Free tools Windows power users keep installed
One-click scans. No signup required.
Anthropic is winning on company-reported financial scale; Cognition is building a more focused bet on autonomous software engineering. That does not establish which coding agent is better. The available evidence contains no controlled, independent head-to-head test of Devin and Claude Code, so the fairest answer to “which AI coding company is winning?” depends on whether you mean company scale, product focus, or proven coding results.
What does “winning” mean in this comparison?
Cognition and Anthropic compete in AI software, but they are not the same kind of company. Cognition centers its business on software-engineering agents, including Devin. Anthropic sells Claude models and products across coding and broader knowledge work, with Claude Code as its coding agent. The comparison therefore has at least three separate measures: company-wide financial scale, breadth and distribution of products, and the quality of coding-agent results.
As an Amazon Associate I earn from qualifying purchases.
The evidence is strongest on the first two. Financial figures below are company-reported valuations, funding totals and run-rate estimates—not audited annual revenue or neutral industry measurements. The coding-product evidence describes positioning and reported adoption, not a controlled comparison of output quality, time saved or safety.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Which company is bigger financially?
Anthropic’s reported scale is much larger, though the announcements are from different dates and use company-specific run-rate estimates. Anthropic announced its Series H on May 28, 2026. Cognition’s latest detailed figures here are from its September 8, 2026 Series E announcement.
#1 Best Overall
| Measure | Cognition | Anthropic |
|---|---|---|
| Funding and valuation | On September 8, 2026, Cognition said it had raised more than $2 billion at a $48 billion valuation. | On May 28, 2026, Anthropic announced $65 billion in Series H funding at a $965 billion post-money valuation. |
| Company-wide run-rate revenue | Cognition said its run rate rose from $492 million in May 2026 to almost $900 million by September. These are company-reported annualized run-rate figures, not audited annual revenue. | Anthropic said run-rate revenue had crossed $47 billion earlier in May 2026. This is a company-reported estimate, not audited annual revenue. |
| Product-specific run-rate revenue | Not stated in the cited Cognition announcements. | Anthropic reported in February 2026 that Claude Code exceeded $2.5 billion in run-rate revenue, more than double its level at the start of 2026. This is a product-specific figure, not company-wide revenue. |
There is also a later, narrower Cognition claim to account for: its homepage lists a September 25, 2026 announcement titled “Cognition Crosses $1B in Annualized Revenue Run Rate.” The headline indicates Cognition subsequently claimed to pass $1 billion, but the announcement’s supporting detail is not established here. It should not be treated as a fully explained or independently verified financial figure. The September 8 and September 25 claims are company statements, not evidence of audited revenue.
Even allowing for the different dates and measurement methods, the reported gap is substantial: Anthropic’s May company-wide run rate was above $47 billion, while Cognition’s September claims were around or above $1 billion. That is evidence of a large difference in reported business scale, not proof that Anthropic’s coding agent is more capable.
Rank #2
What are the companies actually selling?
Cognition: a focused software-engineering agent
Cognition describes Devin as an autonomous software engineer that can plan, write and test code, then ship production code within a customer’s codebase. In its September 2026 announcement, the company also described work on proactive incident investigation, security-vulnerability triage and automations triggered by events. These statements show the direction and intended scope of the product; they do not independently establish how reliably Devin completes those tasks or how much supervision customers need.
Cognition says it operates as an independent agent lab and can choose and combine models, including its own. Its September Series E post put the strategy this way: “We can choose and combine the models best suited to the work, including our own.” That is a claim about its model strategy, not a comparative performance result.
Anthropic: coding within a broader Claude business
Anthropic positions Claude Code as agentic coding software, while its Claude models and products also serve broader knowledge-work uses. This gives Anthropic a wider product scope than a company centered on software-engineering agents. Anthropic said in February 2026 that Claude Code’s run-rate revenue exceeded $2.5 billion, but that product figure should not be confused with the company-wide revenue estimate announced in May.
The available descriptions do not provide a like-for-like specification of task completion, integrations, review requirements, security controls or supported workflows for Devin and Claude Code. Those details would matter in choosing a tool for a particular engineering team, but they cannot be filled in reliably from company positioning alone.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What does enterprise adoption show?
Cognition’s named deployments
Cognition named NVIDIA for chip design, GE Aerospace for aviation, Citi for financial services, Mercedes-Benz for automotive and Modal for AI infrastructure. These are company-reported deployments. The cited announcement does not quantify independently verified outcomes such as engineering hours saved, defect rates or production impact.
Anthropic’s reported customer base and distribution
Anthropic said in February 2026 that more than 500 customers spent over $1 million annualized, that eight of the Fortune 10 were Claude customers, and that enterprise use represented more than half of Claude Code revenue. These are Anthropic’s reported figures, not neutral market-share measurements or proof that those customers use Claude Code in the same way.
Best Value
Anthropic says Claude is available through AWS, Google Cloud and Microsoft Azure, and describes a diversified hardware strategy. That combination of cloud availability and a broader Claude portfolio points to wide distribution. Cognition’s stated independence and ability to select models, meanwhile, is a different strategic advantage: flexibility in building an agent stack. Neither point alone demonstrates superior coding outcomes.
Is either company quietly winning at coding?
That remains unsettled by the evidence here. Company announcements establish what each firm says it offers and reports about adoption; they do not substitute for a controlled test that runs Devin and Claude Code against the same tasks, repositories, tools and success criteria.
A useful comparison would need to disclose more than a pass rate or a polished demo. It would test representative maintenance and feature work, measure whether changes pass tests and review, record human intervention and elapsed time, and examine security failures as well as successful completions. No neutral publisher statistic here compares the companies on market share, revenue quality, productivity or coding-agent win rate.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsSo Anthropic’s financial lead is clear in company-reported figures, while a product-quality winner is not. Cognition’s focused positioning may appeal to teams seeking a software-engineering agent; Anthropic’s broader Claude offering and cloud distribution may matter to organizations standardizing across more kinds of work. A team deciding between them should evaluate its own representative tasks rather than infer coding superiority from fundraising, valuation or revenue.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
