Anthropic’s latest IPO filing reveals that the list-price value of Claude AI used internally by the company exceeded its monthly personnel expenses in April 2026, excluding stock-based compensation. The value of that internal usage then grew more than 250 times between January and September 2026.
The distinction matters: Anthropic is not saying it spent more cash on AI than it paid employees. It calculated the value of its internal Claude usage using the prices charged to customers. The disclosure also says Claude wrote more than 90% of the contributions to Anthropic’s codebase in September, highlighting how deeply the company has integrated its own AI into its operations.
Key Takeaways
- Anthropic’s internal Claude usage, valued at customer list prices, exceeded monthly personnel expenses in April 2026.
- The comparison excludes stock-based compensation.
- The list-price value of internal Claude usage increased more than 250 times between January and September 2026.
- Claude wrote over 90% of contributions to Anthropic’s codebase in September.
- Anthropic’s revenue for July–September 2026 was 86 times its revenue for October–December 2024.
- The company says AI agents help employees handle work that would traditionally require larger teams.
- The disclosure does not mean Anthropic’s actual AI spending exceeded its payroll.
Anthropic’s Internal Claude Usage Has Increased Rapidly
The figures come from Anthropic’s amended registration statement filed with the US Securities and Exchange Commission as part of its planned public listing.
According to the filing, the company measures its internal Claude usage using the list prices it charges customers.
That estimated monthly value first exceeded monthly personnel expenses in April 2026. Anthropic also reported that the value of internal usage grew more than 250 times from January to September.
This indicates that employees across the company are increasingly relying on Claude for their daily work.
Anthropic uses its models across several functions, including research, engineering, sales and operations. Employees can also use multiple AI agents to perform tasks that might otherwise require additional people or larger teams.
However, the filing does not disclose the exact dollar value of internal Claude usage or provide a direct calculation of how much money the company saved through this activity.
Why the AI Usage Figure Is Not the Same as Actual Spending
The headline could suggest that Anthropic is spending more money on its own AI systems than on employee salaries. That is not what the filing establishes.
The company compares two different measures:
| Measure | What it represents |
|---|---|
| Internal Claude usage | Estimated value calculated using customer list prices |
| Personnel expenses | Monthly employee-related expenses, excluding stock-based compensation |
| Actual internal AI cost | Not separately disclosed in this comparison |
| Growth in internal usage value | More than 250 times between January and September 2026 |
Anthropic’s customer list prices represent what customers would be charged for comparable usage. They do not necessarily represent the actual cost of providing those services internally.
The company has not disclosed the specific compute cost associated with its internal Claude usage in this comparison.
There is another important qualification: the personnel expense figure excludes stock-based compensation, which is a significant component of Anthropic’s overall employee-related costs.
Therefore, the disclosure should be understood as a comparison of estimated AI usage value against a particular measure of personnel expenses, not evidence that AI has become more expensive than Anthropic’s entire workforce.
Claude Now Writes More Than 90% of Anthropic’s Code Contributions
One of the most striking details in the filing concerns software development.
Anthropic said that more than 90% of contributions to its codebase in September 2026 were written by Claude.
The company also reported that the average daily volume of code contributed per engineer during the three months ended September 30 was more than 45 times the average level in 2024.
This suggests that Claude is becoming an important part of Anthropic’s engineering workflow.
Instead of relying entirely on engineers to write code manually, the company increasingly uses AI to generate software contributions and support development activities.
However, the 90% figure does not mean that 90% of Anthropic’s engineering workforce has been replaced. It measures code contributions, not the number of employees or the percentage of engineering work completed without human involvement.
Human engineers remain important for reviewing code, defining requirements, testing software and making architectural decisions.
The filing does not provide a corresponding figure for code quality, defects or the amount of human rework required.
Anthropic’s Revenue Has Grown Much Faster Than Its Workforce
The company also highlighted the relationship between revenue growth and employee expansion.
According to the filing, Anthropic’s headcount increased more than fivefold between December 31, 2024, and September 30, 2026.
During the same period, its revenue for the July–September 2026 quarter was 86 times the revenue recorded in the October–December 2024 quarter.
| Business metric | Anthropic’s disclosure |
|---|---|
| Headcount growth | More than 5 times |
| Internal Claude usage value growth, January–September 2026 | More than 250 times |
| Revenue comparison, Q3 2026 vs Q4 2024 | 86 times |
| Code contributions written by Claude in September | More than 90% |
Anthropic attributes part of this operating leverage to Claude’s use across the organisation.
In simple terms, the company says AI is helping its existing employees accomplish more work without expanding its workforce at the same rate as revenue.
These figures also demonstrate why AI companies are increasingly interested in using their own products internally. Doing so can support product development while improving the productivity of their own teams.
AI Helping Build AI Raises Safety Questions
Anthropic’s filing also discusses the potential risks of using AI systems to develop more capable AI models.
The company describes a feedback loop in which one generation of Claude helps researchers and engineers build subsequent generations.
This could accelerate research and development if newer models become increasingly capable of supporting the work involved in creating future systems.
However, Anthropic acknowledges that this process could create considerable safety risks if it sharply accelerates the pace of AI progress.
The company says its Responsible Scaling Policy is designed to monitor and mitigate risks associated with recursive self-improvement.
Importantly, Anthropic does not claim that Claude can independently design and build successive generations of frontier AI models without meaningful human supervision or understanding.
The distinction is important. AI-assisted development is already taking place, but fully autonomous AI research and model development remains a different and much more demanding capability.
The Bigger Picture
Anthropic’s disclosure provides an unusual look at how an AI company uses its own technology to operate its business.
The company’s internal adoption figures suggest that AI tools are moving beyond customer-facing applications and becoming embedded in software development, research, sales and other business functions.
It also raises a broader question for the technology industry: if AI can significantly increase the output of existing employees, companies may be able to grow revenue without expanding headcount at the same pace.
However, measuring productivity is more complicated than counting AI-generated code or calculating the retail value of tokens consumed. The real economic benefit depends on quality, accuracy, operating costs and how much human oversight remains necessary.
Looking Ahead
Anthropic’s IPO filing gives investors an important indicator of how deeply Claude is integrated into the company’s operations. Future disclosures may provide more clarity on the actual cost of internal AI usage, its impact on productivity and whether the company can maintain these efficiency gains as its workforce and customer base expand.
The larger test will be whether other businesses can achieve comparable results. Anthropic has a unique advantage because it develops Claude itself, while most companies must pay external providers for AI services. Its experience may offer a glimpse of how AI-driven productivity could reshape the economics of software development and business operations.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
