Anthropic published a report on 5 November 2026 that studies 207,049 Claude and Claude Code research sessions. It says data analysis accounts for 33.7% of sessions on Claude and 48.1% on Claude Code, against 27.6% of researcher time. It also says computer science makes up 4.3% of the world’s researchers but 32.7% of Claude Code sessions, about 7.6 times its share by our arithmetic.
Business Pill · Verification Cost
An evergreen Business Engineer explainer on verification cost. It teaches the general idea only and does not discuss Anthropic’s report.
The key insight: Anthropic’s report says data analysis accounts for 33.7% of Claude sessions and 48.1% of Claude Code sessions, against 27.6% of researcher time, and that computer science takes 32.7% of Claude Code sessions with 4.3% of researchers, about 7.6 times its share by our arithmetic.
The report says its main sample is 207,049 sessions on consumer plans from August 2 to September 30, 2026. Of those, 128,914 came from Chat and Cowork, which the report calls “Claude,” and 78,135 from Claude Code. That puts Claude Code at about 37.7% of the sample, by our arithmetic.
Anthropic says it used privacy-preserving methods to assign each session a scientific field and a research task. The report compares those shares with researcher time as estimated by SciNet, a benchmark it describes as based on an LLM classifier.
The report says writing and communication accounts for 32.0% of sessions on Claude, against 16.0% of researcher time. That is twice the expected share, by our arithmetic.
Data gathering goes the other way. The report says it takes 28.6% of researcher time but only 11.5% of sessions on Claude and 11.9% on Claude Code. On Claude, that is about 0.4 times the expected share, by our arithmetic.
Which Fields Lead
On Claude, the report names computer science (11.9%), engineering (8.4%) and medicine (7.7%) as the three largest fields by session share. On Claude Code, it says: “Despite making up only 4.3% of the world’s researchers, computer science alone accounts for 32.7% of Claude Code sessions.”
Anthropic built a measure of how well AI can, in theory, perform each research task. It says the top 20% of tasks on that measure get two to four times as much Claude use as researcher time would predict. Its summary of the pattern: “scientists appear to use AI where it works best.”
The Gap: Ideas and Data Gathering
Anthropic also surveyed 528 scientists, primarily in the life and health sciences, who are active adopters of AI. It says respondents rank ideation and hypothesis generation first, and data gathering second, as the tasks that matter most for important results. Those are the two categories the report finds underrepresented in Claude use.
The report says that on Claude Science, idea generation makes up about twice the share of use that it does on Claude. It says biomedical sciences and biology together make up over 50% of Claude Science sessions.
Anthropic’s Scenarios
The report models three scenarios. In the first, with current capabilities widely adopted, it says AI could take on 27–52% of the researcher time needed for a new study, with a median of 38% across fields. It puts the resulting acceleration between 1.35x in the arts and 2.02x in statistics.
The report adds: “Even the most AI-ready fields, whose work is almost entirely cognitive, reach only about a 2x acceleration.” In its hypothetical scenario where AI carries out all non-physical research work, Anthropic says output in fields such as computer science and statistics could rise as much as 15-fold, and bench sciences such as chemistry and biology three- or fourfold.
What Researchers Expect
In Anthropic’s survey, about 40% of respondents expect AI to carry out full research projects by 2030, and about 64% by 2035. The report says 16% say it will never happen.
As we read the report, the usage follows the tasks Anthropic scores as best suited to AI: data analysis and writing rank high, while data gathering, much of it physical, ranks low.
As we read the survey results, the two task categories respondents rank most important for results, ideation and data gathering, are the two the report finds underrepresented in Claude use.
As we read the scenarios, Anthropic’s own model puts current-capability gains at about 2x even for its most AI-ready fields; the 15-fold figure is a hypothetical in which AI does all non-physical research work.
Anthropic, in its 5 November report
“In short, scientists appear to use AI where it works best.”
Three Implications
FOR RESEARCH LABS The report says data gathering takes 28.6% of researcher time but 11.5% of Claude sessions.
FOR AI MODEL MAKERS The report says idea generation makes up about twice the share of use on Claude Science that it does on Claude.
FOR INVESTORS The report says about 40% of surveyed researchers expect AI to carry out full research projects by 2030.
The framework says: “Universal harnesses are likely to dominate highly digital, modular, and verifiable work.”
As we read it, Anthropic’s finding that formal sciences such as computer science, mathematics and statistics use Claude more heavily is consistent with that framing. The framework does not discuss scientific research.
What Is Not Established
The report covers consumer-plan sessions over about two months. It does not say how much research output Claude has produced, and its acceleration figures are Anthropic’s model scenarios, not measured results. The survey sample is weighted toward life and health scientists who already use AI. We have not read the report’s appendix.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Anthropic’s data shows scientists using Claude most for data analysis and writing, and less for idea generation and data gathering. Computer science takes about 7.6 times its share of researchers on Claude Code, by our arithmetic. The report’s own scenarios put current gains at about 2x for the most AI-ready fields.
A note on sourcing. All figures and quotations are from Anthropic’s 5 November 2026 report, How researchers use Claude. The 7.6 times, 37.7%, twice and 0.4 times figures are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
