Anthropic Poised for First Profitable Quarter as AI Costs Surprise Companies
In brief
- Anthropic is strongly rumored to be on the brink of its first profitable quarter, a significant milestone in the AI industry.
- This shift comes amid growing expenses for companies using large language models (LLMs) like Claude Code and OpenAI's Codex.
- Stories are circulating about businesses being surprised by hefty bills from employee usage, highlighting how quickly LLM costs can add up.
- Enterprise customers are now paying API prices without the discounts previously assumed by many users.
- For instance, a moderately heavy user of these tools would spend $2,180 in 30 days using tokens, but under subscription plans like Anthropic's Max or OpenAI's Pro, this same usage costs just $200-a remarkable deal.
- However, companies relying on extensive agent use are now facing API pricing without the discounts they expected.
- This shift marks a new phase for AI adoption, where enterprises are paying closer attention to costs.
- As Anthropic and OpenAI refine their pricing models, developers and researchers should watch for further adjustments in API fees and enterprise plans.
- The future of AI economics is likely to focus more on sustainable, cost-effective usage strategies.
Terms in this brief
- Claude Code
- A specific version or application of Anthropic's Claude AI model tailored for coding purposes. It helps developers by generating code snippets and assisting in software development tasks.
- Codex
- An AI tool developed by OpenAI designed to understand and generate human-like code, similar to how other models handle text. It's used for coding assistance and automating programming tasks.
Read full story at Simon Willison →
More briefs
Anthropic Poised for Profit as AI Costs Shift
Anthropic is reportedly set to record its first profitable quarter, with companies noticing higher expenses from staff using large language models (LLMs). Users on the $100/month Max plan for Anthropic and $100/month Pro plan for OpenAI are seeing significant cost increases. For instance, one user found they would have spent $2,180.16 in API tokens over 30 days, up from a previous estimate of $200. Both companies have recently adjusted pricing, with Anthropic switching its Enterprise plan to $20/seat/month plus API fees and OpenAI aligning Codex pricing with API usage. These changes aim to lock enterprise customers into current higher prices. The shifts reflect the models' product-market fit and likely aim to prepare for IPOs, as both companies seek to maximize profitability ahead of potential public listings.
Investor Sarah Guo's Early AI Bets Pay Off
Sarah Guo invested in AI startups before they were popular. She backed companies like Baseten and Harvey. These companies are now worth billions of dollars. Guo's investments have grown more than tenfold in value. Her portfolio is worth $62 billion. She invested in companies that use AI for law and medicine. Guo will continue to invest in AI startups in the future.
AI Data Centers Raise Electricity Bills
AI data centers are increasing electricity bills for households. They use a lot of energy to run. The cost of building AI infrastructure is high. It may lower costs later. But for now, it is raising electricity bills. In one area, the annual capacity charge was $14.7 billion in 2025. This is up from $2.2 billion two years prior. Households are paying more for electricity. Some communities are fighting back. They are trying to stop new data centers from being built. Electricity bills will keep rising until this issue is solved.
Goldman Sachs Raises S&P 500 Target
Goldman Sachs strategists have raised their target for the S&P 500 Index to 8,000 points. The US benchmark has already jumped almost 10% this year. The increase is due to strong earnings growth, driven by the AI boom. The strategists forecast a 17% return for the S&P 500 Index this year. They also increased their earnings-per-share forecast for companies in the S&P 500 to $340 for 2026. This growth will drive further gains in stocks. The S&P 500 has hit its latest record high, closing at 7,519 points. The strategists expect beneficiaries of artificial intelligence investment to account for half of S&P 500 earnings growth this year. Next year may bring more stock market gains.
Meta Cuts 1400 Jobs in Washington State
Meta will cut nearly 1400 jobs in Washington state. The company will start terminating employees in Seattle, Bellevue, Redmond, and remote positions on July 22. The layoffs affect many positions, including software engineers and data scientists. This is part of Meta's plan to shift thousands of workers into AI-focused roles. The company is spending billions on artificial intelligence infrastructure. The cuts will start soon and affect many workers in the state, with Bellevue taking the largest hit at 699 workers. Meta will continue to change its workforce to focus on AI.