# Ben Thompson argues in Stratechery that model makers are best placed to harness AI [Canonical reading](https://escapemeridian.com/readings/ben-thompson-argues-in-stratechery-that-model-makers-are-best-e106dd89) Ben Thompson’s Stratechery newsletter, published March 13, 2026, argues that the latest evidence points to model makers as the companies best able to harness AI models. The question is whether models would become commodities: Microsoft had bet on integration with OpenAI, while later focusing on infrastructure built around models. Thompson says Copilot Cowork is among the most recent evidence for a different advantage: model makers’ ability to put their own models to use. That distinction matters because it bears on where advantage in the evolving AI value chain may sit—not only in models or the infrastructure around them, but in how model capabilities are integrated. The newsletter points readers to a deeper discussion of those layers on Sharp Tech. It does not explain what evidence, beyond naming Copilot Cowork, supports this shift, or whether that advantage will endure. Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims. ## Original source Ben Thompson · Article Original source date: Mar 13, 2026 (2026-03-13T17:00:00+00:00). Date basis: Date stated on the publisher’s page. Added to Escape Meridian: 2026-11-11T02:23:35.885Z. First observed: 2026-11-10T13:15:55.649Z. Retrieved: 2026-11-10T13:17:38.267Z. - [2026.11: Winners, Losers, and the Unknown](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Limits Text as retrieved on November 10, 2026; originally published March 13, 2026. The newsletter names Copilot Cowork as evidence for its view but gives no further evidence or comparison in this passage. ## Open question What evidence beyond Copilot Cowork would show whether model makers’ ability to integrate their models is a durable advantage over infrastructure built around models? ## Source-stated Ben Thompson says the latest evidence, particularly Copilot Cowork, points to model makers being best able to harness model capabilities. > the most recent evidence — particularly Copilot Cowork — is that the companies who are best able to harness (pun intended) model capabilities are the model makers themselves. - [Supporting source](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Source-stated Thompson describes whether models would become commodities as a long-running question about AI, and says Microsoft once bet on integration with OpenAI. > One of the most important and longest-running questions about AI has been whether or not models would be commodities; Microsoft once bet on their integration with OpenAI - [Supporting source](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Source-stated Thompson says Microsoft later made the bet that infrastructure around models would matter more than the models themselves. > but in recent years has made the best that the infrastructure they can build around models will matter more than models themselves. - [Supporting source](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Source and contributor context Ben Thompson · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## OpenAI OpenAI develops AI models and products and publishes research, release announcements and evaluations. - [Explore this context](/entities/openai) - [Reference](https://openai.com/news/) ## Editorial provenance Version 1. Publication ID: bfa565e1-a512-1215-1ff7-fc01ab5d8acf. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-11T02:19:08.826Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical. Observation: c49ecd5f862d285a4c22613f8fc931139274627e77a066fa62ff8b0e63c6ffbb. Source text hash: 970e3f9671e651e6cf21270f6bdf48f70a115b1da2ed56d913e397701798da66. ## Revision and source-check history ### Version 1 — 2026-11-11T02:23:35.885Z Ben Thompson’s Stratechery newsletter, published March 13, 2026, argues that the latest evidence points to model makers as the companies best able to harness AI models. The question is whether models would become commodities: Microsoft had bet on integration with OpenAI, while later focusing on infrastructure built around models. Thompson says Copilot Cowork is among the most recent evidence for a different advantage: model makers’ ability to put their own models to use. That distinction matters because it bears on where advantage in the evolving AI value chain may sit—not only in models or the infrastructure around them, but in how model capabilities are integrated. The newsletter points readers to a deeper discussion of those layers on Sharp Tech. It does not explain what evidence, beyond naming Copilot Cowork, supports this shift, or whether that advantage will endure. ## Original source Ben Thompson · Article Original source date: Mar 13, 2026 (2026-03-13T17:00:00+00:00). Date basis: Date stated on the publisher’s page. Added to Escape Meridian: 2026-11-11T02:23:35.885Z. First observed: 2026-11-10T13:15:55.649Z. Retrieved: 2026-11-10T13:17:38.267Z. - [2026.11: Winners, Losers, and the Unknown](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Limits Text as retrieved on November 10, 2026; originally published March 13, 2026. The newsletter names Copilot Cowork as evidence for its view but gives no further evidence or comparison in this