How a Company Decides Whether a Billion-Dollar Investment Is Actually Worth It

Every Big Decision Is the Same Decision

Build a new warehouse.Enter a new country.Launch a new product line.On the surface they seem like different problems but in the financial realm they boil down to a single question.You spend a known amount of cash now in exchange for an uncertain cash flow later.Is that trade-off worth making?The discipline that answers this is capital budgeting and it is the machinery behind almost every major investment a company makes

The Time Value of Money

The entire framework is based on one idea. A dollar today is worth more than a dollar a year from now because today's dollar can be invested and earn a return in the meantime. To compare cash arriving at different times you need to convert future dollars to current terms a process called discounting. The pace you use to do this is the heart of the entire analysis and choosing it well is where most of the judgment lies

Net Present Value Is the Answer

Net present value or NPV takes each future cash flow that a project is expected to generate discounts each one to today adds them up and subtracts the initial cost. If the result is positive the project is expected to create value because the discounted cash it produces is worth more than the cash it consumes. If it is negative the project destroys value and should not be financed. NPV is the most important number in capital budgeting for a reason. It answers the real question directly in dollars withthe risk and timing already incorporated

A Worked Example: One Project, Four Answers

The four metrics in this article can be calculated from a small project and seeing that they disagree is the quickest way to understand why VPN wins

The project. It spends $100 million today. It generates $22 million of free cash flow per year for seven years and then nothing. The company's profitability rate is 10 percent

Net present value. Seven years of 22 million discounted at 10 percent is 22 times the seven-year annuity factor which is 4.868. This gives about $107.1 million in present value. Subtract the cost from 100 million and the NPV is positive 7.1 million

Positive so the project creates value. Finance it

Internal rate of return. The IRR is the discount rate at which the NPV would be exactly zero. Here that is approximately 12.1 percent. Since 12.1 percent overcomes the 10 percent hurdle the IRR agrees with the NPV

Recovery period. 100 million divided by 22 million a year is approximately 4.5 years. Whether that is acceptable depends on a rule of thumb rather than economics which is the first indication that something is wrong with the measure

Now change one input and watch how the decision changes. Suppose the project is riskier than the corporate average and the hurdle should be 14 percent instead of 10 percent

Identical cash flows identical project and a swing of almost $13 million in value from a four-point change in a single assumption that no one can directly observe. That's the next concretized caveat: the argument is almost never about the forecast it's about the rate

So the IRR ratio is misleading. Place the project next to a second. Project B costs $10 million and returns $3 million a year for the same seven years. Its present value at 10 percent is 3 times 4.868 or about 14.6 million so its NPV is positive 4.6 million. Its IRR is about 23 percent

The IRR rates B as almost twice as attractive as A. The NPV says that A creates more value. They are both right in what they measure and only one of them measures what the company is trying to maximize which is dollars instead of percentages

A company that is classified by IRR will systematically prefer small projects because a high percentage return on a small base is easy and a high percentage return on a large base is not. Over the years that quietly tilts a company toward incrementalism. If capital is really limited the appropriate tool is NPV per dollar invested instead of IRR and if it is not limited simply do both projects

These are illustrative figures with clean annuity cash flows. Real projects have uneven timing terminal values and tax effects none of which changes the classification problem

The Hurdle Rate Is Where Judgment Lives

The discount rate often called the rate of return is typically constructed from the company's cost of capital plus an additional premium for the risk of the specific project.determined honestly reflects your risk

IRR and Payback, and Why They Mislead

Two other measures appear constantly and both deserve suspicion. The internal rate of return or IRR is the discount rate that would make a project's NPV exactly zero. It's intuitive because it appears as a percentage but it can misclassify projects when they differ in size or have irregular cash flows. The payback period which simply counts how long until the investment is recovered is even weaker because it ignores everything that happens after the break-even point and pretends that a dollar in year five is worth a dollar.today.When the IRR or recovery disagrees with the NPV the disciplined analyst relies on the NPV

Case Study: The Chip Intel Decided Not to Build

The worked example above spends most of its energy on the discount rate because that's what finance teams argue about. The most expensive capital budgeting decisions in history were lost on the other input

In the mid-2000s Apple approached Intel to supply the processor for a new mobile device. Intel was the dominant manufacturer of computer processors with much greater manufacturing capacity than Apple could otherwise use. Apple wanted the chip at a certain price

Intel ran the analysis and refused. Paul Otellini then CEO described the reasoning years later in a 2013 interview: the price Apple would pay was below Intel's cost forecast and the projected volume didn't justify it. Based on the numbers in front of him it was the right answer. His retrospective assessment was that the actual volume turned out to be about a hundred times what had been estimated and that the cost forecast would have been wrong too because volume at that scale changes what it costs to make achip

Apple opted for ARM-based designs. The iPhone became one of the most profitable products ever created and Intel spent the next fifteen years failing to establish a significant position in mobile processors a market that grew larger than the one it dominated

