AI Policy Wiki
Dashboard
⚲
AI Policy Wiki › People & Organizations

Azeem Azhar

high confidence · updated 2026-09-08

Founder and author of the Exponential View newsletter; AI-and-tech-economy analyst known for the AI-bubble framing anchored on compute and rental-price data, the AGI-timelines revealed-preferences test, the May 2026 'backlash growing faster than revenues' thesis, and lead authorship of the June 2026 report The State of the AI Economy.

Contents

Azeem Azhar is a UK-based analyst on the AI and tech-economy frontier and the founder and author of Exponential View, a newsletter and podcast. He previously worked as a journalist (Reuters, BBC, and The Economist's technology coverage) and as a tech entrepreneur (PeerIndex). His 2021 book is Exponential: How Accelerating Technology Is Leaving Us Behind and What to Do About It. His recurring frame contrasts technological exponentials with the slower, more linear speed of institutional response.

Background

Azhar's career spans journalism and technology entrepreneurship before his current work as an independent analyst. He reported on technology for Reuters, the BBC, and The Economist, and founded the company PeerIndex. He publishes the Exponential View newsletter and podcast at exponentialview.co, and authored Exponential: How Accelerating Technology Is Leaving Us Behind and What to Do About It (2021).

Positions and statements

A consistent through-line in Azhar's writing is that institutions — regulators, education systems, and labor markets — move on linear timescales while AI capability and adoption move on exponential timescales. This frame is compatible with both AI as Normal Technology (under its long-tail interpretation) and with more aggressive AGI Timelines views.

On the AI-bubble question, Azhar's framing sits between bear and bull anchors (AI Bubble Debate), closest to the position that the industrial bid runs ahead of social licence while underlying revenue is real. He anchors his bubble-and-not-bubble assessments on physical compute and rental-price data rather than valuation multiples, pushing back against purely financial bubble framings.

Compute economics

An Exponential View data column (May 4, 2026) reported that B200 GPU rental prices grew 114% in six weeks, which Azhar presented as evidence that demand still substantially exceeds supply at the frontier. The figure is cited on Nvidia & TSMC — AI Compute Infrastructure § Compute crunch (Source: exponentialview.co).

The State of the AI Economy (June 2026)

Azhar is the lead author of "The State of the AI Economy," a sixty-six-page report published June 25, 2026 with William Gildea, Hannah Petrovic, Nathan Warren and Marija Gavrilov. Its stated motivation is that the supply side of the AI economy is well documented while "the demand side, what customers are actually paying for and if the revenues are real, has been obscure." Built from a bottom-up, confidence-scored, deduplicated line-level revenue model across more than 1,000 firms, it reports $110 billion of banked trailing-twelve-month generative-AI revenue at a $175 billion annualized pace, global excluding China, and argues that AI demand "is more revenue-validated than any prior platform shift" while reducing the investment case to "whether falling prices can move enough token volume to earn a return on CapEx" (The State of the AI Economy (Exponential View, June 2026)).

The report continues Azhar's practice of anchoring the bubble question on physical and unit-economic data rather than valuation multiples: its central tests are depreciation coverage of capex, the sensitivity of that coverage to assumed chip life, and the price elasticity of token demand.

AGI timelines and revealed preferences

In EV #573 (May 9, 2026), Azhar reframed Clark's Import AI 455 into three testable revealed-preference patterns: research-multiplier hiring, capex over-rotation on physical bottlenecks, and tolerance for "ugly near-term cash burn." The discussion connects to AGI Timelines (Source: exponentialview.co).

Social licence and backlash

Azhar's May 23, 2026 column, "The AI backlash is the only thing growing faster than AI revenues," argues that the operative near-term constraint on AI deployment is loss of public licence rather than capability ceilings. The column drew together the mid-May episodes of the Schmidt commencement booing, the AOC tap-water hearing, and Anthropic-profit reporting as evidence for that thesis. The framing carries over to AI Labor Disruption § "Backlash growing faster than revenues" (Source: exponentialview.co; exponentialview.co).

Labor and AI

An Exponential View column (May 3, 2026) surfaced a Chinese-courts ruling that replacing a worker with AI is not, by itself, a lawful basis for firing — among the first such labor decisions globally. The item is referenced on AI Labor Disruption (Source: exponentialview.co).

AI-economy revenue estimate

Azhar and Hannah Petrovic put annualised revenue for the AI economy at $229 billion as of the end of August 2026, up 3.5 times in one year, with trailing twelve-month revenue of $140 billion against $44 billion in August 2025, in an estimate published September 7, 2026. They state the figures are drawn from their own research rather than from reported company results. The same post records that Snowflake cut full-year product gross-margin guidance from 75% to 74%, saying fast-growing AI workloads "carry a lower contribution margin today," and that AI-attributed United States job cuts fell in August, ending a five-month run as the most-cited reason for layoffs while still accounting for 22% of the year's cuts (Source: exponentialview.co). See AI Bubble Debate, Inference Economics and Token Pricing, AI Labor Disruption.

Relationships

Sources

Categories: People & Organizations | media | analyst | ai-economics | exponential-view
What links here (5)