AI & The Dynamo Doctrine
It’s easy to lose sight of where we are in the AI Supercycle.
In that regard, Jensen Huang’s ability to zoom out, connect the dots, and articulate the broader transformation underway is truly exceptional.
I highly recommend watching the talk below, along with the accompanying article it inspired.
The piece builds on Huang’s perspective while incorporating several frameworks and mental models I’ve been developing to better understand and navigate the AI Supercycle.
Every industrial age is defined by a machine that consumes one substrate and emits another. The substrate that gets emitted — not the machine that emits it — is what reshapes the economy. Three hundred years of industrial history compress into three of these conversions, and the third one is being installed in real time, in concrete and copper, on every continent simultaneously.
The machine is the dynamo. The substrate is intelligence. The cocoon is forming now.
The Substrate Doctrine — Atoms, Electrons, Tokens
A substrate becomes infrastructure when three conditions hold. It must be generable on demand (a machine can produce it). It must be transportable at near-zero marginal cost (a wire, a fiber, a rack). It must be universally consumable (everything plugs in). Three substrates have ever passed all three tests at planetary scale.
Electrons. A coil rotated through a magnetic field emits an invisible force. Within a century, that force cocoons every continent in a grid. Total addressable market: every joule of useful work humanity wants done.
Bits. Electrons routed through switches at near-luminal speed emit packets. Within three decades, those packets cocoon the planet in a network. Total addressable market: every act of human coordination.
Tokens. Electrons fed into massively parallel arithmetic emit numbers that encode meaning — text, code, images, video, protein structures, control signals, decisions. The substrate is intelligence. The economy is reorganizing for the third time. Total addressable market: every act of cognition humanity wants performed, plus the cognition of the machines themselves.
The implication of the lineage is not that AI is “like” electricity. The implication is that AI is on the same curve as electricity — same generational capital requirements, same regulatory eventualities, same commodity endpoint, same hundred-year reorganization of everything downstream. We are not in the chatbot phase. We are in the dynamo phase. The chatbot is one light bulb.
A second, deeper claim sits underneath the lineage: anything with structure is convertible to tokens. Proteins fold predictably. Cells behave predictably. Genes express predictably. Physics is predictable. Materials are predictable. The token is therefore not just a unit of language — it is the universal carrier of any learnable structure in the universe. The substrate is wider than language by orders of magnitude.
The Primitive Inversion — From Retrieval to Generation
For sixty years, computing was a retrieval architecture. You wrote, you stored, you fetched. Every screen ever looked at served something pre-recorded by someone, somewhere, earlier. The most valuable real estate of the prior internet was the index — the catalog of what had already been made.
That primitive has inverted. Output is now produced originally, in real time, contextually, for one consumer, once. Every pixel, every word, every frame, every recommendation — generated, not retrieved. This is not a refinement of the old computing. It is a different machine emitting a different substrate. Hence the rename: data centers do not produce intelligence; AI factories do.
The unit economics of an AI factory are unlike anything in software history. One rack carries seventy-two accelerators, weighs two tons, costs around four million dollars, and contains roughly one and a half million parts. One gigawatt of factory capacity costs around fifty billion dollars to build and emits three hundred to four hundred billion dollars of token output per year — a revenue-to-capex multiple of six to eight times at the asset level. The total capital flowing into the stack in 2026 is on the order of one trillion dollars, against an implied steady-state ecosystem of roughly twenty trillion dollars per year. This is not a software business — it is a fab. This is the unit economic that breaks every capex-normalization forecast in the market. Forecasters are pricing buildings; operators are pricing production lines.
The second-order consequence is retrieval decay. Every business priced on retrieval economics — search advertising, content aggregation, social feeds, recommendation networks — is in slow runoff. Retrieval businesses paid pennies per query because they re-served stored content with marginal cost approaching zero. Generation businesses meter compute per query because each output is originally produced. The cost curve is different. The pricing model is different. The defensibility is different.