Note which entry failed. The discount rate was irrelevant. The cost of capital was irrelevant. The decision was destroyed by a volume forecast produced by capable people using the best information available for a product category that did not yet exist

The Channel Tunnel is the same failure seen in the other direction. Construction was originally estimated at around £2.6 billion and came to around £4.65 billion approximately 80 percent more. Passenger and freight traffic was then well below the forecasts on which the financing had been based. The tunnel worked and works today. Eurotunnel's original shareholders lost the vast majority of their money and the company had to restructure its debt more than once

Two projects opposite mistakes same lesson. Intel underestimated a market by two orders of magnitude. The tunnel sponsors underestimated the cost by 80 percent and at the same time overestimated demand. In neither case would a better discount rate have helped and in both cases the finance function spent its energy on the rate because the rate is the part that can be argued with authority

Where NPV Itself Breaks

The VPN is the correct default and has three genuine failure modes worth knowing about

It's supposed to be now or never. A standard NPV compares doing the project today with not doing it and ignores the value of waiting for information. Many real investments are better understood as options: spending a small amount to preserve the right to commit later once the uncertainty is resolved. A pharmaceutical trial an exploratory well and a pilot facility have a value that a simple NPV records as zero or negative and the literature on real options exists precisely because of this gap

Brutally punishes the distant future. The discount is exponential. With a 14 percent hurdle a dollar arriving in year twenty is worth about 7 cents today. That's arithmetically correct and means that any project whose profitability is genuinely durable - fundamental research infrastructure with a fifty-year lifespan decarbonization - will have a hard time clearing a corporate hurdle no matter how valuable it is. The framework has a structural bias toward the short term that is attributed to managerial short-termism when it's actually built into the mathematics

Companies use rates of profitability much higher than their own cost of capital. Surveys of corporate practices have repeatedly found that companies apply hurdles in the mid-teens while their actual weighted average cost of capital is closer to 8 or 9 percent and those hurdles are notoriously rigid barely moving when interest rates change

My opinion is that the discount rate gets more attention than it deserves and the forecast gets much less mainly because arguing about the rate feels like finance and arguing about the forecast feels like questioning the business

The Real Work Is in the Assumptions

Because each input is a forecast the honest part of the job is not running the formula once and declaring an answer. It's about testing the model asking how the NPV moves when volume price or cost is worse than planned and identifying one or two assumptions on which the entire decision really depends. A recommendation that a project is worth doing unless a specific factor fails by more than fifteen percent is much more useful to a leader than a single confident number

How I Would Actually Present a Capital Request

If I were to bring one of these into a room I would organize it around the decision and not around the model

I would start with the break-even assumption instead of the NPV. Saying that this project works unless the volume is more than 18 percent below plan gives the leader something to argue with based on his own knowledge of the business. Saying that the NPV is $7.1 million invites acceptance or fighting over the discount rate and neither is helpful

Second I would show the rate of return sensitivity from the table above every time without being asked. It's the one input that the audience is most likely to disagree with and putting the range in front of them turns an argument into a choice

Third I would indicate what input the answer actually depends on and I would expect it to be a volume or price forecast rather than the rate. Intel is the reason. If the recommendation is strong with respect to the discount rate and weak with respect to the demand estimate the meeting should be about the demand estimate

Fourth I would explicitly ask if the decision should be made now because if not the option to wait has value and the NPV has not counted

Fifth I would write down what would prove me wrong and when we would know it so I could qualify the decision later. Almost no capital requests include this and almost all should

This Is the Core of the Job

When a finance team is asked to identify capital investment requirements or evaluate a new initiative this framework is the machinery that does the work. A company that spends approximately $200 billion in capital in a single year is not making a gigantic decision. It is performing thousands of these analyses each of which is a small bet that the money returned from a project will overcome the obstacle the company set for itself. Learn how to build and defend a VPN and you can sit in the room where they are made.those bets

The Bottom Line

Every major investment boils down to spending known cash now for uncertain cash later and net present value responds to that directly in dollars with risk and opportunity already within it

The arithmetic shows why it works and where the arguments occur. A project that costs $100 million and returns $22 million a year for seven years has a positive NPV of 7.1 million at a 10 percent hurdle approximately zero at 12 percent and negative 5.7 million at a 14 percent hurdle. Same cash flows $13 million swing from an unobservable input. And a smaller projectwith a 23 percent IRR may create less value than a larger one with a 12 percent IRR which is why percentage ranking silently biases a company toward small bets

Intel is the reminder that the rate isn't usually what kills you. It refused to make the original iPhone processor because the price was below its expected cost and the volume didn't justify it and the volume turned out to be about a hundred times greater than any estimate. The Channel Tunnel was about 80 percent over budget with less traffic than expected and wiped out its original shareholders. Stress test the forecast mention the assumption on which the decision depends and be honest that the discount rate is the partthe one everyone argues about because it's the part that feels like finances