passage. ## Open question What evidence beyond Copilot Cowork would show whether model makers’ ability to integrate their models is a durable advantage over infrastructure built around models? ## Source-stated Ben Thompson says the latest evidence, particularly Copilot Cowork, points to model makers being best able to harness model capabilities. > the most recent evidence — particularly Copilot Cowork — is that the companies who are best able to harness (pun intended) model capabilities are the model makers themselves. - [Supporting source](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Source-stated Thompson describes whether models would become commodities as a long-running question about AI, and says Microsoft once bet on integration with OpenAI. > One of the most important and longest-running questions about AI has been whether or not models would be commodities; Microsoft once bet on their integration with OpenAI - [Supporting source](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Source-stated Thompson says Microsoft later made the bet that infrastructure around models would matter more than the models themselves. > but in recent years has made the best that the infrastructure they can build around models will matter more than models themselves. - [Supporting source](https://stratechery.com/2026/winners-losers-and-the-unknown/) ## Source and contributor context Ben Thompson · Publisher The publisher is identified here. No individual author has been inferred. ## Source media Media retrieval status: unchecked. ## Terms and references Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement. ## OpenAI OpenAI develops AI models and products and publishes research, release announcements and evaluations. - [Explore this context](/entities/openai) - [Reference](https://openai.com/news/) ## Editorial provenance Version 1. Publication ID: bfa565e1-a512-1215-1ff7-fc01ab5d8acf. Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5. Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-11T02:19:08.826Z. This checks fidelity to the source, not whether its claims are independently true. Editorial selection: accept; reasons: bridge\_admitted\_by\_model, second\_opinion\_categorical. Observation: c49ecd5f862d285a4c22613f8fc931139274627e77a066fa62ff8b0e63c6ffbb. Source text hash: 970e3f9671e651e6cf21270f6bdf48f70a115b1da2ed56d913e397701798da66. ### uncertain — 2026-11-11T02:19:08.826Z — anthropic/claude-opus-5.5 Publication: bfa565e1-a512-1215-1ff7-fc01ab5d8acf attribution — pass: This is a weekly roundup with commentary in the authors' own words, so it counts as one piece rather than a bare listing. The integration paragraph is signed by Ben Thompson, and the reading attributes the argument to him correctly. The basketball and China items come from Andrew Sharp, and the reading does not draw on them. > If none of that makes sense, Andrew and I do a much more extensive deep dive on these different layers of the evolving AI value chain on this week’s episode of Sharp Tech. — Ben Thompson summary — uncertain: Copilot Cowork is the central evidence, but the reading never explains it. It also leaves out Anthropic, the model maker whose integration Thompson credits; the source says Microsoft is building a new bundle on top of it. A general reader cannot tell which model maker is meant or why a Microsoft product counts as evidence for model makers. Smallest repair: say Copilot Cowork is a Microsoft product built on Anthropic's integration. The reading also drops 'commoditize its complements' and the '(pun intended)' voice, but those losses are minor. > Microsoft is seeking to commoditize its complements, but Anthropic has a point of integration of their own; it’s good enough that Microsoft is making a new bundle on top of it. date — pass: originalPublished is 2026-03-13T17:00Z, and the reading states March 13, 2026, which matches the UTC day. The issue number 2026.11 is consistent with that date. The reading uses no later material, presents no forecast as an outcome, and has no hindsight framing. The kind is publisher-page, so no archive disclosure is required. > 2026.11: Winners, Losers, and the Unknown support — pass: All three claims quote the source verbatim and keep it attributed, including the source's typo 'best' for 'bet'. The point about where advantage in the AI value chain may sit stays within Thompson's framing of integration and value-chain layers. The stated limits, that no further evidence is given in this passage and durability is unknown, are fair and do not overreach. > the most recent evidence — particularly Copilot Cowork — is that the companies who are best able to harness (pun intended) model capabilities are the model makers themselves. media — pass: No media was selected, and no verified candidates were supplied. Selecting none is appropriate. ## Recorded references Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter. No public references recorded here yet. [Reference data (JSON)](/readings/e106dd8974b2d35614c78795e9c0683970aef72f0c8c593dc58f7a520efb05d3/references.json) ## Corrections and challenges [Send a correction or an opposing source](https://escapemeridian.com/corrections/) [JSON record](https://escapemeridian.com/readings/ben-thompson-argues-in-stratechery-that-model-makers-are-best-e106dd89.json)