The third-order consequence is the agentic demand shock. Today, intelligence is consumed by roughly one billion humans through chat interfaces. In the agentic phase — already underway — intelligence will be consumed by roughly 100 billion agents running continuously, talking to each other, executing tasks, calling tools, and coordinating workflows. That is a hundredfold demand step change layered on top of the existing consumer demand. The factories under construction are not overbuilt. They are underbuilt relative to a demand curve that has not yet materialized.
The Nine-Layer Cake — A Deeper Reading
The standard investor framing of the AI stack has five layers — energy, compute, infrastructure, models, applications. It is a useful public-facing compression, but it hides at least four separations that have become structurally consequential. A more accurate reading of the stack has nine layers.
The five-layer framing collapses what are now separate profit pools. Networking has separated from compute — rack-scale interconnects consume more than twenty percent of system cost and are growing roughly two hundred percent year-on-year, with one company holding monopoly economics on the interconnect.
Foundries have separated from silicon — allocation power at the leading-edge fab and the high-bandwidth memory oligopoly is now more consequential than chip design itself.
Governance has separated from generic policy and trust — it is now a priced variable in enterprise procurement and a binding constraint on cross-border deployment. Silicon itself has bifurcated into a generalist substrate, hyperscaler custom silicon, and merchant custom challengers. Each of these separations is its own market with its own incumbents, moat, and bottleneck.
Bottom-to-top, the nine layers are:
1 · Energy & Physical. Generation, transmission, water, cooling, land, permits. The first serious grid investment cycle in roughly a century. Power is the binding constraint for everything above it. Rare-earth supply, transformer steel, and grid-scale storage chemistry are the physical chokepoints that govern the pace of the entire stack. Capital intensity is extreme; margin is regulated; the moat is physical and political.
2 · Foundries & Packaging. The fab and the package — the physical conversion of designs into silicon. Allocation power at the leading-edge foundry and the HBM oligopoly is now a more consequential bottleneck than chip design. Advanced packaging (CoWoS and equivalents) is the binding constraint on supply for the next two generations of accelerators. Capital intensity is the highest in the entire economy. Margin is high at the frontier and structurally protected by twenty-plus-year capability gaps to the next-best competitor.
3 · Silicon. Chip design itself — generalist accelerators, hyperscaler custom silicon (custom TPUs, training accelerators, inference accelerators), and merchant custom challengers. The layer has bifurcated into three structurally different positions: the dominant generalist who sets the software stack, the captive customs who optimize for one workload, and the merchant customs who specialize on inference or specific architectures. Margin concentrates at the generalist; volume disperses to the customs.
4 · Networking & Protocols. Rack-scale interconnect, optical fabric, switching silicon, and the protocols stitching factories together. This layer carries monopoly economics on the leading edge — the interconnect is the most under-priced asset in the public market. Networking is now growing faster than compute itself because every additional accelerator demands a non-linear increase in interconnect bandwidth.
5 · Compute Capacity. The actual deployable rack-and-power-contract bundle — what hyperscalers and neoclouds sell. The difference between owning silicon and being able to deploy it at scale is now its own market. Capital intensity is extreme. Margin is high while scarcity holds and will normalize toward utility-like over a decade as the physical layer catches up.
6 · Foundation Models. The visible language models cover one slice. The industrial layer is everything else — protein, cell, genome, materials, physics, climate, robotics control, autonomous driving, code, design, audio. Anything with structure is learnable; anything learnable becomes a model; every model is a vertical market. The layer is shaped by three structural philosophies: maximalist capability frontiers (capability is the moat), trust and safety frontiers (refusal and reliability are the moat), and integrated platform frontiers (distribution into existing ecosystems is the moat). These are not interchangeable strategies — they produce structurally different products and command different multiples. The safety-positioned frontier in particular commands a refusal premium — enterprise buyers pay materially more for a model whose refusal behavior they can underwrite.
