chance of becoming enormous when seven conditions converge.
1. A previously scarce capability becomes abundant
Automobiles made powered personal movement widely available. Computers made calculation and information processing abundant.
Look for sharp changes in the supply of a basic capability:
- intelligence
- energy
- labor
- biological design
- transportation
- manufacturing
- trust
- communication
2. Performance follows a compounding curve
The strongest technologies improve through cumulative production and research. Costs fall, adoption rises, and adoption finances further improvement.
A one-time technical breakthrough may create a product. A durable improvement curve can create an era.
3. There is an expensive but urgent initial use
Early products rarely win by being broadly adequate. They win by solving one problem for which customers will tolerate high cost and inconvenience.
Good beachhead markets have:
- acute pain
- measurable value
- concentrated buyers
- short feedback cycles
- high willingness to pay
4. Complementary technologies are becoming ready
The automobile needed fuel distribution and roads. Personal computing needed semiconductors, displays, storage, and software. Smartphones needed wireless networks, batteries, touchscreens, and compact sensors.
A famous idea may fail repeatedly until its complements mature. Timing often means noticing that several independent curves are crossing.
5. The product becomes a platform
A platform permits people outside the original company to discover uses the inventor never imagined.
Examples include:
- roads and standardized vehicles
- personal computers and software
- smartphones and apps
- cloud computing and APIs
Third-party experimentation expands the search space. That is why platforms often grow faster than vertically planned systems.
6. Institutions adapt
Large industries require changes in law, finance, education, standards, insurance, and public infrastructure.
Institutional friction slows adoption, but it also creates opportunity. Companies that solve compliance, financing, installation, training, or integration can become as important as the headline inventors.
7. The technology changes behavior, not just efficiency
A small innovation helps people do the same thing slightly better. A foundational innovation changes:
- where people live
- how firms are organized
- what skills are valuable
- which goods are economically possible
- how people spend time
That is the difference between a feature and an industrial revolution.
The leading candidates in 2026
No one can know which industries will rival automobiles and computers. But several have the right structural characteristics.
1. Artificial intelligence
AI is the clearest candidate because it makes a fundamental input—certain forms of cognitive labor—cheaper and more abundant.
Its strongest analogy is probably not the automobile. It is the computer itself. AI is a new computing layer that changes how software is created and used.
Potential effects include:
- software that operates through goals rather than explicit commands
- widespread automation of administrative and analytical work
- personalized education and tutoring
- faster scientific research
- lower-cost media and design
- semi-autonomous organizations
- new human-computer interfaces
- much greater demand for computation and electricity
The uncertainty is not whether AI will matter. The uncertainty is where durable profits will settle. They may accrue to model developers, chipmakers, energy suppliers, data owners, application companies, or firms that redesign existing workflows.
A common mistake is to bolt AI onto an old product. The larger opportunities may come from rebuilding a process around the assumption that competent machine reasoning is cheap and continuously available.
2. Robotics and autonomous machines
AI affects the information world first because software can be copied cheaply and deployed instantly. Robotics extends it into the physical world.
Promising markets include:
- warehouses
- factories
- agriculture
- mining
- construction
- defense
- delivery
- elder care
- household work
Robotics has moved more slowly because physical systems face safety requirements, variable environments, maintenance costs, and expensive hardware. But better perception and general models may reduce the amount of task-specific engineering required.
If robots become trainable rather than painstakingly programmed, the industry’s economics could change sharply.
3. Programmable biology
Biology may be moving from observation toward engineering. The relevant advances include:
- cheap sequencing
- gene editing
- automated laboratories
- computational protein design
- cell therapies
- synthetic biology
- AI-assisted drug discovery
The long-run possibility is not merely better medicine. It is the use of cells and biological systems as manufacturing platforms.
The constraints are substantial: regulation, clinical timelines, biological complexity, and safety. This means transformation may be slower than in software, but individual breakthroughs can be far more valuable.
4. Electrification, storage, and new energy systems
Cheap, abundant energy has historically expanded the frontier of economic activity. Important developments include:
- solar power
- batteries
- electric vehicles
- grid software
- advanced geothermal
- next-generation nuclear power
- perhaps eventually commercial fusion
The opportunity is larger than producing electricity. A changing energy system requires transmission, storage, permitting, financing, control software, power electronics, and new industrial processes.
AI itself may make energy more strategically important by increasing electricity demand. A constraint in one transformational industry often creates the next major opportunity.
