We'd all like the problem of having more money than we know what to do with.

Tech companies, of course, have been faced with that happy dilemma for years and continue to be awash in cash.

Intel

(INTC) - Get Report

for example, had a bit more than $7 billion on hand at the end of September, while

Microsoft's

(MSFT) - Get Report

hoard topped $9 billion.

And it's not just the mega-caps that generate huge amounts of cash. Even smaller companies have rivers of greenbacks flowing into the coffers. Take

BEA Systems

( BEAS), which reported earnings earlier this week. The business software vendor has more than $1 billion on hand, including short-term investments and restricted cash.

So, what's a "poor" company to do? Historically, tech companies tended to plow much of their earnings back into the business; this is not surprising, considering the rapid pace of innovation in the industry. And that, of course, is why few technology stocks pay a dividend.

But as the tech sector matured, two other strategies for dealing with excess cash have come to the fore: share buybacks and acquisitions. Both have their uses, but investors don't always reap the benefits they might expect.

In a thought-provoking note published this week, Merrill Lynch semiconductor analyst Joe Osha says "the continued emphasis on stock buybacks as opposed to dividends within the semiconductor industry is not translating into gain for shareholders." Here's some of his research:

Average dividend yield for companies in the Philadelphia Semiconductor Index is 0.5%, while companies on the

Dow

paid 2.2%.

Of the $20.3 billion that the SOX components have spent on buybacks and dividends this year, 84% have flowed to buybacks. The comparable number for the Dow is just 37%.

Semiconductor companies generate more than twice as much free cash flow relative to their asset bases as do Dow companies.

That would be OK, says Osha, if buybacks were translating to gains for shareholders, but they're not. The SOX has underperformed the Dow this year, increasing by about 1% vs. 14% for the Dow. And that's been true for some time. Since the beginning of 2002, the SOX has declined by 7% while the Dow climbed by 22%.

"After five years of listening to managements' claim that they believe their own stock represents a good investment opportunity, we'd think that investors would start asking for the money instead," Osha concludes.

Interesting points, to be sure, but there's plenty of room to disagree.

Becker Capital Management's Pat Becker Jr. says that because the semiconductor industry is so cyclical, it doesn't lend itself to dividends. After all, once a company declares a dividend, it's pretty much forced to pay it -- or take a pounding -- even if it's on the downside of the cycle.

More troubling for Becker is the effect that stock options can have on share price. Over the last decade, electronic design automation companies (cousins to the chipmakers) such as

Cadence Design Systems

(CDNS) - Get Report

and

Mentor Graphics

(MENT)

have roughly doubled their revenue. But the dilution caused by the options has kept share prices down, he says.

Alan Gayle, senior investment strategist for Trustco, notes that buybacks give earnings per share a small artificial bump by diluting the stock. But as long as companies report strong net income numbers -- which they have been -- investors will support buybacks as a way to increase shareholder value. But should earnings head south, they're likely to prefer the certainty of dividends.

As it turns out, not all semiconductor conductor companies are created equal when it comes to dividends. Consider

Microchip Technology

(MCHP) - Get Report

.

Unlike Intel, which lives in the digital world, Microchip develops analog products. The difference? Analog doesn't have to be on the cutting edge of technology, so Microchip and similar companies don't have to spend nearly as much money on capital improvements as their digital brothers, says Standard & Poor's chip analyst Clyde Montevirgen. And that makes the business somewhat more predictable and there's more money leftover for shareholders. In Microchip's case, a yield of about 3%.

Moreover, as technology becomes more of a commodity -- and demand becomes even more widespread than it is today -- the semiconductor cycle will become less extreme, says Montevirgen. At that point, dividends could begin to make more sense in the sector.

Growth Is Growth -- Or Is It?

Acquisitions -- as shown by software companies spending hundreds of billions of dollars on M&A in the last few years -- are great for the top line. But for how long?

Sanford Bernstein analyst Charles Di Bona took a hard look at BEA's recent quarter in which the business software company reported that on a trailing-12-month basis (TTM), license revenue is accelerating into the mid-teens.

But when recent acquisitions are backed out, growth declines to single digits, he says. Including revenue from Plumtree and Fuego, BEA's TTM license revenue in the third quarter grew 15%; without them, growth was only 6.9%.

So, what's the difference? The higher growth rate isn't sustainable. "Once you get to the anniversary date of an acquisition, there's no

increase over the previous year since revenue from the acquisition is then included, Di Bona says. BEA's core businesses, he says, are growing in the low- to mid-single digits.

That may sound like a quibble, but Wall Street demands strong license revenue growth from software companies. When it disappears, shares take a big hit. Indeed, BEA took a 16% hit this week when it reported a

weaker-than-expected third quarter.

And note this: The Plumtree acquisition has passed its anniversary date, and Fuego will reach that mark in January.