Anthropic's Depth Catches Its Breadth — Extensive Claude Use Tops 30% of Firms by End of 2027

The Prediction

By December 31, 2027, the share of organizations reporting extensive use of Anthropic's Claude — as measured by IDC or a directly comparable enterprise survey methodology — will exceed 30 percent, up from roughly 19 percent reported in June 2026.

This is a depth prediction, deliberately separated from breadth. In June 2026, Anthropic crossed OpenAI in business adoption on the Ramp AI Index (34.4 percent to 32.3 percent), but the same week's IDC data showed only about 19 percent of organizations using Claude extensively. Breadth crossed over; depth did not. The claim here is that the depth line follows the breadth line within roughly eighteen months — that the shallow adoptions of 2026 convert into load-bearing, extensive use rather than stalling in pilot.

Why 60% confidence and not higher

The bull case is land-and-expand: breadth is a deposit on future depth, and a predictable fraction of shallow adoptions mature as teams find uses and switching costs accumulate. Anthropic's API-and-enterprise orientation skews its customer base toward the kind of backend integration that deepens.

The bear case is the production chasm. The Composio 2026 data showed roughly 97 percent of executives deploying agents but only about 12 percent of initiatives reaching production at scale. If most "land" dies in pilot, extensive-use share stays flat under a still-growing logo count — breadth running ahead of a depth line that refuses to follow. Anthropic's reported moves to ship products competing with application partners (Figma, Canva) could also cap the embedded, sticky usage that depth depends on.

At 60 percent confidence, this is a lean toward conversion, not a sure thing.

How This Will Be Evaluated

• Confirmed if IDC or a comparable enterprise survey reports extensive/heavy Claude usage above 30 percent of surveyed organizations at any point through Q4 2027.
• Failed if the best available comparable measure remains at or below 30 percent (i.e., depth fails to meaningfully outgrow the ~19 percent June 2026 baseline).
• If no comparable methodology is published by the target date, the prediction is graded against the closest available enterprise depth-of-use measure and noted as a partial/indirect evaluation.

Related

• Adoption Crossed Over, Depth Didn't: The Number That Will Price Two AI IPOs
• Anthropic's $965B S-1 and the Bubble Test