5. Space infrastructure
Reusable rockets have lowered launch costs and enabled large satellite networks. Potential markets include communications, Earth observation, defense, navigation, and specialized manufacturing.
Space is real and growing, but claims about asteroid mining or mass settlement remain highly speculative. Near-term↳ Asteroid Mining Profitability opportunities are more likely to arise from useful services delivered to customers on Earth.
How to recognize a genuine growth story
Do not begin by asking which technology sounds most futuristic. Ask whether its underlying economics are changing.
Useful questions include:
Is a critical cost falling predictably?
Track cost per unit of useful output:
- cost per inference
- cost per robot-hour
- cost per genome
- cost per kilowatt-hour stored
- cost per kilogram launched
Raw performance benchmarks matter less than delivered economic value.
Who is using the bad, expensive version today?
This reveals the beachhead market. If customers tolerate an awkward product because the alternative is worse, there is real demand.
Which complement has just become available?
A technology may have existed for decades but become investable because of a new battery, chip, dataset, regulation, manufacturing method, or distribution channel.
Does adoption generate more improvement?
Strong feedback loops include:
- more users producing more data
- more production lowering unit costs
- more developers producing more applications
- more infrastructure making adoption easier
- more adoption attracting capital and talent
Are users changing workflows?
A flood of demos is weaker evidence than companies reorganizing operations around the technology. Durable adoption appears in budgets, headcount plans, training, and process design.
Is a supply chain or ecosystem forming?
Watch for specialized vendors, standards, training programs, insurance products, financing arrangements, and regulatory frameworks. These “boring” developments often indicate that an industry is becoming real.
What happens at one-tenth the price?
This is one of the best exercises. Do not extrapolate only from the current customer base. Ask which entirely new customers and uses appear after a large cost decline.
How to capitalize without predicting the entire future
There are several defensible positions in an emerging industry.
Build the bottleneck
The most valuable layer is often the constraint that everyone else encounters:
- chips for AI
- charging for electric vehicles
- launch capacity for satellites
- manufacturing tools for biotechnology
- integration and safety systems for robotics
But bottlenecks migrate. Today’s scarce component may become tomorrow’s commodity.
Build complements
A technology’s growth creates demand for things it cannot provide itself:
- implementation
- security
- data infrastructure
- training
- maintenance
- financing
- compliance
- workflow software
These businesses can be less glamorous and more durable.
Serve an unattractive early market
Start where the technology is already economically justified, even if the market looks narrow. Early markets supply learning that later entrants cannot instantly reproduce.
Own distribution or workflow
Technical advantages often erode. A company embedded in customer operations, with trusted distribution and proprietary workflow knowledge, can retain value after the core technology becomes cheap.
Avoid confusing industry growth with company profits
Railroads, automobiles, airlines, semiconductors, and the internet all transformed society. That did not make every participant a good investment.
A sector can grow spectacularly while:
- competition destroys margins
- capital requirements consume returns
- standards shift
- incumbents commoditize suppliers
- value moves to another layer
Correctly predicting the future of technology is not enough. You must also predict the structure of value capture.
A practical way to train your judgment
Maintain an “industrial change” notebook. For each candidate technology, update six items:
- Capability: What has recently become possible?
- Cost curve: What is getting cheaper, and how fast?
- Beachhead: Who pays for it today?
- Complements: What else must exist for mass adoption?
- Bottleneck: What currently limits growth?
- Second-order effects: What new businesses become possible if adoption succeeds?
Then make explicit predictions with dates and numbers. For example:
- deployment cost will fall below a given threshold
- a specific customer segment will adopt it
- regulation will permit or restrict a particular use
- the bottleneck will shift from hardware to energy or distribution
Review those predictions annually. The goal is not to prove that you were right. It is to discover which variables you consistently misunderstand.
The deepest lesson
Automobiles and computers became enormous because they did not merely satisfy existing demand. They changed the set of things people could reasonably want.
That makes the earliest stages difficult to evaluate with conventional market sizing. Before mass automobile adoption, asking consumers how much they spent on cars would reveal almost nothing. Before personal computing, there was no established household market for software.
The strongest signal is therefore not a grand vision by itself. It is a grand vision attached to a working economic wedge:
- a primitive product that someone urgently wants,
- a cost curve that can improve,
- complements moving into place,
- and a path from one application to a platform.
Look where a basic capability is becoming dramatically cheaper. Then look one layer beyond the celebrated invention—toward the bottlenecks, complements, infrastructure, and new behaviors it will create. That is usually where the future becomes visible before it becomes obvious.