Meta’s capex jump buys AI time—but it risks starving FCF while Microsoft turns spend into service-margin growth

On the surface, Meta and Microsoft both accelerated AI infrastructure investment. The second-order difference is the funding arithmetic: how much free cash flow each company can convert into capex without compressing near-term EPS and shareholder returns.

In Meta’s latest trailing-twelve-month view, net income is $68.1B while capital expenditure runs to -$89.3B, producing $40.98B in free cash flow. In Microsoft’s latest trailing-twelve-month view, net income is $133.7B against capital expenditure of -$64.6B, producing $71.6B in free cash flow. That gap matters because the market prices the first visible outcome: whether capex is “growth-normal” (Microsoft-like) or “FCF-negative today to buy tomorrow” (Meta-like).

A hyperscaler’s capex has a predictable physical supply-chain shape. Money first buys the things that remove bottlenecks: compute + networking (GPUs/accelerators and their switching/interconnect), then the energy and site layer (power distribution, cooling, and connectivity transport). Advertising/consumer-internet economics get hit only indirectly—through how fast EPS and margins are pressured by accelerated depreciation and/or cash burn.

So when Meta’s spend reads as more FCF-straining, the order of operations changes: investors tend to reward companies whose products sit inside the critical path (data center buildouts and connectivity) and penalize those whose revenues depend on “consumer/ads demand staying intact while capex rises.”

Markets often react less to “AI spend amount” and more to the path from spend to earnings. In short: the same capex magnitude can produce different outcomes depending on (1) how much of it is funded by operating cash flow versus incremental financing, and (2) how much near-term earnings quality is preserved.

Using the same data lens for both firms, Meta’s free cash flow is positive but notably lower than Microsoft’s in the latest trailing view, implying a tighter cash buffer. Microsoft’s stronger cash generation relative to capex suggests investors may treat its buildout as a throughput investment that eventually converts into higher-value cloud/services earnings.

Below are investable linkages that follow directly from the funding arithmetic and the physical capex path.

Upstream beneficiaries tend to be the compute/network layer. Downstream beneficiaries tend to be the connectivity and critical site infrastructure that enables deployments at scale. The losers (or higher-risk names) tend to be businesses whose near-term earnings quality is more exposed to consumer/advertising softness if capex pressure causes hyperscaler budget reshuffles or incremental platform monetization delays.

Note: the article uses listed-company financials from the platform’s data tools. The specific “same-24-hours EPS and guidance ranges” claim in the brief could not be independently confirmed from primary filings within this run because the SEC HTML pages were not accessible via the web navigation tool (URL scheme error).