Heavy Capex Spending: Valuing Hyperscalers and Other Big Spenders

Definition

Capital expenditure (capex) is money a company spends on long-lasting assets: buildings, machines, data centres, power plants. Hyperscalers (the companies building the cloud and AI data centres, such as Alphabet, Meta, Microsoft and Amazon) are the best-known example right now. When a company suddenly spends far more than usual, its free cash flow (the cash left after that spending) can shrink or turn negative, even though the business is earning well. A valuation that only looks at free cash flow then treats every dollar of new spending as lost. For these companies KashVector splits the value in two: the existing business, valued on the cash it makes after the spending needed to keep it running, plus the new spending, valued at a return on that money. It also shows the return on the new spending that today's share price implies. This is a description of the method and of what the price assumes, not a view on whether the spending will pay off.

Formula

Value = Existing business + Σ Growth capex × ((R ÷ w) ÷ (1 + w) − 1) ÷ (1 + w)ᵗ
Existing business = The normal discounted cash flow valuation, run on operating cash flow minus maintenance capex
Maintenance capex = The spending needed to keep today's business running: the larger of depreciation and the company's own capex level (as a share of revenue) before the surge
Growth capex = Capex above maintenance, assumed to continue at its last-12-months level for 5 years (t = 1 to 5)
R = The yearly return each dollar of growth capex earns, from the year after it is spent, for good
w = The company's cost of capital (the same discount rate the DCF uses)

The Base case sets R equal to w: the new spending earns exactly its cost of capital, so it is worth what it costs and nothing more. Capex is still counted in full; nothing is assumed to pay off. Bear assumes a return 3 points below the cost of capital, Bull 3 points above. The implied return is the R that makes the Base value equal today's share price. When our value for the existing business alone is already above the price, the implied return is shown as "not meaningful". When it comes out far above the cost of capital, part of the gap is usually the market expecting faster growth from the existing business than the model assumes, not only returns on the new spending.

How to interpret it

Below the cost of capital
The price assumes the new spending earns less than it costs: the market is pricing the spending as value-destroying, or is sceptical of it
Near the cost of capital
The price assumes the spending roughly pays for itself, and the company is priced close to its existing business
Far above the cost of capital
The price assumes the spending earns well above its cost, or that the existing business grows faster than the model assumes, or both
Not meaningful
The existing business alone is valued above the price, so the price implies nothing positive about returns on the new spending

Worked example

Worked example: Meta Platforms (5 November 2026)

Capex, last 12 months$104B, 43% of revenue, 4.2x depreciation
Maintenance capex (2022-23 level of 23% of revenue)≈ $56B a year
Existing business cash flow (operating cash flow − maintenance)≈ $77B a year
Base value per share (new spending at its cost of capital)$656 vs a price of $630
Implied return on the new spending≈ 5.9% a year vs a 10.0% cost of capital
At that price the market was assuming Meta's new spending earns less than its cost of capital, and was paying roughly the value of the existing business. A plain free-cash-flow DCF on the same day valued the stock at about $136, because it counted the whole build-out as lost cash.

Common mistakes

Valuing a company mid-build on its free cash flow alone
Free cash flow subtracts all capex, including spending meant to grow the business. During a surge that makes a healthy company look worthless and gives meaningless answers on the most-watched stocks.
Adding the capex back and calling it "owner earnings"
Capex is a real cost. Adding it back assumes the spending pays off without asking how well. This method charges it in full and asks what return the price needs.
Using depreciation alone as maintenance capex
Depreciation lags a fresh build-out (new assets have not started depreciating fully), so it understates what keeping the existing business running costs and overstates its cash generation.
Reading a high implied return as a forecast
It is what the price assumes, not what will happen. A very high figure usually means the market also expects faster growth from the existing business.
Applying it to a company that borrows to build
A company whose capex is far above its operating cash flow, or that is already in financial distress, carries a different kind of risk. The method only applies when the spending is mostly paid for from the company's own cash flow and the Altman Z-Score is outside the Distress Zone.

Frequently asked questions

When does KashVector use this method?

When all six of these hold for a company that would otherwise be valued with a plain DCF: capex is at least 15% of revenue, capex is at least twice depreciation, free cash flow is negative or below half its best recent year, capex is no more than 1.5 times operating cash flow, the Altman Z-Score is not in the Distress Zone, and no sector-specific model (banks, insurers, REITs, airlines, regulated utilities, cyclical companies) already applies. It is decided from live data for every company, in any sector.

Is this only for AI companies?

No. Alphabet, Meta, Microsoft and Amazon qualify today because of their data-centre spending, but so can a power producer building new plants, a manufacturer building factories, or a defence company expanding capacity, as long as the six tests pass.

Why does the Base case assume a return equal to the cost of capital?

It is the neutral assumption: the spending creates no value and destroys none. It keeps the cost of the spending in the valuation without guessing how well it will pay off. The implied return then shows what the market itself is assuming.

Why is the implied return sometimes "not meaningful"?

If our value for the existing business alone is above the share price, no positive return on the new spending is needed to justify the price, so the solved figure would be zero or negative and says nothing useful.

Related terms

See the heavy capex valuation for any stock