Anthropic is on track to turn a profit before OpenAI

AI might be the most expensive playground in tech history, but its two biggest players are playing completely different financial games.

Anthropic is on pace to turn a profit ahead of ChatGPT maker OpenAI, according to internal investor documents reported by The Wall Street Journal.

Anthropic, known for its Claude model that has become a staple among software developers and corporate teams, now expects to break even in 2028.

OpenAI is looking at a much longer, bloodier road. The company projects its operating losses will balloon to roughly $74 billion, pushing its estimated profitability timeline out to 2030.

To make matters wilder, OpenAI's projected $74 billion loss does not even factor in its latest wave of computing partnerships. CEO Sam Altman previously said that financial commitments from new cloud and chip agreements total a staggering $1.4 trillion over the next eight years.

Pragmatic software vs. giant infrastructure bets

The filings lay bare two fundamentally opposed worldviews.

OpenAI is doubling down on massive capital expenditures, pouring billions into proprietary computing power, state-of-the-art chips, and sprawling data centers. It is also dishing out hefty stock-based compensation to court deep-pocketed backers.

Anthropic is deliberately keeping its feet on the ground. The startup is aiming to keep its ongoing costs roughly in line with incoming revenue.

Instead of chasing every shiny object, Anthropic has focused heavily on enterprise contracts and coding assistants. Crucially, it has steered completely clear of capital-draining products like image and video generation tools. Because of that discipline, Anthropic is projected to enjoy healthier profit margins than OpenAI over the next five years.

Why it matters

For enterprise buyers and coders, Anthropic is proving that a generative AI partner does not need to bankrupt itself to deliver world-class intelligence. If your business depends on Claude for backend logic, document processing, or code refactoring, seeing a clear path to break even by 2028 brings real peace of mind.

For the wider tech ecosystem, Wall Street's patience for runaway capital expenditure is wearing thin. Investors are loudly asking when the hundreds of billions poured into data centers will yield actual returns. Anthropic is proving that selling plain, high-utility text models to paying businesses can build a sustainable company much faster than building an all-encompassing supercomputer.

We think Anthropic's refusal to build video and image generators is a brilliantly pragmatic move. Media generation is a bottomless compute sink with notoriously messy monetization. But it also carries a long-term risk: by staying narrow, Anthropic risks ceding the next generation of multimodal computing to OpenAI. OpenAI looks to be betting the farm that immense scale will eventually create an unassailable moat, but carrying $74 billion in operating losses leaves zero room for error if macro sentiment sours.