Anthropic's IPO could be just weeks away.

The maker of the popular Claude AI models is reportedly preparing to go public as soon as late September or early October.

And as its Wall Street debut gets closer, expectations for the company are only getting bigger.

Anthropic was valued at $380 billion back in February. By May, that figure jumped to $965 billion.

Now, some investors are even projecting a valuation upwards of $2 trillion when it finally goes public.

At that price, Anthropic would pull off the largest IPO in history and immediately become one of the world’s most valuable companies.

It all sounds so exciting. But didn't we just do this with SpaceX?

A few months ago, investors were clamoring to get into what would be the biggest IPO to date. But today, SpaceX trades double digits below its IPO price.

It's a good reminder that there's usually no reason to rush into a blockbuster IPO on day one.

Anthropic may prove to be worth the wait. Its growth has been extraordinary, with annualized revenue reportedly reaching roughly $65 billion by the end of July.

But a $2 trillion valuation comes with some enormous expectations. And there's one question in particular that we need to consider…

What If AI Becomes a Commodity?

Anthropic's lofty valuation assumes more than rapid AI adoption. It assumes Anthropic will capture a huge piece of the economic value created by that adoption.

Those are two very different things.

Chinese developers such as DeepSeek and Alibaba's Qwen are already showing how quickly AI economics can change.

Increasingly capable models are available at a fraction of the cost of today's frontier systems. Open-source models are improving too.

Meanwhile, smaller models increasingly can run locally on PCs, phones, and corporate servers.

I’m not concerned about whether Claude suddenly becomes obsolete. I don't expect that.

The question is how much customers will pay for frontier intelligence when cheaper alternatives become good enough.

Imagine a company that currently sends nearly every AI task through Claude.

Two years from now, perhaps it reserves Claude for the 20% of jobs that require the very best model, while routing everything else through cheaper open-source or local models.

AI usage could explode while Anthropic's pricing power falls.

That's a dangerous possibility when investors are discussing a $2 trillion valuation.

It reminds me a lot of Netscape.

During the early internet boom, AOL and Netscape looked like dominant gateways to the digital world. But they weren't.

Microsoft bundled Internet Explorer into Windows and crushed Netscape. Google eventually conquered browsers with Chrome.

Today, browsers are mostly free.

The internet still became far larger than almost anyone imagined. But much of the economic value moved elsewhere, to search, advertising, cloud computing, e-commerce, devices and applications.

AI could follow a similar path.

There's another warning sign: users don't appear particularly loyal to individual AI models.

Power users jump between Claude, ChatGPT, Gemini and other models depending on which works best for a particular task.

Others use several.

Millions more are happy using whichever capable model they can access for free.

Anthropic may eventually build a powerful moat around Claude. But should investors assume that outcome while potentially paying $2 trillion for the stock?

I wouldn't.

Competition isn't the only uncertainty, either.

Businesses are still grappling with questions around confidential information and ownership of AI-generated work.

Earlier this year, a federal judge ruled that documents a defendant created using Claude weren't protected by attorney-client privilege or the work-product doctrine.

AI also faces growing public resistance over everything from potential job losses to the enormous power and water demands of data centers.

None of this means Anthropic or AI will fail. It just means the future is often more difficult to price than we realize.

Let Wall Street Go First

Anthropic could be absolutely right about AI changing the world. However, investors could still be wildly wrong about what Anthropic is worth.

That's the distinction I want you to remember when this IPO arrives.

There is no prize for buying Anthropic on day one.

Let Wall Street fight over the initial shares. Let the hype settle. Let insiders sell. Let the lockups expire. Let Anthropic report several quarters as a public company.

Most importantly, let's see how its growth, margins and pricing hold up as Chinese, open-source and local models improve.

Maybe Anthropic eventually proves that Claude deserves one of the largest valuations on Earth.

If it does, there will still be time to invest.

For now, I'll do what I've recommended with other blockbuster IPOs: wait until the market has had a chance to determine what the company is actually worth.

And I'll be watching closely.

When the numbers, competitive landscape, and valuation tell me the risk/reward has changed, you’ll be the first to know.