Chapter 8

What to Watch

The evidence assembled so far describes a well-defended, fairly-priced advertising franchise run by an owner spending heavily on a return no one can yet prove. At $669 the price implies only modest long-run earnings growth — provided the AI build converts to cash. Whether it does is measurable, quarter by quarter. This chapter reconciles the case into shared facts, lays out three scenarios, and names the specific line items that would confirm or break the read.

The four tensions

Every chapter so far lands on the same structure: a fact that both a buyer and a skeptic can point to, read in opposite directions. Reconciling the case is not a matter of picking a side but of naming, for each fact, the evidence that would settle it. The four that carry the most weight:

Sources: FY2025 Annual Report, segment results and cash flows [1] [2]; 2026 guidance from the Q4 2025 earnings call [3]; founder voting power from the 2026 proxy [4].

The rows are not independent. Each of the first three rests, in the end, on the fourth: the margin, the multiple, and the tolerance for founder control all resolve according to whether the AI infrastructure build earns its keep. The reconciliation, in other words, has one measurable fulcrum — capital intensity — and the rest of this chapter treats it as such.

What the price implies

The multiple stack laid out in Valuation leaves an expectations question a value buyer will ask directly: how much growth does $669 already assume. A reverse discounted-cash-flow answers it as arithmetic. Take consensus 2026 net income of roughly $84B, discount future earnings at a 9% cost of equity, and assume the company is worth 15 times its earnings a decade out. Solving for the earnings growth rate that reproduces today's ~$1.72 trillion market value gives about 6% a year; at an 18x exit multiple, under 5%.

Source: derived from consensus estimates (2026E EPS $32.84, revenue $252.9B) and a reverse DCF at a 9% discount rate; price $669.21 as of 2026-07-11.

That is a modest bar. Consensus has revenue growing about 26% in 2026 and 20% in 2027, and even after stripping out the tax-charge reversal, normalized earnings grow around 11% near-term — all comfortably above the ~5% a decade the price embeds. On an earnings basis, the stock is not asking the buyer to underwrite heroic growth.

The catch is in the phrase "on an earnings basis." The reverse-DCF treats reported earnings as if they were distributable cash. They are not, while capex runs at $69.7B and climbs toward $125B: free cash flow was $46.1B in 2025, less than 40% of operating cash flow, and the wedge is the build [5]. The benign expectations read holds only if free cash flow re-converges with earnings once capital intensity normalizes. If it stays near the guided ~50% of revenue, the earnings the price is capitalizing never fully arrive as cash, and 20x earnings is a richer price than it looks. The expectations check and the capex question are the same question.

Where earnings and cash diverge

The divergence is visible in three years of results. Operating income has compounded — $46.8B in 2023, $69.4B in 2024, $83.3B in 2025 — while free cash flow rose and then fell back, ending 2025 roughly where it began the period despite far higher profits [6] [7].

Source: FY2025 Annual Report, segment results and cash flows [8] [9]. FCF = operating cash flow less purchases of property and equipment.

Management's own framing is the bull's strongest card: it guides 2026 operating income above 2025's $83.3B despite the step-up in spending, which implies the income statement absorbs the depreciation wave even as free cash flow stays suppressed [10]. The bear's answer is that depreciation and amortization ($18.6B in 2025) still lags capex by a wide margin, so the heaviest non-cash charges from the 2025–2026 build have not yet reached the income statement [11]. Both can be true for a while; which one governs by 2027 is a matter of record, not opinion.

Three scenarios

The scenarios below are not forecasts. They are the three ways the fulcrum can resolve, each tied to a capital-intensity path and the earnings and cash outcomes that follow. The point is to make the paths distinguishable in advance, so the reader can tell early which one is unfolding.

Sources: derived from FY2025 results and 2026 guidance [12] [13]; peer-intensity range from Hyperscaler Capex.

The evidence does not yet separate these three. Meta has never run capex near 50% of revenue, so there is no company history that says whether 2026 is a peak or a new level, and the depreciation from the build lands mostly after the current disclosures. What the evidence does give is an early-warning system: each scenario shows up first in a small number of line items, well before it shows up in the share price.

Watch items

These are the checkable signals, each anchored to a line in a filing that appears every quarter, with the current reading and the threshold that would move the read. None requires a forecast — only attention to the next few statements.

Sources: FY2025 Annual Report — segment results [14], cash flows [15], capex guidance [16], and commitments [17].

Two of these deserve emphasis because they are the earliest and the hardest to spin. Capex as a share of revenue is disclosed alongside guidance and cannot be reframed: it is either turning down toward the peer range or it is not [18]. And the gap between operating and free cash flow is the build made visible in one subtraction each quarter [19]. A buyer who tracks those two lines will know which scenario is unfolding a year or more before the multiple does.

One item sits outside the income statement. The founder controls 60.8% of the vote, so the decision to keep spending — or to stop — rests with one person and cannot be forced by outside holders [20]. The check that exists is behavioral, not structural: the $31.6B returned in 2025 is the evidence that the owner still treats free cash flow as partly the shareholders'. If capital returns fall while capex climbs, the alignment argument set out in Control and Pay weakens on the one metric that has held it up.