Anthropic reports 900% net revenue retention and ships AI spend reporting tools

Anthropic recorded an operating profit in the second quarter and reported 900% net revenue retention to investors this week, according to financials that show its AI coding tool, Claude Code, spending from the same customers in Q3 2025 grew nine times by Q3 2026. The company paired those numbers with a product announcement: on Wednesday it launched a suite of reporting tools designed to give enterprise customers a granular view of what their token spend actually buys.

The stakes are straightforward. Large customers want demonstrable return on investment from AI deployment, and most of the Fortune 500 are not yet seeing major gains, even as usage among venture-backed firms drives token spend upward. Anthropic's answer is to hand managers the measurement layer themselves.

What do the new reporting tools do?

The new suite draws on internal analytics and employee surveys. Managers can view a team's work over a given date range, what the team produced, how much it cost, and where its workflows broke down. The reports also offer recommendations on how to improve results.

Anthropic says it will not use this data for model training by default. The company is also creating pooled group budgets, allowing teams to better manage token spending across an organization.

The move is a clear attempt to demonstrate the return on investment of Anthropic's tools, and to show customers how to improve that return — a major concern among large customers who are still working out how best to deploy AI across their businesses.

How big is the spend, exactly?

Jellyfish, a company that builds tools for software developers, released a report yesterday based on the use of agentic coding tools by 276,000 of its customers. Claude Code and other tools built on Anthropic models remain the most widely used in the survey, responsible for nearly a third of all proposed code changes submitted for review.

The spending curve is steeper than the adoption curve. Spending on Anthropic-based tools has grown 10x since the beginning of the year, and their cost now equals about 2% of total engineering cost.

"Even if you're saying 2% is still small, that 10x growth is growth that most CFOs didn't plan for," Nicholas Arcolano, the chief scientist at Jellyfish, told TechCrunch. "We're certainly getting to the point where people are realizing, okay, it's not enough to just spend a bunch. You now have to optimize it, make it work."

Why does this story matter for the enterprise AI market?

Anthropic's new reports speak directly to that optimization challenge. The company leads among enterprise customers, but what makes an agentic coding tool truly sticky is difficult to identify.

The release of OpenAI's Astra model this summer looked like a potential threat to Anthropic's position. So far, it has not slowed Anthropic's adoption numbers, according to Jellyfish's data.

"It took a long time for us to see Codex really pop in our data at all, and it has come up with Astra," Nikolas Albarran, the AI product lead at Jellyfish, told TechCrunch. He noted that reading X, the social platform, you would get the distinct impression that the competition is more heated than the data shows. "Everyone in the in the Valley is like, 'Oh, Codex is better.'"

Why switching costs favor incumbents

In reality, Albarran said, "Anthropic was just there with great enterprise motions, and they had an incumbent weight of their own — this stuff just worked and worked well, and a lot of people just put a lot of effort into standardizing workflows around it. And there's not a big enough reason to switch."

The stickiness question goes beyond models and workflows. The realities of business mean that once a frontier lab has embedded itself in a customer, switching is harder than it is for an individual programmer or a small organization.

Albarran points to the investment all the major AI labs are making in forward-deployed engineering (FDE) organizations — teams that embed AI consultants inside customer organizations to help them get off the ground. Anthropic's new reporting tools can be read as a kind of FDE-lite, a lighter version of that hands-on help, delivered as software.

"We're at a point now where it's mature enough, where once you're in, it's going to be harder and harder to switch because you've spent the effort," he said. "Even if they're all mostly the same, they're all going to be slightly different, have their own slight quirks of how they handle agents to ND files, etc., and so that cost to switch is not zero."

The economics behind the announcement

The context for Wednesday's launch sits in the numbers Anthropic shared with investors. A second-quarter operating profit marks a financial milestone for the frontier lab. The 900% net revenue retention figure — spending from the same customers in Q3 2025 growing nine times by Q3 2026 — indicates that existing enterprise customers are expanding their commitments rather than churning, the metric software investors watch most closely.

The new reporting suite addresses the next phase of that expansion. As Jellyfish's data shows, agentic coding spend has grown 10x since the beginning of the year and now represents about 2% of total engineering cost. A 10x cost curve that CFOs did not budget for creates demand for exactly the kind of visibility Anthropic is now selling as a feature: what a team produced, what it cost, where the workflow failed, and what to change next quarter.

Anthropic's bet, laid out in Wednesday's release, is that the vendor which proves the return on token spend keeps the account — and at 900% net revenue retention, the company is currently proving it.