7 · Agentic Harness. The orchestration substrate that turns model output into operational output — tool calling, memory, planning, retries, evaluation, guardrails, sandboxing, multi-agent coordination. This is the layer that most application-layer companies confuse with their product. It is in fact a horizontal infrastructure layer in its own right, comparable to the cloud-native infrastructure layer of the prior decade. The companies that own the harness — and the persistent state that agents accumulate inside it — will become the next generation of infrastructure incumbents.
8 · Distribution. Applications, surfaces, and channels — the layer that touches the user. $100B of venture capital deployed in 2025 alone, the largest year in the history of the asset class. It is also the most fragmented layer and the one with the highest failure rate. Most application-layer companies built on a single model capability are absorbed by the next model release. The breakouts here own a workflow, not a feature, and reprice from per-seat to per-token or per-outcome.
9 · Governance. Policy, sanctions, export controls, deemed-export rules, AI safety regulation, model accountability, antitrust. This was treated for years as a soft layer — a horizontal constraint, not a market. It has now become a control plane perpendicular to the entire stack. A single jurisdictional directive can switch off a frontier model across every customer on the planet in seventy-two hours. Governance does not sit on top of the stack — it cuts through it. This is the most consequential reframe in the deeper reading: Layer 9 was never a layer. It is the dimension along which every other layer is conditional.
Two reading rules govern the stack. Each layer is constrained by the one below it — you cannot ship a model without compute, you cannot ship compute without silicon, you cannot ship silicon without foundries, you cannot ship foundries without energy. Each layer is commercialized by the one above it — silicon needs models to monetize, models need a harness, the harness needs distribution. Read top-down to see what gets priced. Read bottom-up to see what gets bottlenecked.
Where profit pools concentrate. Read the stack as a column to see one company’s full vertical posture — only a handful of operators in the world are deeply present across more than three layers, and those that are command the largest market capitalizations on Earth. Read the stack as a row to see who owns one layer in isolation. The diagonal between vertical integration and horizontal monopoly is where the supercycle’s profit pools live. Capex concentrates at L1–L5. Margin concentrates at L3 (silicon), L4 (networking), L6 (models), L7 (harness), and L8 (distribution). Demand concentrates at L8. Bottleneck concentrates at L1. Control concentrates at L9.
Any portfolio that touches only one layer carries dependency risk on all the others. The complete investment thesis specifies which layer is being financed, which adjacent layer the position depends on, and which jurisdictional governance regime the position assumes.
Task vs Purpose — The Cognitive Jevons Paradox
There is a single mechanism that controls how the labor market reorganizes in the intelligence age. It is the same mechanism that controlled the steam age, the electrification age, and the computerization age, and it has been misread every time. The mechanism is the distinction between task and purpose.
A task is the thing a person does in a given hour. A purpose is the reason the task exists. Automation collapses tasks. It does not collapse purposes. And when the underlying purpose is demand-unconstrained — when there is more of it wanted than is currently supplied — automating the task expands employment in the role, not contracts it.
Twelve years ago, a leading voice forecast the extinction of the radiology profession on the strength of superhuman computer vision. Computer vision did saturate radiology. Every reading is now augmented by it. And the number of radiologists is up, scan volume is up, departmental profitability is up. The mechanism: AI made each radiologist more productive, which lowered the cost per scan, which expanded the volume of patients who could be admitted, which increased departmental revenue, which justified hiring more radiologists. The same forecast made about software engineers in 2024 — “ninety percent of coding will be automated” — is being falsified in real time by record hiring at the foundation model labs.
This is a cognitive version of Jevons’ paradox: in 1865, William Stanley Jevons observed that more efficient steam engines did not reduce British coal consumption — they expanded it, because lower cost per unit of work expanded the universe of work being done. Cognitive labor sits in the same regime. Cheaper cognition expands the demand for cognition. The plumber becomes a bathroom designer. The carpenter becomes a home architect. The salesperson becomes an interior consultant. The accountant becomes a strategic advisor. The radiologist becomes a clinician with infinite scan capacity.
