According to reports, AI startup Anthropic is on the verge of securing a new round of substantial financing, with its valuation projected to soar to around $900 billion.

Should the deal be finalized successfully, Anthropic will eclipse OpenAI, emerging as the most valuable AI startup on a global scale. OpenAI made an official announcement in March 2026, stating that it had completed $122 billion in financing, with a post-money valuation reaching $852 billion.

This is not the first instance where Anthropic has outpaced OpenAI.

Back in April of this year, Anthropic’s official website revealed that the company’s annualized run-rate revenue had surpassed $30 billion. In contrast, as of the end of February 2026, OpenAI’s annualized revenue reportedly exceeded $25 billion.

Delving into which of these two titans is momentarily leading or overtaking the other holds minimal significance. The truly crucial aspect is that this scenario unveils at least two significant insights:

The Greatest Strength Can Also Be the Heaviest Albatross

The first counterintuitive notion is that in the AI landscape, the company boasting the largest user base is not necessarily the most valuable.

In February of this year, OpenAI officially disclosed that ChatGPT boasted over 900 million weekly active users and over 50 million consumer subscribers. Data from November 2025 indicated that over 1 million enterprises were utilizing OpenAI’s commercial products, with ChatGPT for Work seats exceeding 7 million.

900 million weekly active users is an astonishingly massive figure.

For many individuals, their initial encounter with AI was not through Claude, not through Gemini, nor through agents, but through ChatGPT.

However, the crux of the matter lies in the fact that AI is not a conventional internet product.

In the past, we were accustomed to evaluating AI companies using the same criteria as internet companies.

The company with a larger user base was deemed more valuable.

900 million users are undoubtedly an asset, but they also represent a liability. In the past, internet companies dreaded having no users; today, AI companies are starting to fear an excessive number of people using their services for free.

Reuters reported that 80% of Anthropic’s revenue stems from enterprise clients, whereas only 40% of OpenAI’s does.

Anthropic’s official website disclosed that in February 2026, the company had over 500 clients with annualized spending exceeding $1 million; by April, this figure had surpassed 1,000, doubling in less than two months.

Annualized spending does not imply that clients have already disbursed $1 million in cash; it is an annualized projection. For instance, if an enterprise client spends $100,000 on Claude this month, it would be annualized to $1.2 million.

Such high-spending enterprise clients are worth thousands of ordinary individual subscribers.

Furthermore, as of March 2026, Anthropic accounted for over 73% of companies purchasing AI tools for the first time, while OpenAI only held 26%.

OpenAI resembles an airline with 900 million passengers, but many are purchasing discounted tickets. Anthropic is akin to a business-class airline with significantly fewer passengers, but each seat commands a hefty price.

Coding: The Simplest Scenario to Validate AI Commercialization

Why can Anthropic so swiftly tap into enterprise budgets?

A pivotal answer lies in: Coding.

Programming currently stands as one of the most mature 2B scenarios for AI commercialization.

The rationale is straightforward.

Whether AI composes high-quality articles is subjective.

Whether AI crafts compelling PPTs is subjective.

Whether AI offers engaging chat companionship is subjective.

But code is unforgiving. It either functions or it doesn’t.

Anthropic’s official website disclosed in December 2025 that Claude Code’s run-rate revenue reached $1 billion six months after its public release.

This elucidates why Anthropic can sustain higher-quality revenue with a smaller user base.

Claude Code is not merely a tool; it is evolving into an integral part of enterprise production systems. And enterprises are most inclined to pay for production systems.

However, OpenAI is not lagging far behind. Recently, it is re-entering the coding arena with Codex.

This is quite a dramatic turn of events.

The inaugural AI programming model was actually developed by OpenAI itself. In 2021, OpenAI unveiled the initial Codex; GitHub Copilot also initially leveraged OpenAI Codex technology in its early stages. However, following ChatGPT’s meteoric rise, OpenAI’s focus shifted entirely to the universal chat portal.

What Claude Code capitalized on was the void left by OpenAI’s strategic pivot in focus over the past few years.

In September 2025, Codex’s usage was merely about 5% of Claude Code’s; by January 2026, it had caught up to 40%. In April 2026, Codex’s weekly active users surpassed 4 million, and its annualized revenue also exceeded $1 billion in early 2026.

The formidable aspect of OpenAI is that once it reallocates its resources, its catch-up pace is remarkably swift.

According to reports, OpenAI also envisions integrating ChatGPT, Codex, and a built-in browser into a unified desktop application capable of autonomously executing complex tasks for users.

This represents an endeavor to upgrade ChatGPT from a mere chat portal to an Agent operating system capable of orchestrating Codex, browsers, local computers, and cloud services.

AI influencer Kuai Dao Qing Yi posits that Codex’s evolution signifies a direction where AI programming tools are genuinely transitioning from being “developer-exclusive” to “accessible to the general populace.”

Therefore, Codex’s resurgence is not merely OpenAI filling the void in AI programming; it is also a pivotal leap in attempting to convert its 900 million chat users back into productive users.

END This article is an original work by "Intelligent Evolution Theory." Welcome to follow.