Ben Thompson is a technology-business analyst and the founder and author of Stratechery (2013–present), a subscription publication on platform strategy, competition, and the business of technology. He is the originator of the analytical frameworks Aggregation Theory and the FANG Playbook, and is a frequent commentator on AI-industry strategy and competition. He hosts the Sharp Tech, Sharp China, and Dithering podcasts and is based in Taipei, Taiwan, with periodic US travel.
Analytical frameworks
Thompson's frameworks are widely used in commentary on AI-industry platform power, competition, and incumbents-versus-entrants dynamics.
Aggregation Theory (2015) describes a dynamic in which the collapse of distribution costs makes user experience the primary moat, producing winner-take-all aggregators. He refined it in Defining Aggregators (2017) into a three-level classification by supply relationship (supply-acquisition, supply-transaction-cost, and zero-supply-cost) plus a Super-Aggregator subset (Google, Facebook). The framework supplies vocabulary used to analyze whether OpenAI, Anthropic, Cursor, and others are aggregators, integrators, or a new category.
The FANG Playbook (2016) treats Facebook, Amazon, Netflix, and Google as a single archetype, each of which subsumed rather than disrupted incumbents by owning the consumer entry point. It appears in The FANG Playbook — Ben Thompson (Stratechery, 2016) and recurs as an analogy for whether AI labs are following the same path.
In What Clayton Christensen Got Wrong (2013), Thompson argues that Clayton Christensen's low-end disruption theory applies only to business-to-business markets. He distinguishes new market disruption, which he accepts, from low-end disruption, which he rejects for consumer markets where user experience cannot be overshot. He cited Benedict Evans's "PC market was mostly corporate" framing in making the argument. Thompson applies this position to whether DeepSeek, Llama, gpt-oss, and similar releases constitute low-end disruption of frontier labs; his position is that they do not, because in consumer AI user experience matters more than price.
Positions and statements on AI policy
Thompson is a frequent foundational-essay source on contemporary AI policy. His positions are paraphrased below as arguments, not as findings.
In Anthropic and Alignment (March 2026), Thompson characterizes Anthropic's standoff with the Department of War over autonomous weapons and mass surveillance as "intolerable and misaligned with reality." He argues that AI on Dario Amodei's stated trajectory rivals nuclear weapons, and that the US would be incentivized to destroy any private actor seeking veto power over the military. The essay is a companion piece to AI Promise and Chip Precariousness (Feb 2025), which addresses Taiwan and TSMC strategy. In the same essay he argues that "open source is ultimately safer" (Anthropic and Alignment §"Complex Systems").
In Attenuating Innovation (AI) (October 2023), Thompson critiques Biden's AI Executive Order 14110 as regulatory capture in the form of compute-threshold reporting; the essay is a reference point for the position that regulation favors incumbents.
In Training AI is Not Fair Use? (April 2025), Thompson offers a mixed analysis of the Bartz v. Anthropic decision and the broader debate over intellectual property and training data.
In AI and the Human Condition (March 2026), Thompson argues that AI's defining cultural impact will fall on relationships and the experience of being human rather than on productivity.
In Mythos, Muse, and the Opportunity Cost of Compute (April 2026), Thompson compares the release strategies of Anthropic's Claude Mythos Preview, OpenAI's GPT-5.5 ('Spud'), and Meta's Muse Spark (Meta Superintelligence Labs) as strategic-positioning bets about scarce compute.
In "Who's Afraid of Chinese Models?" (July 20, 2026), written after Moonshot's Kimi K3 release and the Hugging Face agentic breach, Thompson argues the U.S. administration should loosen cybersecurity restrictions on Anthropic's Fable and its peers and level the field for U.S. open-weight developers, citing Hugging Face's account that U.S. frontier-API guardrails blocked its incident-response forensics while a self-hosted Chinese open-weight model did the work (Who's Afraid of Chinese Models? (Ben Thompson, Stratechery, July 2026)). See Open-Weight Frontier Models, Defensive AI Paradox.
Disclosures
Thompson lives in Taipei and discloses his Taiwan ties when commenting on Taiwan, TSMC, and China geopolitics. Stratechery is a paid-subscription publication; Thompson does not take outside investment, though he sometimes discloses positions when commenting on specific tickers.
Relationships
- related: Aggregation Theory — Thompson's 2015 framework, anchor concept page
- related: Benedict Evans — frequent counterpart and co-citer in the same analytical milieu (Thompson cited Evans's "PC market was mostly corporate" framing in What Christensen Got Wrong)
- related: Clayton Christensen — primary target of Thompson's 2013 critique
- related: Dario Amodei — primary target of Thompson's 2026 "Anthropic and Alignment"
- contradicts: Anthropic and Alignment — Ben Thompson (Stratechery, March 2026) frames Anthropic's position on autonomous weapons and surveillance as untenable
- supports: Open-source AI — Thompson is an open-source advocate; argues "open source is ultimately safer" (Anthropic and Alignment §"Complex Systems")
- supports: Anthropic v. US Government — Thompson's framing prefigures the legal posture of that suit