Three predictions follow directly. First, aggregate cognitive employment grows, not shrinks — the roles change shape, the headcount expands. Second, roles defined by task automate while roles defined by purpose elevate — a clear definition of purpose is now a labor market moat. Third, the geographic and educational concentration of cognitive work flattens — the prior dominant programming language was known by roughly two percent of humanity, the new programming language is human language known by roughly one hundred percent, and the population of programmers therefore becomes the population of humans. Returns to scale move from technical populations to user populations.
Practical Implications
The doctrine is structural. The implications are operational. They differ sharply depending on where the reader sits in the economy.
For capital allocators. Position across the dependency chain, not inside a single layer. Capex is flowing into the bottom five layers ahead of margin compounding at silicon, networking, models, harness, and distribution — capital runs the layer cycle, margin follows with a twelve-to-twenty-four-month lag. Energy is the binding physical constraint; governance is the binding political constraint. Long both bottlenecks. Retrieval-economics businesses are in slow runoff and need to be repriced as melting equity. Application-layer plays demand extreme defensibility tests, because most are absorbed by the next model release. The next set of category-defining markets sits at the harness layer — agentic infrastructure — and at the vertical foundation model layer (protein, materials, robotics control, finance, legal, design). The screen across the stack: is the company at the frontier of its layer, anchored in a real-economy workflow, embedded in a use case the model cannot replicate, and defensible against the next capability release and the next governance regime.
For operators and founders. Reprice. The per-seat pricing model assumes humans are the consumers of software; in the agentic phase, agents are. Per-token, per-outcome, and per-workflow pricing models replace per-seat. Move from feature to workflow — features get absorbed in the next model release, workflows compound through data feedback loops. Build products that are consumable by an agent calling your API on behalf of a user, not only by a human clicking a button. Data feedback loops are the moat, not the model itself. Treat retrieval-economics competitors as in slow runoff and design around their decay. Assume Layer 9 is a live risk — the governance environment around your jurisdiction, your suppliers, and your customers can change in days, not quarters.
For builders. The cost of building has collapsed. Distribution, judgment, and workflow ownership are the new bottlenecks. The harness layer — memory, orchestration, evaluation, planning, tool calling — is the new infrastructure stack. The companies that win the next decade are not the ones who train models but the ones who own the persistent state agents accumulate as they operate. Code is no longer the artifact; the orchestrated workflow is the artifact. Optimize for build velocity at the small scale and for orchestrated reliability at the large scale.
For knowledge workers and professionals. Audit your role. Distinguish your task from your purpose. The half-life of your role is determined by which side of that line you sit on. Become an orchestrator: direct agents rather than execute tasks personally. Pay for the best available models — the compounding gap between augmented and unaugmented operators is measured in months, not years. Purpose elevates; task automates. Build defensibility through judgment, taste, relationships, and end-to-end ownership of an outcome that requires more than one cognitive substrate.
For governments and sovereigns. Domestic AI capacity becomes a strategic asset on par with energy or military capability. Nations without factory build-out are structurally dependent on those with it. Energy policy is AI policy: grid permits, transmission expansion, and generation mix are now industrial strategy. Layer 9 — the governance control plane — is the lever the most powerful states will use to discipline the entire stack. The countries that build credible, predictable governance regimes attract capital; the countries that wield governance as a unilateral switch lose investability over time. The framework for governing intelligence is not yet written. The states that move first on it set the framework everyone else inherits.
What’s Coming Next
The next twenty-four to thirty-six months are the period during which the doctrine becomes visible to the median observer. Several specific transitions are already underway and will cross into the mainstream in this window.
The agentic phase scales from demo to deployment. Enterprise workflows automated by agent swarms move from proof-of-concept to production at scale. The first generation of agent-native software companies announces material revenue. The conversation shifts from “can agents do this” to “which workflows are still human-only.”
