Anthropic spend reports give enterprise customers a team-by-team view of what their Claude budgets are buying, and the Wednesday launch lands as those budgets grow faster than finance departments planned. Timing matters here. Managers can pick a date range and see what a team produced, what it cost, and where its workflows broke down, and each report adds recommendations for fixing the weak spots.
Internal analytics and employee surveys feed the data, according to TechCrunch. Pooled group budgets arrive alongside the reports. Teams can manage token spending as a group, and Anthropic says it will not train models on the report data by default.
Claude Tools Now Sit Behind Nearly a Third of Proposed Code Changes
Jellyfish, which surveyed 276,000 of its customers, found that Claude Code and other Anthropic-based tools account for nearly a third of proposed code changes submitted for review. Spending on those tools grew 10x since January. Their bill now equals about 2% of total engineering cost.
A proposed code change measures activity, not value. Pull requests submitted for review say nothing about whether the code merged, shipped, or broke something later, so a report that pairs cost with workflow breakdowns comes closer to what finance teams need than a raw count of output.
Nicholas Arcolano, chief scientist at Jellyfish, said "that 10x growth is growth that most CFOs didn't plan for." Cheaper tokens have not shrunk the bill. Anthropic launched Claude Opus 5.5 with 40% lower costs, yet spending on Anthropic-based tools kept climbing, which suggests lower prices are pulling in heavier use rather than smaller totals.
For engineering leads, Claude token costs now behave less like a tooling fee and more like cloud compute, scaling with adoption instead of headcount. Reports tied to individual teams put those numbers in front of the managers who control the budget. Workflow gaps become visible too.
Anthropic Needs Proof of Value Before Its IPO
Anthropic recorded an operating profit in the second quarter and reports net revenue retention of 900%, TechCrunch reported, meaning the same customers now spend about nine times what they did a year earlier. Investors weighing Anthropic's IPO filing will ask whether that spending holds. CFOs ask the same question now.
Large-company adoption is the harder sell. Most Fortune 500 firms are not yet seeing major gains from AI, per TechCrunch, while venture-backed startups drive much of the token spend. Anthropic spend reports aim at buyers who need a documented return before they approve more enterprise AI spending.
Reports Could Raise the Cost of Leaving Claude
Reporting can double as retention. Nikolas Albarran, AI product lead at Jellyfish, described the tools as a lighter version of forward-deployed engineering, the hands-on consulting model AI labs are investing in. Once a customer embeds the tools, he said, "it's going to be harder and harder to switch."
Anthropic has also bet $100 million on engineers who can put AI to work, so the reports appear to fit a wider push to embed Claude inside customer teams. A tool that records each team's output, cost, and broken workflows also builds a history that any rival would have to recreate. Competitors start with a blank page.
Rivals are not locked out. OpenAI shipped its Astra model over the summer, and Jellyfish says it has not yet slowed Anthropic's adoption, though Codex has gained some ground. Anthropic's enterprise relationships and standardized workflows still give it an edge, Albarran said, and he noted that switching costs are not zero.
Employee Surveys Feed the Reports, and the Vendor Scores Its Own Work
Employee surveys feed the reports, so developers' own answers about where Claude helps and where it stalls become input to a budget decision. Teams that rate the tools poorly could see recommendations to change their workflows, and teams that rate them well could make the case for larger commitments. Survey answers now carry budget weight.
Buyers will probably want to check the figures against outside measurement, because the vendor is also the one scoring its own product. That is an awkward referee. Anthropic spend reports hand buyers a written case for staying, and Anthropic's IPO disclosures will show whether customer spending keeps growing at this pace.
