Anthropic and OpenAI at the Edge: The Race for Profit Has Become a Matter of Survival

The world's largest AI companies face a choice: become profitable or burn through investor money along with it. Commentators call what's happening a "monetization cliff" — the moment when years of losses must finally give way to real business. Anthropic and OpenAI are built on a foundation of hundreds of billions of dollars in venture capital.

In parallel, multi-billion dollar programs have been launched to build data centers, purchase chips, and develop infrastructure. The logic is simple: at some point profit must appear — otherwise the bubble bursts. Most CEOs in the industry don't hide it: some players will suffer a spectacular failure, some will survive, but the money put into the game is too large for anyone to walk away voluntarily.

AI agents became the turning point. Products like Claude Code and Cowork from Anthropic, as well as the open-source OpenClaw and Codex from OpenAI changed the companies' understanding of how many resources are needed to operate. Agents are valuable for users right now — they perform tasks autonomously, write code, manage workflows.

But this value comes at a price: agents burn through tokens several times faster than regular chatbot requests. Companies weren't prepared for this pace of consumption. It's already leading to harsh decisions.

A month ago, OpenAI unexpectedly shut down Sora, its video generation application — despite an active licensing deal with Disney worth $1 billion. The reason turned out to be mundane: the service required too much computing power, and it's needed for Codex. Product profitability turned out to be more important than a high-profile partnership.

Anthropic acted similarly, but targeted a different audience. The company announced that users actively using the OpenClaw framework through a standard subscription can no longer do so under the previous terms. They're being moved to a usage-based payment model — significantly more expensive.

Essentially, the company introduced price discrimination for the most resource-intensive customers. Both moves are symptoms of the same disease. Companies are forced to manage a deficit of computing power at a time when demand is growing faster than infrastructure.

The agent use case turned out to be economically far more expensive than assumed during planning. On the horizon are two of the largest IPOs in tech industry history. Anthropic and OpenAI are preparing to go public, and pressure on profitability metrics has never been higher.

This week, financial projections from both companies became known: hundreds of billions of dollars in revenue and a path to profit by the end of the decade. But these very numbers raise the main question: are they realistic? The answer depends on several variables.

Will companies be able to maintain current growth rates? Will enough corporate clients be found willing to pay for agents at the needed scale? Won't competition from open models destroy the price premium on which these forecasts are built?

What's happening is not just management decisions by two startups. It's the moment when the AI industry stops being a story about technology and becomes a story about money. Every product closure, every tariff tightening — a signal to the market: companies can no longer afford unprofitable growth.

Investors are waiting for returns, the market is waiting for profits, and the entire ecosystem of developers is watching to see who will stand and who will fall off the cliff.