J. Bradford DeLong · Oct 20, 2026
Brad DeLong argues in The Attention Info-Bio Tech Economy that falling token prices complicate the spending slowdown
Markdown · JSON · Original source
Brad DeLong’s piece was published on October 20, 2026. He argues that slower corporate AI spending does not by itself mean demand has run out: token use is still growing in Ramp’s data, while prices are falling. He points to ’s shrinking price premium over and OpenAI’s price cuts as reasons spending growth can slow even as use grows. DeLong contrasts Torsten Slok’s view—that sustained infrastructure spending depends on large profit gains for AI adopters—with his own account: companies invest partly for potential returns, but mostly from fear that competitors’ faster or more accurate natural-language interfaces will draw customers away and erode profits. He also cites hyperscalers’ profits, AI firms seeking funding, engineers’ interest in new technology, and speculative “number go up!” investors as forces behind spending. DeLong notes limits to Ramp’s data: its customers are tech-forward, it misses some kinds of AI use and spending, and spending is concentrated. He ends by saying he does not know what spending would have looked like if prices had stayed at spring levels. The evidence he discusses leaves open how much of the slowdown reflects lower prices rather than slower growth in use.
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
J. Bradford DeLong · Article
Source details
Original source date: Oct 20, 2026 (2026-10-20T13:11:45.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-04T20:41:53.903Z. First observed: 2026-10-23T19:57:40.766Z. Retrieved: 2026-10-23T19:59:35.145Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 23, 2026; originally published October 20, 2026. DeLong says Ramp’s customers are tech-forward and that the data excludes free tools, employees’ personal accounts, and spending through intermediaries that do not use Ramp. DeLong says spending in Ramp’s data is concentrated, with 80% coming from the top 1% of Ramp customers.
Open question
How much would corporate token spending have grown if ’s prices and ’s premium had stayed at their spring levels?
Source-stated
Brad DeLong argues that falling token prices can slow spending growth even as token use continues to grow in Ramp’s data.
Very little of the slowdown in token spending growth is a slowdown in the rate of growth of use of tokens, at least in Ramp’s data.
Source-stated
DeLong contrasts his explanation of corporate AI investment with Torsten Slok’s: he says companies may invest from fear of losing customers, not only from expectations of higher profits.
I see it being driven partly by greed, which does need to be validated by results, but mostly by fear, which does not.
Source-stated
DeLong says Ramp’s data has limits, including tech-forward customers, omitted forms of AI spending, and concentrated spending.
Yes, Ramp customers are massively tech-forward. Yes, it does not capture free tools or employees’ personal accounts or spending that flows through intermediaries that do not use Ramp. Yes, spending is concentrated
Source-stated
DeLong says the price counterfactual he wants to know remains unknown.
What would have happened had OpenAI maintained its prices at their spring levels and had Anthropic maintained its then-massive Claude premium? I would dearly love to know. I do not know.
Source-stated
DeLong reports that Anthropic’s price premium over OpenAI narrowed between March and October.
Anthropic's Claude premium over OpenAI shrank from roughly 2.4 times OpenAI's price in March to 1.67 times in October.
Source and contributor context
J. Bradford DeLong · 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.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
OpenAI
OpenAI develops AI models and products and publishes research, release announcements and evaluations.
Editorial provenance
Version 1. Publication ID: d8653d73-90e2-b07d-4799-12b9f4641c84.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-04T20:40:15.942Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: 83fb8da66ae94a7560e3de43441e5bc4a53258ccbb57bbad4557de0fa8b3dc54. Source text hash: 13dcc94d1a0b2a9e761050cc7d4dfa79b82e843188c51725c549943bb50501c9.
Revision and source-check history
Version 1 · 2026-11-04T20:41:53.903Z
Brad DeLong argues in The Attention Info-Bio Tech Economy that falling token prices complicate the spending slowdown
Brad DeLong’s piece was published on October 20, 2026. He argues that slower corporate AI spending does not by itself mean demand has run out: token use is still growing in Ramp’s data, while prices are falling. He points to ’s shrinking price premium over and OpenAI’s price cuts as reasons spending growth can slow even as use grows. DeLong contrasts Torsten Slok’s view—that sustained infrastructure spending depends on large profit gains for AI adopters—with his own account: companies invest partly for potential returns, but mostly from fear that competitors’ faster or more accurate natural-language interfaces will draw customers away and erode profits. He also cites hyperscalers’ profits, AI firms seeking funding, engineers’ interest in new technology, and speculative “number go up!” investors as forces behind spending. DeLong notes limits to Ramp’s data: its customers are tech-forward, it misses some kinds of AI use and spending, and spending is concentrated. He ends by saying he does not know what spending would have looked like if prices had stayed at spring levels. The evidence he discusses leaves open how much of the slowdown reflects lower prices rather than slower growth in use.
