The capex cycle

For a business that grows by building things — a factory, a plant, a hotel, a branch network — future growth is bought years before it arrives. The capital expenditure spent today sits on the balance sheet as work-in-progress, commissions into a producing asset some years later, ramps up over more years still, and only then shows up as the revenue everyone was waiting for. The lag between the money leaving and the revenue arriving is the capex cycle, and reading it is how you see growth before it prints.

But the cycle carries a trap as well as a promise. The same rising capex that builds durable growth in one company builds a value-destroying glut in another. Whether an expansion is good news depends on two things the announcement does not tell you: the return the last cycle earned, and whether the rest of the industry is expanding at the same time.

Because of the lag, capex is a leading indicator — one of the few places the accounts show you tomorrow's growth today. A company that has quietly commissioned new capacity is about to grow into it; a company whose capex has dried up is telling you its growth is about to plateau, whatever this year's numbers say. Reading the cycle lets you anticipate the inflection instead of discovering it in a result.

The cycle also decides whether growth is worth having at all. Capacity added at the top of a cycle, when everyone is optimistic and building, tends to commission into a downturn and earn poorly. Capacity added counter-cyclically, when rivals are retrenching, tends to commission into recovery and earn well. The timing of the capex, not just its size, separates value-creating from value-destroying expansion — which is why this module feeds directly into growth that destroys value and the J-curve later on.

The cycle runs through stages, and each leaves a trace in the accounts:

• Announcement — capex guidance, a capacity number, a project cost. Intent, not yet spend.
• Build — cash flows out (investing cash flow turns sharply negative) and accumulates as capital work-in-progress (CWIP) on the balance sheet. Nothing is producing yet; CWIP is not depreciated.
• Commissioning — the asset moves from CWIP into gross block, depreciation begins, and production starts. This is the inflection to watch: CWIP falls, fixed assets jump, depreciation steps up.
• Ramp-up — utilisation climbs from low to high over several quarters or years. Early on, the asset carries full depreciation and interest against low output, so margins and returns dip before they rise — the J-curve.
• Peak utilisation — the asset runs full, returns are at their best, and the question becomes whether to start the next cycle.

Two reconciliations do most of the work. First, tie CWIP to commissioning: CWIP that rises and then converts into gross block with rising depreciation is a real build; CWIP that rises for years and never converts is a stalled or troubled project — or a place where costs are being parked. Second, tie capex to the return it earned: compare the incremental operating profit a completed cycle produced against the capital it consumed. That realised return on the last cycle is your forecast for the next.

A composite specialty-chemicals maker, Kadamba Specialty illustrative, guides to ₹1,200 crore of capex over three years to double a product line. [illustrative] Follow the trace. Year one and two: investing cash flow swings deeply negative and CWIP climbs by roughly the spend, while revenue and depreciation are flat — nothing is producing yet. Year three: CWIP falls sharply, gross block jumps by ~₹1,150 crore, depreciation steps up, and the first quarter of new volume appears. Years four and five: utilisation ramps from ~45% to ~85%, and the segment's margin — depressed at first by full depreciation on low output — recovers and then exceeds its old level.

The judgement is not "they spent ₹1,200 crore." It is: did the last line they built earn its keep? If Kadamba's previous expansion earned, say, ~20% on the capital it used, this one deserves the benefit of the doubt. If the last one earned single digits and never reached its promised utilisation, the new ₹1,200 crore is a reason for worry, not excitement — same announcement, opposite reading, decided by the record.

The same rising capex inverts in meaning across sectors, and the pivot is industry structure. In a differentiated, branded or specialty business, one company's expansion adds capacity it can fill at its own price — capex is disciplined growth. In a commodity, every producer reads the same high prices and expands at once, so the aggregate new capacity commissions together into a glut that crushes utilisation and price for all of them — capex is collective value destruction. The identical line on the cash-flow statement is a green light in one sector and a red one in the other.

Reading the capex cycle tells you that growth is coming, roughly when, and — from the last cycle's return — roughly how good it is likely to be. It does not tell you the demand will be there when the capacity commissions; a plant built for a market that softens in the intervening years ramps slowly and earns poorly however well it was constructed.

It does not tell you the project will finish on budget and on time. Cost overruns and commissioning delays are common, and they quietly lower the return on the whole cycle — the denominator grows and the revenue arrives late.

And it does not, by itself, distinguish maintenance capex from growth capex. Some of what a company spends merely keeps existing capacity running; only the growth portion adds output. Treating all capex as growth capex overstates the runway and understates the true maintenance cost of the business.

The first trap is cheering capex in a commodity glut. A wave of expansion announcements across an industry reads, company by company, as growth — but in aggregate it is the market building its own downturn. The individual announcement is signal; the industry-wide simultaneity is the signal that overrides it, and the crowd usually misses it because it reads each company alone.

The second is treating CWIP as guaranteed growth. Rising CWIP is money spent, not capacity earned. A project that never commissions, or commissions years late and over budget, converts that CWIP into a poor return — and CWIP that rises for years without converting can be where costs are being parked to flatter the P&L.

The third is ignoring the return on the last cycle. Excitement about a new expansion routinely overlooks that the last one earned single digits and missed its utilisation target. Capital allocation is a habit; the record of the previous cycle is the most honest forecast of the next, and it is the number the announcement is designed to make you forget.