The reason META looks cheap is simple: the market believes Mark Zuckerberg is asking shareholders to fund a moonshot using a money printer, and Wall Street has decided to focus on the moonshot.
At roughly $1.52 trillion of market value, META sits on top of a Family of Apps franchise that generated $55.9 billion of Q1 2026 revenue and $26.9 billion of operating income in that segment alone. That is not a broken business. That is a toll bridge with very good traffic.
The valuation gap exists because the market is not doubting the ad engine. It is doubting the capital allocation.
Start with the obvious offense. Meta told investors it expects to spend approximately $125 billion to $145 billion on capital expenditures in 2026. In the first quarter alone, it spent $19.0 billion on property and equipment purchases, largely for servers, data centers, and network infrastructure. On top of that, the company disclosed $237.7 billion of non-cancelable contractual commitments as of March 31, 2026, mostly tied to cloud capacity, servers, data centers, and Reality Labs hardware, and then added that another roughly $24 billion of infrastructure commitments were signed in April 2026.
That is the sort of number that makes even growth investors sit up straighter. It also explains why a company that still gushes cash can look optically cheap. When management signals that tomorrow’s depreciation bill will be wearing steel-toed boots, the market marks down today’s earnings.
Then there is Reality Labs, the division Wall Street treats like an expensive hobby with line-item disclosure. Meta’s 2025 annual report says Reality Labs reduced overall operating profit by approximately $19.19 billion in 2025, and management expects 2026 Reality Labs operating losses to remain similar to 2025. Worse, the company explicitly says Reality Labs will continue to operate at a loss for the foreseeable future.
That phrase—“for the foreseeable future”—is not exactly catnip for multiple expansion.
So why argue the stock is still below intrinsic value? Because the market may be capitalizing the pain and undercapitalizing the earning power.
Meta’s core business remains absurdly profitable. For full-year 2025, the company reported $200.97 billion of revenue and $83.28 billion of operating income, a 41% operating margin. Family daily active people averaged $3.58 billion in December 2025. In other words, the core machine is still doing what great franchises do: it compounds while everyone argues about the science project in the basement.
Even Q1 2026, with all the spending anxiety, was not exactly a funeral march. Meta generated $32.2 billion of operating cash flow in the quarter and ended March with continued access to enormous liquidity, including ongoing cash generation and a stated belief that available funds and operating cash flow are sufficient to fund operations, AI investments, and capital returns for the foreseeable future. It also still paid a dividend of $1.35 billion in Q1, even while pausing buybacks during the quarter, with $25.03 billion of repurchase authorization remaining.
That matters. A stock can be cheap because the business is deteriorating, or cheap because investors dislike what management is doing with the cash. META looks more like the second category.
The bear case, to be fair, is not dumb. There are at least four reasons the market is skeptical.
First, capex has stopped being a line item and become a worldview. Spending up to $145 billion in 2026 forces investors to ask whether AI economics are being proven or merely asserted.
Second, these commitments are sticky. Meta had $131.05 billion of non-cancelable contractual commitments at year-end 2025, which rose to $237.67 billion by March 31, 2026. Once you build the empire, you have to feed it.
Third, management has not earned unlimited benefit of the doubt on long-duration bets. Reality Labs consumed about $21.40 billion of total investments in 2025, and the company says many of those products may only be fully realized in the next decade. Public markets famously hate waiting.
Fourth, Wall Street remembers that Meta’s stock is not priced on GAAP alone. It is priced on trust that AI infrastructure, open-source models, wearables, messaging monetization, and ad tools will eventually throw off returns above the cost of capital. Management may be right. But “eventually” is not a valuation multiple.
My read: the market is punishing META because investors are treating AI and Reality Labs spending as if it were permanently value-destructive. That may be too harsh. A company producing $83.28 billion of 2025 operating income and nearly $26.8 billion of Q1 2026 net income should not need heroic assumptions to justify more than a market-level multiple—unless you think management will keep shoveling cash into projects that never clear the hurdle.
That is really the whole debate. Not whether Meta has a great business. It does. The debate is whether Zuckerberg is investing like an owner with a long runway or spending like a man who found the company credit card and also the future.
What to watch: over the next few quarters, will Meta show clear incremental revenue, margin, or efficiency benefits from its extraordinary AI infrastructure buildout—or will investors keep seeing a wonderful ad business forced to subsidize an ever-larger experiment?