The quality earnings put.

The trade I own

I own a March 2027 $660/$850 call spread in Meta. The long call is deep in the money at 0.70 to 0.80 delta so it behaves like the stock. The short call sits at my first target of $850. As each target approaches and the thesis remains intact, I roll it up to the next leg.

Meta is the quality leg of the AI book. Every other AI name I own is a bet on the buildout continuing. Meta is the one where the business underneath could pay for the whole thing by itself: 81% gross margins, $130 billion of trailing operating cash flow, and an ad franchise doing $60 billion quarters at 28% growth.

Four letters: Muse

Muse launched on September 8. Twenty two days later it had crossed 5 million US downloads, per Sensor Tower. ChatGPT took 56 days to get there. Grok took 103. Claude took 492. Only 23 apps in history have hit 5 million downloads in their first 22 days, and 85% of them are mobile games. Muse now counts 2.7 million daily active users, per TickerTrends, from just 3.5% of the WhatsApp world. Apply that penetration globally and the numbers are far larger. The headroom is massive.

So what is it worth? OpenAI is raising near a $1.4 trillion valuation on roughly a $70 billion run rate. Anthropic is targeting $2 trillion on over $100 billion. That is 20 times revenue. Meta trades at $1.85 trillion, about 8 times revenue, priced as an advertising business. At a 20x AI multiple, Muse at ChatGPT parity adds $240 billion of value, 13% upside before any re-rating of the core business. Apply ChatGPT's 4.2% conversion to the full WhatsApp base and it is $670 billion, 36%. At a 6% bull case, $950 billion, 51%.

Citi agrees with the direction. The bank models Muse at $27 billion in revenue by 2030, with about $23 billion from transactions and $4.5 billion from subscriptions. On a two-year horizon that implies roughly $12-15 billion in revenue, or $240-300 billion at a 20x AI multiple, right on the parity case above.

Self-funded, not debt-funded

The misunderstanding is the cash flow statement. Last quarter Meta printed $784 million of free cash flow, down from over $8 billion a year earlier, and guided 2026 capex to $130 to $145 billion. The stock fell 9% after hours. The market read it as the AI build eating the company. Read the whole statement instead. Trailing twelve months: $130 billion of operating cash flow, $92 billion of capex, $38.5 billion of free cash flow left over. The capex is lumpy, $31 billion landed in a single quarter against $31.9 billion of operating cash, so one quarter's FCF rounds to a rounding error. That is timing, not the business breaking. No equity raised. No debt binge. Net cash on the balance sheet.

This is what separates Meta from every other AI capex story. The hyperscalers are funding their buildouts with bonds and off balance sheet vehicles. Meta funds its buildout the old fashioned way: the Family of Apps prints $23.4 billion of quarterly operating income at a 39% margin, and the check for the data centers clears against it. Eighty one cents of every revenue dollar survives cost of revenue. After Reality Labs and a $2.4 billion legal charge, the consolidated operating margin was still 31%. The earnings put is the ad business.

The structure

The long call gives me stock like exposure at a fraction of the capital. The short call is where the thesis says the stock should go. Its theta should cover the long call's theta. The spread defines the loss before I enter and forces the price target into the trade.

The honest drag: Reality Labs

The division did $431 million of revenue last quarter against a $4.6 billion operating loss. Cumulative losses since 2020 are past $85 billion. That is real money. Two things keep me in. First, the fire is contained: first half 2026 losses were $8.6 billion versus $8.7 billion a year earlier, and the division's costs grew 3% in a quarter when the company's total costs grew 55%. Second, the glasses are actually selling: 7.4 million AI glasses shipped last year, 85% global share. It is the only consumer AI hardware with real adoption. I am not paying for Reality Labs. I am tolerating it inside an ad business that covers the loss many times over.

Four legs, four targets

My first price target is $850 in six months. The second is $1,150 in twelve months. The third is $1,400 in eighteen months. The fourth is $1,700 in twenty four months. The first leg is my live spread. The later legs are modeled and illustrative, showing where the structure goes if the thesis plays out and I keep rolling the short call up.

The structure is designed to move with the thesis. As the stock approaches $850 and the thesis remains intact, I roll the short call up to $1,150. As it approaches $1,150 with the thesis intact, I roll again toward $1,400, then $1,700. The target path is set before emotion enters the trade.

What could go wrong

• 01The 2027 capex number is the whole fight. If Zuckerberg guides another step up with no revenue attached, the market will not wait for the timing argument.
• 02Regulatory risk is not theoretical. $2.4 billion of legal charges hit this quarter's EPS.
• 03Advertising is cyclical. In a real downturn the 28% growth goes away and the multiple follows.
• 04Muse monetization is early. The market may keep penalizing the capex until Muse revenue shows up in the numbers.
• 05WhatsApp rollout is the clock. Today Muse is in 3.5% of the WhatsApp world. Delays in India, Brazil and Indonesia push the revenue out.

Flavio