#Gate直通IPO第三期Anthropic Let’s discuss Claude’s real-world experience and Anthropic’s prospects from the perspectives of actual use and industry logic—no hype.

🛠️ Claude in real-world use: like a reliable senior colleague

Whether they’re programmers or work in corporate content, most people around me now use Claude as their go-to tool. Its biggest strengths come down to a few things:

• Long texts and code are its clear strengths
You can drop in a few thousand lines of project code, and it can pinpoint bugs and refactor components accurately without losing track of details in complex projects like React and Python.
Give it a chaotic 50k-character technical white paper, and it can distill the core architecture while flagging contradictory technical specifications—its big-picture grasp is a cut above other models.

• No fluff, no making things up—it treats you like an adult.
Unlike other AIs, it won’t start by flattering you. If it finds a flaw in your proposal, it will point it out directly with data. If it’s unsure, it will simply say, “I’m not sure.” Its rate of making things up or cutting corners has been driven almost to 0.
With the same $200 Pro subscription, no one around me has hit a usage limit or had their work interrupted. Its peak-time stability is far better than competitors’.

• Its weaknesses are also obvious
Its conversational style is rather cold. If you’re looking for emotional support, it’ll give you an analysis of the physiological mechanisms instead—it won’t hand you a tissue.
It can occasionally be overly cautious, and in legal reasoning it may end up “defending its previous judgment,” making it insufficiently flexible.

🚀 Anthropic’s prospects: It has picked the ultimate market for AI to replace professional services

It’s not really competing with OpenAI for consumer traffic; it’s going straight after a huge slice of the traditional professional services market. Its growth logic is exceptionally solid:

• Its enterprise subscription base is already up and running
At the end of October, annualized revenue surged to $81 billion, up another $10 billion from September. Enterprise customer revenue retention is as high as 900%—meaning that once a company adopts it, core business processes across the whole company migrate to the platform, and usage grows exponentially.

• Goldman Sachs’ $685 billion revenue forecast isn’t pie in the sky.
By 2030, enterprise revenue will reach $586 billion, spanning software development, legal documents, accounting compliance, and drug discovery—fields that have traditionally relied on large amounts of professional labor. Replacing labor costs in each niche industry represents a market opportunity worth tens or hundreds of billions.
That scale is already approaching the current revenue levels of Amazon and Nvidia. This isn’t about creating a new market from scratch; it’s about taking a direct slice of the trillion-dollar traditional professional services market.

• The only real hurdle: sustaining years of heavy spending
It has already committed to spending $900 billion on compute procurement over the next few years. Free cash flow outflow will reach $65.7 billion in 2027, and it won’t achieve positive free cash flow until 2030.
But it has seven major compute players—including Google, Nvidia, and Microsoft—as cornerstone shareholders. Capital and compute are fully integrated, so there’s no need to worry about its funding drying up midstream.

All in all, Claude is now “the ultimate productivity tool for professionals,” and Anthropic is the AI company with the best chance of breaking into the world’s top revenue tier through enterprise services.