Azeem Azhar’s first bottom-up measure of the generative-AI economy puts annualized software revenue at $175 billion — and the growth curve is running three times faster than the mobile or internet waves did at the same stage.
State of the AI Economy — Exponential View, June 2026
On June 25, 2026, Azeem Azhar’s Exponential View published the State of the AI Economy — the first publicly available, bottom-up, deduplicated measure of generative-AI commercial revenue. The headline figures: $110 billion in booked revenue over the trailing twelve months, and an annualized run rate now exceeding $175 billion. The ~1.6× gap between those two numbers is not rounding error — it is the encoded growth velocity of the market in real time.
The methodology matters as much as the numbers. The dataset covers app-layer, foundation-model, and infrastructure-hosting revenue across 1,000-plus companies. It is deduplicated — so API revenue that flows through an app is not double-counted at the model layer — and it is global ex-China. Critically, it excludes chip manufacturing: this is a software and services number, not an Nvidia hardware figure. When people cite the $175B, they need to carry that asterisk.
Paul Graham amplified the report on social media, zeroing in on the growth-rate comparison: generative AI revenues are expanding roughly three times more rapidly than the mobile or internet waves did at an equivalent stage of development. That framing — not the absolute dollar figure — is the more important signal for anyone trying to calibrate how fast the platform curve is moving.
How the Numbers Stack Up
Trailing 12 Months — Booked Revenue
$110B in deduplicated, global ex-China GenAI software and services revenue across app, foundation model, and hosting layers. Excludes chip manufacturing.
Current Annualized Run Rate
$175B — not booked revenue, but the pace at which the market is generating revenue right now. The ~1.6× gap to the trailing figure encodes the acceleration.
Paul Graham Amplifies — June 2026
Graham spotlights the 3× growth-rate advantage over mobile and internet at comparable stages — reframing the story from “big number” to “historic velocity.”
The Structural Gap — Still Open
Industry capex and burn continues to run ahead of $110B in booked revenue. Demand is real and compounding; the subsidy era has not ended.
The key insight: The $175B run rate is not a forecast — it is the current velocity of a market that is already here. The more consequential number is the 3× growth advantage over mobile and internet, because it tells you the compounding period is shorter than any prior platform shift. Every strategic decision calibrated to a “mobile-speed” AI rollout is already running behind.
The Structural Read
For two years, skeptics have argued that AI revenue was overstated, double-counted, or narrowly concentrated in a handful of enterprise pilots. The Exponential View dataset — bottom-up, deduplicated, cross-stack — is the first serious answer to that argument. The demand is real, it is broad, and it is compounding.
But here is where the honest structural read diverges from the bullish headline: $110 billion in software revenue still sits below the industry’s total capex and operational burn. The hyperscalers are spending at a pace that the current revenue base does not yet justify on a standalone basis. Which means the AI economy in 2026 has two simultaneous truths — the AI Supercycle is confirmed, and the Subsidized AGI Economy is still running. Revenue is proving the demand signal; spend is betting on the compounding effect making the gap irrelevant within a short window.
Using the Map of AI lens: the revenue mass is not evenly distributed across the stack. Infrastructure hosting and foundation-model API revenue are growing fastest in absolute terms, but app-layer monetization is where margin will eventually concentrate — because that is where the end-user relationship lives. The companies that establish distribution and recurring contracts at the app layer now are building the moat that infrastructure spend is, paradoxically, subsidizing for them.
“These revenues are growing roughly three times more rapidly than the mobile or Internet waves.”
Map of AI — Stack Position Analysis
The Revenue Gravity Moves Up the Stack
Infrastructure and foundation-model layers are generating revenue at scale today — but they are structurally price-competitive and increasingly commoditized. The durable margin capture will happen at the app layer, where switching costs, data network effects, and user relationships compound. The $175B run rate is today’s number; the map of who captures tomorrow’s margin looks very different from who is winning the infrastructure race right now.
Three Implications
IMPLICATION 1 — The Supercycle Is Not Hype, It Is a Measurable Phenomenon
A deduplicated, bottom-up $110B in trailing software revenue — across 1,000-plus companies, ex-China, ex-chips — is the end of the “AI revenue is vaporware” argument. The demand is distributed, not concentrated in one vendor’s ARR. That breadth is what makes the 3× growth-rate comparison to mobile and internet structurally meaningful rather than a marketing talking point.
IMPLICATION 2 — The Subsidy Era Is Not Over, and That Changes Competitive Strategy
Because hyperscaler capex continues to exceed the current software revenue base, infrastructure pricing remains artificially suppressed. App-layer builders are effectively operating in a subsidized cost environment. The strategic question is not whether to exploit that — every rational actor should — but how to structure the business so it survives when the subsidy normalizes. Lock in long-term compute contracts and user relationships now, before the cost floor rises.
IMPLICATION 3 — The 3× Velocity Collapses the Strategic Planning Window
Mobile and internet gave incumbents years to adapt before platform-native competitors became existential threats. At 3× the growth velocity, that window compresses to months, not years. Enterprises that are still in “pilot mode” or treating AI as a 2027 budget line are not being cautious — they are ceding compounding ground to competitors who are already on the revenue curve. The Exponential View data makes the urgency quantitative, not rhetorical.
Business Engineer Framework
The Map of AI — Where $175B Is Actually Being Made
The Exponential View data confirms the revenue is real — but it does not tell you which layer of the AI stack captures durable margin. The Map of AI breaks the full ecosystem into nine layers across 200+ companies, showing where pricing power concentrates, where commoditization is accelerating, and where the next $175B will compound. If the Supercycle is confirmed, the next strategic question is position — and this is the tool for answering it.
Exponential View just handed the market its first credible, methodology-sound revenue baseline for the generative-AI economy: $110 billion booked, $175 billion annualized run rate, growing at three times the pace of mobile and internet — and still, structurally, running below the industry’s total capex burn. Both halves of that sentence are true simultaneously, and the companies that build strategy around only one of them will be wrong. The AI Supercycle is no longer a thesis; it is a measured, compounding, historical fact. The Subsidized AGI Economy is the only honest explanation for why the numbers can be this large and the structural gap still open. Navigate both, or get caught by either.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