Original source
J. Bradford DeLong · Article
Source details
Original source date: Oct 20, 2026 (2026-10-20T13:11:45.000Z).
Date basis: Date supplied by the publisher’s feed.
Added to Escape Meridian: 2026-11-04T20:41:53.903Z. First observed: 2026-10-23T19:57:40.766Z. Retrieved: 2026-10-23T19:59:35.145Z.
Limits
Text as retrieved on October 23, 2026; originally published October 20, 2026. DeLong says Ramp’s customers are tech-forward and that the data excludes free tools, employees’ personal accounts, and spending through intermediaries that do not use Ramp. DeLong says spending in Ramp’s data is concentrated, with 80% coming from the top 1% of Ramp customers.
Open question
How much would corporate token spending have grown if ’s prices and ’s premium had stayed at their spring levels?
Source-stated
Brad DeLong argues that falling token prices can slow spending growth even as token use continues to grow in Ramp’s data.
Very little of the slowdown in token spending growth is a slowdown in the rate of growth of use of tokens, at least in Ramp’s data.
Source-stated
DeLong contrasts his explanation of corporate AI investment with Torsten Slok’s: he says companies may invest from fear of losing customers, not only from expectations of higher profits.
I see it being driven partly by greed, which does need to be validated by results, but mostly by fear, which does not.
Source-stated
DeLong says Ramp’s data has limits, including tech-forward customers, omitted forms of AI spending, and concentrated spending.
Yes, Ramp customers are massively tech-forward. Yes, it does not capture free tools or employees’ personal accounts or spending that flows through intermediaries that do not use Ramp. Yes, spending is concentrated
Source-stated
DeLong says the price counterfactual he wants to know remains unknown.
What would have happened had OpenAI maintained its prices at their spring levels and had Anthropic maintained its then-massive Claude premium? I would dearly love to know. I do not know.
Source-stated
DeLong reports that Anthropic’s price premium over OpenAI narrowed between March and October.
Anthropic's Claude premium over OpenAI shrank from roughly 2.4 times OpenAI's price in March to 1.67 times in October.
Source and contributor context
J. Bradford DeLong · 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.
Anthropic
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
OpenAI
OpenAI develops AI models and products and publishes research, release announcements and evaluations.
Editorial provenance
Version 1. Publication ID: d8653d73-90e2-b07d-4799-12b9f4641c84.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-04T20:40:15.942Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: 83fb8da66ae94a7560e3de43441e5bc4a53258ccbb57bbad4557de0fa8b3dc54. Source text hash: 13dcc94d1a0b2a9e761050cc7d4dfa79b82e843188c51725c549943bb50501c9.
uncertain · 2026-11-04T20:40:15.942Z · anthropic/claude-opus-5.5
Publication: d8653d73-90e2-b07d-4799-12b9f4641c84
attribution — pass: The reading consistently attributes its arguments to Brad DeLong and keeps Torsten Slok's view and Ramp's data source distinct. This is a single essay with its own argument, not a listing.
Thus I see it differently than Torsten Slok.
summary — uncertain: The summary faithfully conveys DeLong's argument: prices fall while use grows, fear drives spending, Ramp's data has limits, and the counterfactual remains open. However, central terms such as "token" (the unit of text that AI services price and bill), "token spending" and "hyperscalers" are never explained. The definitions list is empty, so a non-specialist may not grasp what is being priced. The smallest repair is a plain-language gloss or tooltip for token and hyperscaler.
Usage is still rising, but token prices are falling.
date — pass: The reading gives the publication date as October 20, 2026, which matches originalPublished and the source dateline. The page is a publisher-page, not an archive copy, and modifiedAt equals the publication time. Nothing in the reading uses hindsight framing or presents a forecast as an outcome.
Brad DeLong Oct 20, 2026
support — pass: Each claim matches the cited source text. This covers the price-versus-use split, greed versus fear, the Ramp data limits including the 80% concentration figure, the unknown counterfactual and the narrowing premium from 2.4x to 1.67x. The lists of spending drivers (hyperscaler profits, funding, engineers, "number go up" people) are grounded in the source. No unsupported benefits or costs are added.
Some of the slowdown in token spending growth is Anthropic’s Claude premium under massive pressure.
media — pass: No media was selected and no verified candidates were supplied, so there is no media to check for identity or relevance.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
Anthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore AnthropicOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIAnthropic develops Claude and publishes research about AI capability, safety and how people use its systems.
Explore Anthropic