Anthropic Posts Q2 Operating Profit as Claude Code Spending Balloons

Anthropic recorded an operating profit in the second quarter and reported 900% net revenue retention — meaning spending by the same customers grew ninefold from Q3 2025 to Q3 2026, according to financials the company shared with investors this week. On Wednesday it doubled down on that momentum with a new suite of reporting tools designed to show enterprise customers exactly what their token spend is buying.

The numbers behind Claude Code, Anthropic's AI coding tool, anchor the story. The company has focused on getting enterprise users to stick around, and the new tools are built to make that loyalty measurable. Whether they settle the ROI question or merely reframe it remains to be seen.

What do the new reporting tools actually do?

Using 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 will also offer recommendations on how to improve results. Anthropic says it will not use this data for model training by default — a default, not a ban, and customers would be wise to read the terms closely.

Anthropic is also creating pooled group budgets so teams can better manage token spending. That is a direct response to a pain point: businesses have been hunting for this kind of information as they work out how to deploy AI tools, and while usage among venture-backed firms is driving token spend, most of the Fortune 500 are not seeing major gains from AI deployment.

The takeaway: this is a clear attempt to demonstrate return on investment — and to show customers how to improve it themselves.

How big has agentic coding spending become?

Jellyfish, a company that builds tools for software developers, released a report Wednesday based on usage by 276,000 customers. The findings:

• Claude Code and other tools built on Anthropic models remain the most widely used, accounting for nearly a third of all proposed code changes submitted for review.
• Spending on those tools has grown 10x since the beginning of the year.
• Their cost now equals about 2% of total engineering cost — roughly $2 out of every $100 spent on engineering.

"Even if you're saying 2% is still small, that 10x growth is growth that most CFOs didn't plan for," Jellyfish chief scientist Nicholas Arcolano 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."

The takeaway: the spend line is still small, but its slope is what unnerves finance departments.

Has OpenAI's Astra dented Anthropic's lead?

Not so far, according to Jellyfish's data. The release of OpenAI's Astra model this summer looked like a potential threat to Anthropic, but it has not slowed Anthropic's adoption numbers. Nikolas Albarran, Jellyfish's AI product lead, told TechCrunch: "It took a long time for us to see Codex really pop in our data at all, and it has come up with Astra."

Looking at X, Albarran said, you would get the impression the competition is more heated — "Everyone in the Valley is like, 'Oh, Codex is better.'" In reality, he argued, "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 takeaway: perception on social media and measured enterprise usage are telling two different stories.

Why is switching so hard for enterprises?

It is not just a question of models and workflows. 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 major AI labs are making in forward-deployed engineering (FDE) organizations — teams that embed AI consultants inside customer companies to get them 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. "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," Albarran 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 takeaway: Anthropic is not just reporting value to customers — it is raising the cost of leaving.

The open question is whether better reporting will satisfy CFOs or simply make the ninefold spending growth harder to justify when the optimization gains run out.