Hey Traders,

In August 1995, Netscape went public and its stock more than doubled on the first day (I remember the moment well!).

Netscape’s Navigator browser was how most people saw the internet for the first time, and it held most of the market.

Then Microsoft put Internet Explorer on every Windows computer and gave it away, because Windows paid the bills.

Netscape was still charging businesses for a license. In 1998, Netscape agreed to sell itself to AOL, and by 2001, Internet Explorer had more than 90% of the market.

The first mover lost as the company that already had the users won.

Meta Platforms (META) and its new AI agent, Muse, look a lot like that story, with one twist: Muse may be the Netscape moment for AI agents. But Meta isn't Netscape here; It's Microsoft.

Part one of this series showed that the smart part of AI is getting cheap, so the value moves to whoever owns the habit. Microsoft owned the desktop. Meta owns the scroll. About 3.6 billion people use Meta's apps every day, and Muse already lives inside WhatsApp. Ten days after launch it was the No. 1 free iPhone app in America, ahead of ChatGPT, and it's only available in the U.S. and Canada so far. Most of those 3.6 billion people can't get it yet.

Microsoft could give the browser away because Windows paid the bills. Meta can practically give the smart part away because ads pay the bills: $60.8 billion in revenue last quarter, almost all of it from advertising.

Muse is free for up to 100 million tokens a week. Developers can use Meta's Muse Spark model for $4.25 per million output tokens, against $20 for Anthropic's Opus 5.5, and Meta cuts that price about 95 percent, to 20 cents, for customers who let it learn from their prompts. Zuckerberg has said the pricing from "some of the other labs is very extreme and has very high margins."

It's the oldest play in Facebook's book. Users never paid for Facebook with money. They paid with their attention. Now the smart part is nearly free, and they pay with their habits.

On launch day Zuckerberg said, "we think that this thing is actually going to make you money and save you money." Meta says roughly one in three early users connected Muse to a business account. At the end of September it launched Muse for Small Business, which plugs into QuickBooks, Shopify, Stripe and Slack, and Zuckerberg announced Meta Enterprise Platform, which he called "the next major pillar" of the business.

Muse helps a small business find customers, the business spends more on Meta ads to find more of them, and when Muse buys something for a user, Meta wants a cut. At Connect, Zuckerberg said, "over time we will profit by taking a small fee from transactions." Meta gets paid on the way in and on the way out.

The risks are real. Amazon.com (AMZN) blocked Muse from shopping on its site, saying the agent didn't identify itself and appeared to store customers' logins, which Meta denies.

Expect more fights like that as every walled garden decides which bots get through its door.

Trust may matter more than the market thinks. The bill is big, too. Meta plans $130 to $145 billion of capital spending this year, free cash flow shrank to $784 million last quarter, and analysts now see it turning negative this year. Meta's last big enterprise product, Workplace, never lived up to its early promise, and Meta pulled the plug on it.

Still, the lesson from 1995 holds. Netscape got there first and lost. Meta doesn't need the smartest model. It needs to make life easier for 3.6 billion people before anyone else does. Capture them early, and they have little reason to leave.

Part three covers positioning: who wins when everyone has an agent, who gets squeezed, and how to own Meta with defined risk while it's still spending heavily.

Here for a good time… AND a long time,

Hans