OpenAI vs Anthropic: How Claude Code Flipped the AI Leaderboard
Anthropic's revenue growth and valuation now surpass OpenAI's, driven by Claude Code. Here's what went wrong for OpenAI and what it's doing to recover.
Anthropic has overtaken OpenAI in both revenue growth and valuation, with its worth now approaching $1 trillion on the back of coding tool Claude Code. OpenAI, which sparked the current AI boom with ChatGPT, misread the market by focusing on consumer chatbot subscriptions and side projects like video generation and consumer hardware while Anthropic quietly captured the developer and enterprise market. OpenAI is now scrambling to catch up with new models, leadership changes, and a revised go-to-market strategy.
What happened
| Data point | Detail |
|---|---|
| Anthropic valuation | Nearing $1 trillion |
| Anthropic revenue trend | Growth rate recently surpassed OpenAI’s |
| Anthropic IPO timeline | Planning a fall IPO, meetings with potential investors underway |
| OpenAI IPO timing | May wait until next year; filed one week after Anthropic |
| OpenAI’s new hires | Denise Dresser (former Slack CEO) as first chief revenue officer |
| OpenAI Codex public launch | Announced publicly in May; underperformed expectations |
| Anthropic reasoning model | Sonnet 3.7, released a few months after OpenAI’s fall 2024 reasoning models |
OpenAI launched a series of reasoning models in fall 2024 that could work through problems step by step before responding. Those models were built to perform well on coding challenges from high-school competitions, not the messier, open-ended work of building real software. Anthropic’s answer, Sonnet 3.7, was shaped around what a February 2025 blog post called “real-world tasks” that reflected how businesses actually use AI. Developers noticed the difference.
OpenAI gave employees early access to a new version of its enterprise coding tool, Codex, expecting strong uptake among technically sophisticated engineers. Usage came in below expectations. When the company announced Codex publicly in May, it underperformed again. Software engineers chose Claude Code instead, describing Codex as too slow and clunky.
What went wrong inside OpenAI
The root problem, according to the Wall Street Journal’s reporting, was a strategic bet that consumer ChatGPT subscriptions would drive revenue growth as AI became more embedded in daily life. While that bet was being placed, OpenAI also spread attention across a video generator, consumer hardware, and chip development.
Anthropic, smaller and more focused, built the tool that enterprise customers and developers actually wanted to pay for. Several of OpenAI’s largest investors have privately raised concerns about the company’s cash burn relative to its growth rate, and some have hedged by investing in Anthropic as well.
Internal distractions made things worse. Executives were dealing with a talent raid led by Meta CEO Mark Zuckerberg and a strained relationship with Microsoft, OpenAI’s largest investor. The departure of Fidji Simo, who had been brought in as CEO Sam Altman’s heir apparent after joining from Instacart in August, triggered a reshuffle among Altman’s top deputies.
How OpenAI is trying to recover
On the product side, OpenAI has released new models focused on coding and professional work. President Greg Brockman has been put in charge of overhauling the product lineup. A new internal team called “coding ninja” was created to ensure models are built around how customers actually use them, not just benchmark performance.
On the commercial side, OpenAI struck a deal with Amazon to distribute AI tools through the cloud giant’s customer base. It hired Dresser as its first chief revenue officer. Sales staff are offering volume discounts and other incentives to win enterprise contracts, though that risks compressing margins at a sensitive pre-IPO moment.
Pricing for the Codex enterprise product is still being worked out. Executives are weighing price cuts to gain market share while watching the effect on profit margins. Altman is also meeting with Trump administration officials and lawmakers in Washington to preview a new model as federal AI regulation debates continue.
Altman posted on X earlier this month: “We did not have our best last 12 months ever, which is mostly my fault, but we are about to have our best 12 months to date.”
Why it matters
This is a meaningful realignment, not just a PR story. The AI coding market turned out to be far more valuable than consumer chatbot subscriptions, and Anthropic identified that earlier. Businesses that have been defaulting to ChatGPT for internal tools and developer workflows now have a legitimate reason to evaluate Claude Code directly.
The IPO race is also worth watching. Anthropic filed first and is accelerating investor meetings, reportedly emphasizing its lead over OpenAI. OpenAI filed a week later but hedged on timing in its filing, noting it might be “a while” because there are “things we want to do that are likely easier as a private company.” If Anthropic lists first at close to a $1 trillion valuation, it sets a benchmark that shapes how the entire sector is priced.
For a broader view of how the big AI players are spending and what they’re earning, our coverage of Silicon Valley’s unease about the OpenAI-Anthropic duopoly adds useful context here.
Our take
The lesson is not that OpenAI built bad products. It is that focus beats breadth when markets are still forming. Anthropic made one bet on developers and enterprise, executed well on it, and now has the valuation and the growth story to match. OpenAI tried to be everything: consumer app, hardware company, chip designer, video platform. That is a hard thing to do well at any stage.
For businesses evaluating AI coding tools right now, the honest answer is to test Claude Code against Codex on your actual codebase, not synthetic benchmarks. The gap that frustrated developers earlier this year may narrow as OpenAI ships its revised lineup, but “we’re getting better” is not a reason to wait if a competitor’s tool already works better for your team today.
If you’re thinking about how to integrate AI coding tools or automated workflows into your business, the AI integration work we do for clients starts exactly there: real use cases, not demos.
What to do about it
- Run a head-to-head test of Claude Code and OpenAI Codex on a real internal project, not a toy problem.
- Review your current OpenAI enterprise contract terms, especially if volume discounts are being offered proactively. This is leverage.
- Watch Anthropic’s IPO roadshow materials when they become public. They will contain competitive positioning data that is harder to find elsewhere.
- Track how OpenAI prices Codex over the next quarter. Price cuts signal urgency and may create a better entry point for enterprise deals.
The AI coding market is now the main battleground. Pick your tools based on what your developers actually ship with, not brand recognition.
Frequently asked questions
Has Anthropic overtaken OpenAI in valuation?
According to reporting from the Wall Street Journal, Anthropic's valuation is now nearing $1 trillion and its revenue growth rate has recently surpassed OpenAI's, driven largely by the success of its Claude Code developer tool.
Why did OpenAI Codex fail to compete with Claude Code?
Developers found Codex too slow and difficult to use after its public launch in May. OpenAI's reasoning models were trained on high-school competition coding tasks rather than the open-ended, real-world software development work that enterprise customers actually need.
When will OpenAI go public?
OpenAI may delay its IPO until next year, according to people familiar with the plans. The company filed to go public one week after Anthropic but hedged on timing in its filing, saying it might be 'a while' because some planned activities are easier to execute as a private company.
Who is Denise Dresser at OpenAI?
Denise Dresser is the former CEO of Slack. OpenAI hired her to become the company's first chief revenue officer as part of its push to win enterprise customers.