The harness layer crystallizes as its own infrastructure category. Memory, orchestration, evaluation, sandboxing, and multi-agent coordination consolidate into a defined vendor stack. The first generation of harness-layer incumbents emerges — equivalent in role to what databases or observability platforms were to the prior software era. Several of the next decade’s largest software companies are founded here.
Energy becomes the visible binding constraint. Every major AI announcement is qualified by power availability. Hyperscaler capex announcements are paired with power purchase agreements rather than chip orders. Grid permits and transmission expansion become front-page industrial strategy. Nuclear restart, small modular reactor permitting, and natural gas peakers all receive accelerated treatment.
Governance becomes the visible control plane. Cross-border model availability becomes a function of jurisdictional posture, not commercial agreement. The first wave of model-level export controls, sanctions, and deemed-export rules is enforced at scale. Enterprises begin underwriting model choice against governance risk as a first-class procurement criterion. The refusal premium widens.
The model layer bifurcates into three structural philosophies. Capability-frontier labs compete on raw performance at the top of the market. Trust-and-safety frontiers compete on refusal-quality and enterprise underwriting. Integrated-platform frontiers compete on distribution into existing software ecosystems. The middle compresses. The same lab often serves more than one philosophy, but the three positions are increasingly priced differently.
The pricing model breaks the seat. Per-seat SaaS pricing, dominant for two decades, begins its terminal phase. Major incumbents announce per-token, per-outcome, or per-workflow pricing. The seat is no longer the unit, because the user is no longer the only consumer of software — the agent is. Software accounting and procurement reorganize around this.
Vertical foundation models proliferate. Protein, materials, robotics control, finance, design, legal, audio, video, manufacturing — each becomes its own model market with its own emerging incumbents. The “two model companies” framing of 2024 is decisively wrong; by 2027, there are dozens of vertically-anchored model companies with billion-dollar revenue lines.
Physical AI crosses the chasm. Humanoid robots and warehouse automation move from pilot to production deployment. The token-to-action gap closes. Robotics control becomes a structured model market in its own right, attracting capital at scale. The first generation of robot-native operating companies emerges.
Sovereign AI becomes an asset class. Nation-state-level factory build-outs are announced with national strategic framing. National model trusts and sovereign compute pools become standard infrastructure for mid-tier economies. AI capability is treated as a strategic asset comparable to energy reserves or military capacity.
The wrapper shakeout completes. The 2024–2025 vintage of AI application companies shakes out audibly. The majority fail. The breakouts that survive own workflows, not features, and are repriced as category-defining companies. Venture capital recalibrates around defensibility tests rather than novelty.
Cognitive labor reorganization becomes visible. The first wave of purpose-versus-task restructuring at major firms becomes public. Headcount expands in cognitive professions while role shape changes. The first generation of operators who combine product judgment, build velocity, and agent orchestration becomes the dominant high-leverage hire. The talent geography of the prior software era visibly disperses.
Each of these is already in motion. The window during which they remain non-consensus is closing.
Key Takeaways
The Substrate Doctrine — Industrial ages are defined by the substrate their machines emit. A substrate becomes infrastructure when it is generable, transportable, and universally consumable. Intelligence is the third one to pass all three tests.
The AI Supercycle — A substrate cocoon is a generational capital wave by definition. The current trillion-dollar deployment is the early grid build, not the peak. Same arc as electrification, same arc as networking, same century-long downstream reorganization.
The Primitive Inversion — Computing’s core primitive flipped from retrieval to generation after sixty years. Every business priced on retrieval is in slow runoff. Every business priced on generation is on fab-style unit economics.
The Factory Multiple — A fifty-billion-dollar AI factory producing three to four hundred billion dollars per year in token output is a six-to-eight-times revenue-to-capex multiple at the asset level. This is the number that breaks every capex-normalization forecast.
