Capex Cycles and Free Cash Flow: A 2-3 Year Framework

Capex and free cash flow move on different clocks. The cash goes out now; the cash comes back later — sometimes much later, sometimes not at all. If you own a stock through a heavy capex cycle without understanding where it sits on that clock, you are basically guessing at FCF for the next two to three years.

This post is a framework for mapping that clock. It works for hyperscalers building data centers, utilities building transmission, semis building fabs, telcos rolling fiber, and industrials retooling plants. The names change; the cash flow pattern does not.

The Four Phases of a Capex Cycle

Every capex cycle moves through roughly the same four phases. Knowing which phase a company is in tells you what FCF should look like over the next eight to twelve quarters.

Phase 1 — Announcement. Management guides capex up, usually with a multi-year number. The stock reacts immediately; FCF has not moved yet. This is where consensus estimates lag the most, because analysts often only model the next year's number and treat the out-years as "normalizing."

Phase 2 — Build. Capex hits the cash flow statement. Depreciation has not caught up yet (assets are still under construction or just placed in service), so reported earnings look fine while FCF compresses hard. The gap between net income and FCF is the tell — if it widens by more than working capital can explain, you are in Phase 2.

Phase 3 — Ramp. New assets start producing revenue, but utilization is below steady state. Depreciation is now fully loaded, which drags EPS. FCF starts recovering as capex tapers, but it is rarely linear. This is the phase where investors get impatient and the stock often does the worst, even though the fundamentals are turning.

Phase 4 — Harvest. Capex falls back toward maintenance levels. Revenue from the new asset base is at or near full utilization. FCF inflects — sometimes sharply. This is when buybacks, dividend hikes, or deleveraging show up.

The useful question is never "is capex high?" It is "which phase, and how long until the next one?"

Maintenance vs. Growth Capex: The Number That Actually Matters

Companies rarely break this out, but you can usually back into it. Maintenance capex is roughly the level needed to keep the existing asset base producing. A reasonable proxy: depreciation expense, adjusted for inflation in equipment costs (typically 1.1x to 1.3x D&A for asset-heavy businesses).

Growth capex is everything above that. The reason this split matters: only growth capex should be evaluated against an incremental return. Maintenance capex is a cost of staying in business — if a company cannot cover it from operating cash flow, that is a different problem entirely.

A practical test: take trailing five-year average capex and divide by trailing five-year average D&A. A ratio near 1.0 means the company is roughly in steady state. A ratio of 1.5-2.0x signals an active growth cycle. Above 2.0x — think hyperscalers in periods of heavy investment, or chip equipment cycles — and you are in a meaningful build phase where FCF will be structurally lower than the run-rate suggests.

Modeling FCF Through the Cycle

The lazy approach is to take next year's guided capex, subtract it from operating cash flow, and call it FCF. That works in steady state. It does not work mid-cycle.

A better approach has three inputs:

Capex trajectory. Use management's multi-year guide where it exists. Where it doesn't, anchor to the announced project pipeline and assume a three-to-four-year build profile. Be skeptical of "capex will normalize next year" — it usually doesn't.
Revenue contribution timing. Assets placed in service today rarely contribute meaningfully to revenue for 12-24 months. Map the in-service dates, not the spending dates.
Incremental margin on the new revenue. New assets often run at lower margins initially (ramp inefficiencies, customer concessions, depreciation drag). Steady-state margins are a 2-3 year ask, not a Day 1 outcome.

When you put these together, you get a FCF curve that dips, troughs, and recovers — not a straight line. The dip is the buying opportunity if the project economics are real. It is the value trap if they are not.

Red Flags That the Payoff Won't Come

Not every capex cycle ends in a Phase 4 harvest. The patterns that should worry you:

Capex keeps rising past the originally guided peak. This is the single biggest red flag. Either the projects cost more than planned, or management is chasing a moving target.
Revenue growth lags the asset growth by more than a year past the original ramp guide. The assets are built; the demand isn't there.
Return on invested capital trends down through the cycle, not up. A healthy cycle should show ROIC troughing in Phase 2-3 and recovering in Phase 4. If it keeps falling, the new capacity is value-destructive.
Maintenance capex itself drifts up faster than inflation. This suggests the asset base is getting more expensive to run, which eats the harvest before it arrives.

What to Watch Next

Pull the capex-to-D&A ratio for any heavy-asset name in your portfolio. Anything above 1.5x means you should be modeling phases, not run-rate FCF.
Find the in-service date for the largest projects in the pipeline. That date — not the announcement date — is when FCF math starts to change.
Track guided peak capex versus actual. Quarterly updates that creep the peak higher are your earliest warning that Phase 4 is getting pushed out.
Watch capital return announcements. Buybacks and dividend hikes during Phase 2-3 are management telling you they see the Phase 4 cash flow even if the print doesn't show it yet.