South Korea unveiled a combined ₩2,000 trillion (~$1.3 trillion) investment plan by Samsung Electronics and SK Hynix over the next decade, covering new semiconductor fabs, AI data centres, and advanced packaging. The market's reaction was immediate and brutal — Samsung fell 4.7% and SK Hynix dropped 3.1% on the news.
A company announces $1.3 trillion in investment. Its stock falls. That feels backwards — until you understand how a CFO actually reads capex.
Capex (capital expenditure) is money spent today to build assets that generate cash flows tomorrow. The catch: 'tomorrow' is uncertain, and 'today' is very real. When Samsung and SK Hynix announced this plan, investors weren't cheering future fabs — they were discounting future cash flows. That's the core of NPV (Net Present Value): take every dollar of expected future cash inflow, discount it back to today using a required rate of return, and subtract what you're spending. If NPV > 0, go for it. If not, you're destroying value.
Here's what spooked the market. A fab (semiconductor fabrication plant) costs $20 billion or more to build. Returns arrive years later. The discount rate — which reflects risk, interest rates, and opportunity cost — is the killer. The higher it is, the less those future cash flows are worth today. With the Fed holding rates elevated and geopolitical risk (Middle East conflict, US-China chip war) running hot, discount rates are high. That shrinks NPV fast.
The IRR lesson: investors aren't asking 'will this generate revenue?' They're asking 'will the return on this capital beat what we could earn elsewhere?' Ten years is a long horizon. A lot can go wrong — AI demand could plateau, rivals could catch up, yields could disappoint.
For founders: every time you raise money and plan 'investments,' your investors are running exactly this math. Spend capital in ways where the IRR clearly clears your cost of capital — or be ready to defend why it will.
📚 Learn the concept: NPV & IRR
▶ Play the 90-second CFO game All daily posts