The Agentic Demand Shock — Intelligence demand is moving from roughly one billion human users to roughly one hundred billion agent users on the same physical substrate. The factories under construction are underbuilt for a demand curve that has not yet arrived.
The Nine-Layer Stack — The investor five-layer compression hides four separations that have become structurally consequential: networking from compute, foundries from silicon, governance from generic trust, and silicon itself bifurcating. The accurate reading has nine layers, each constrained by the one below and commercialized by the one above.
Layer 9 Is a Control Plane, Not a Layer — Governance no longer sits on top of the stack. It cuts through it. A single jurisdictional directive can disable a frontier model globally in seventy-two hours. Every thesis must underwrite governance risk as a first-class variable, not a residual.
The Cognitive Jevons Paradox — Automating cognitive tasks expands cognitive demand when the underlying purpose is demand-unconstrained. Task-defined roles automate; purpose-defined roles elevate.
The Cognitive Equalizer — When the programming interface becomes natural language, the population of programmers becomes the population of humans. The talent geography of the next twenty years looks nothing like the last twenty.
Recap: In This Issue!
Intelligence Is Becoming Infrastructure
The core thesis is simple: every industrial era is built on a new substrate.
The Industrial Revolution scaled physical work through electricity.
The Internet scaled information and coordination through networks.
AI is scaling cognition through intelligence.
The implication is that AI should not be viewed as another software cycle. It is the emergence of a new economic substrate that can be generated, distributed, and consumed at scale.
We Are in the “Dynamo Phase,” Not the Application Phase
Most people focus on chatbots and AI applications.
The bigger story is the construction of the underlying infrastructure:
Energy
Data centers
Chips
Networks
AI factories
Just as the light bulb was only one application of electrification, ChatGPT is only one application of a much larger intelligence infrastructure buildout.
Computing Has Shifted from Retrieval to Generation
For decades, computers primarily retrieved information that already existed.
AI introduces a new primitive:
Generation.
Instead of finding content, systems create content in real time:
Text
Code
Images
Video
Decisions
This changes the economics of the internet and puts pressure on business models built around retrieval and aggregation.
The AI Stack Is Deeper Than Most Investors Realize
The common view of AI focuses on models and applications.
The article argues the real stack includes:
Energy
Foundries
Silicon
Networking
Compute
Models
Agent Infrastructure
Distribution
Governance
The key insight is that every layer depends on the one below it and monetizes through the one above it. Bottlenecks and profits emerge in different places.
Agents Create the Next Demand Shock
Today’s AI economy serves roughly a billion humans.
Tomorrow’s AI economy may serve tens of billions of autonomous agents:
Executing workflows
Calling tools
Coordinating tasks
Interacting with other agents
This suggests today’s infrastructure may ultimately prove insufficient rather than excessive.
The Cognitive Jevons Paradox
The common assumption is:
AI automates work, therefore jobs disappear.
History suggests a different outcome.
When the cost of a capability falls, demand often rises.
The critical distinction is between:
Tasks → automated
Purpose → elevated
The highest-value professionals will increasingly act as orchestrators, directing intelligence rather than manually executing every task.
Governance Is Becoming a Strategic Layer
Regulation, export controls, and national AI policies are no longer peripheral concerns.
They are becoming a control plane for the entire ecosystem.
Future AI leadership will depend not only on technological capability but also on:
Energy access
Compute sovereignty
Regulatory frameworks
Geopolitical positioning
AI is increasingly becoming a matter of national strategy, not just corporate competition.
Bottom Line
The most important takeaway is that we are not witnessing the rise of a new software category. We are witnessing the emergence of a new industrial substrate.
If electricity transformed physical labor and the internet transformed information flow, AI is positioned to transform cognition itself. Most discussions today focus on applications, while the real story is the construction of the intelligence infrastructure that will power the next several decades of economic activity.
With massive ♥️ Gennaro Cuofano, The Business Engineer