Why Is Unlimited Subscription AI Suddenly Asking You to 'Pay As You Go'?

Imagine going to an all-you-can-eat buffet where you pay a fixed monthly fee, and suddenly they tell you, “From now on, you have to pay extra for certain premium dishes based on how much you eat.” A similar situation has just unfolded in the artificial intelligence (AI) industry. On June 15, 2026, the prominent AI company Anthropic abruptly changed the pricing model for its core AI tools.

Why It Matters

This decision is not merely a trivial pricing update. Currently, Anthropic is pushing for a massive $1 trillion (approximately 1,300 trillion KRW) initial public offering (IPO). Yet, at the same time, the company has raised safety concerns, even calling for a temporary pause in AI development.

Simply put, they are stepping on the accelerator to attract massive investments while simultaneously pumping the brakes by warning of the risks. Right at the center of this complex and contradictory situation is this very ‘pricing change’.

The Explainer

So, what exactly changed? The core issue revolves around ‘automated AI tools’. These are AI programs like Claude Code that make judgments and operate on their own without requiring step-by-step human instruction or intervention.

Previously, a monthly subscription granted unlimited access to these powerful tools. However, Anthropic has now excluded these automated tools from the unlimited benefits, shifting to a ‘pay-as-you-go’ system. To use an analogy, a rental car that you drive yourself is still a flat rate, but a self-driving taxi that takes you to your destination on its own will charge you based on the distance traveled. Imagine having to pay an assistant exactly for the hours they worked independently while you were asleep.

Where We Stand

The primary reason for this shift is the immense computing resources that AI consumes. AI tools that sequentially process complex tasks without human intervention devour vast amounts of power and server costs—far beyond what we might imagine.

From Anthropic’s perspective, as they prepare for an IPO and need to present a healthy financial state, maintaining an uncontrollably expensive unlimited service would have been realistically impossible.

What’s Next

Going forward, it is highly likely that other major AI companies will follow in Anthropic’s footsteps. As ‘AI Agent’ technologies—which make their own judgments and take continuous actions—advance, providing them on an unlimited basis imposes an astronomical financial burden on companies. Ultimately, we are heading into a strict ‘pay-as-you-go era’ where we will pay varying costs depending on the complexity and volume of tasks performed by the AI.

AI’s Take

Anthropic’s contradictory behavior—hitting the brakes by calling for safety while stepping on the accelerator for massive capital through an IPO—reveals the current reality where the AI industry’s astronomical costs collide with its fears. This move may ultimately be a necessary growing pain ahead of the true commercialization of automated tools.

References