(FORTUNE Magazine) - Here's a new world record you haven't heard about. If AT&T buys BellSouth as announced, the deal will create the largest agglomeration of capital invested in any company anywhere -- about $280 billion of it.

While that is a truly awesome achievement, it isn't necessarily one to be proud of. In fact, it's a strong reason to suspect that this deal, like most megamergers, will be a loser for the acquiring company's long-suffering investors.

The key idea here is the most fundamental principle in business: A company has to earn a return on capital greater than the cost of capital. Otherwise it's failing. Your local shoe retailer and dry cleaner understand this perfectly, but when people rise to the top levels of giant enterprises, some strange force clouds their minds and they become blind to that principle. And then watch out.

Here's how it works in the AT&T-BellSouth deal. According to the financial consulting firm Stern Stewart, AT&T (Research) has about $193 billion of capital invested in it. That represents the total amount of capital put into the company over the years by equity investors, lenders, and the company itself in the form of retained earnings.

AT&T wants to buy BellSouth (Research) for about $87 billion, which is the value of the new stock AT&T will issue plus the BellSouth debt it will take on. So the combined company would have some $280 billion of capital invested in it. Now the big question: What kind of return could it earn on that investment?

The discouraging news is that for years neither company has come anywhere near earning its cost of capital (or what it must return to reward investors for the use of their money). Right now, according to Stern Stewart, AT&T and BellSouth both face the exact same capital cost: 9.1 percent. But last year AT&T earned a return on capital of only 3.1 percent. BellSouth earned 5.9 percent. These are two drowning swimmers that have decided to clutch each other.

Can they do together what they can't do individually? Let's get out the calculator. The combined company would have a total capital cost of about $25.5 billion a year (9.1 percent of $280 billion). That's the minimum it would have to earn in order to create value rather than destroy it, as both companies are doing today.

For finance wonks, the appropriate earnings measure here is net operating profit after tax, or NOPAT. So we now must ask whether the combined company could earn NOPAT of $25.5 billion, which would just match its capital cost and elevate it to financial mediocrity, way above where either company is today.

The outlook isn't promising. AT&T's NOPAT last year was $5.1 billion. BellSouth's was $3.4 billion. That totals $8.5 billion, which is distressingly short of the goal. But of course this merger, like all of them, promises synergies -- cost savings, new revenues from cross-selling, and such. Specifically, the companies have announced that they expect synergies "to reach an annual run rate exceeding $2 billion in the second year after closing."

Companies in general have a dismal record of overestimating synergies, but let's be generous. Let's assume that "exceeding $2 billion" means $3 billion. Heck, assume it means $10 billion. It doesn't matter. There's just no way that NOPAT gets remotely close to matching the combined company's capital cost, which means this outfit looks like a value destroyer for years to come. It's no surprise that AT&T stock fell on the news of the deal.

The proposed merger is especially worth analyzing because it typifies the sad story of most megamergers, only magnified. Like so many of them, it's a strategically sensible combination at a financially insane price (Exhibit A being the 2001 marriage of my esteemed employer, Time Warner -- the parent of FORTUNE and CNNMoney.com --and AOL).

The reason this deal shows the phenomenon magnified is that it's the result of serial megamergers. AT&T (formerly SBC) bought Pacific Telesis in 1997, Ameritech in 1999, and the old AT&T in 2005, bulking up on capital every time. That's how this latest announced deal could produce the biggest mountain of capital anywhere, even though it won't result in the biggest revenues or profits.

Understand that in theory there's nothing wrong with a company's having tons of capital. In fact, as long as it can earn a knockout return on it, the more the better. But in practice, the greatest wealth creators get along on relatively modest amounts -- Microsoft: $30 billion; Procter & Gamble: $53 billion; even big old GE: $122 billion.

Producing a sufficient return on $280 billion of capital is a feat no one has ever attempted, and in the best of circumstances it would be extraordinarily difficult. In this case it looks impossible.

You can reach Geoffrey Colvin at gcolvin@fortunemail.com