(MONEY Magazine) – Three years after the tech and telecom bubble burst, wiping out trillions in shareholder value, technology stocks are roaring back. Even though the Nasdaq is still far from its March 2000 high, it's up 45% so far this year, and some market watchers have begun to worry about the possibility of an echo bubble.

This month I asked three Wall Street tech watchers about the recent run-up in technology stocks--have we come too far too fast? Their responses were mixed, but each was compelling.

A NEW TECH BUBBLE?

Steve Milunovich, chief global technology strategist at Merrill Lynch, is the most cautious of the group. He's worried that we are back in a mini-bubble. "I think there's a bit of a disconnect between the fundamentals and the stock prices," he says. He attributes much of the stock gains to the Federal Reserve's policy of keeping interest rates low. "Now it's possible that fundamentals will improve sufficiently next year to allow earnings to grow into the current valuations," he allows. But he warns that "valuations are stretched, and if there is any disappointment in the economy and in technology earnings next year, we could have a severe correction."

Thomas Smith, director of information technology research at Standard & Poor's, is more upbeat. "Nasdaq put on its antigravity boots," he quips. "It just won't fall." He sees a broad uptick for the whole sector. "In some cases with tech companies, we've already begun to see some actual tangible improvement in earnings," he says.

Tobias Levkovich, chief U.S. equity strategist for Citigroup Smith Barney, says there are a number of indicators that suggest we are not back in bubble territory. "First of all, many of the stocks that were at $100 and went down to $2 are not back at $100," he points out. Second, "there are higher levels of investor enthusiasm for stocks today than you had before, but the money flows into equity mutual funds are about half of what they were in the early part of 2000." Equity fund flows are at about $20 billion a month right now vs. $40 billion a month in March 2000.

"If you took the Nasdaq as a statistic," Levkovich continues, "it peaked somewhere close to 5200, it went down to 1100, and now we're at 1900, and all of a sudden there's a bubble? We're still more than 60% below the peaks."

Valuations in and of themselves are not the issue, according to Levkovich. "Investors are still kind of hung over with the fear that valuations are what made them make the grand investment mistake in 2000." He contends that the real issue is the inputs into valuations--namely, the earnings side of the P/E. Unlike in the '90s, he says, investors are now paying high valuations on trough earnings rather than peak earnings. "In 2000, instead of paying high valuations for earnings that were supposed to come through in the future, people paid high valuations on what proved to be peak earnings, and as a result, as the earnings flow started to fall so did stock prices."

"A RELUCTANT RUN-UP"

Milunovich has his own theory about why the Nasdaq is posting stronger gains than the S&P and the Dow. "People remember how much money they made a few years ago," he says, "and even though a lot of people got burned, they're anxious to make it back and tend to go back to the same names."

Levkovich, however, attributes the Nasdaq's stronger gains, at least in part, to recent strength in capital spending. "The area we've been waiting for is capital spending, and IT [information technology] is roughly half of capital spending." Levkovich refers to the tech-stock surge as a reluctant run-up, in which "the analysts are chasing what's been happening in the marketplace and the earnings trends." After predicting recoveries in the second half of 2000, 2001 and 2002, he says, analysts were a little sheepish: "This time they came in saying, 'We've been burned three times; I'm not coming in with a significant second-half expectation for recovery,' and then, of course, it happened."

LOOKING AHEAD

All three agree that some parts of the tech sector will perform better than others--and that semiconductors should be among the winners. Smith, who describes semiconductors as "a leading-edge industry," likes Analog Devices, Intel, Texas Instruments and Microchip Technology. Of the chip equipment makers, he prefers Teradyne and Cadence Design Systems. "In the first year of the comeback, lower-quality names out-paid some of the bigger ones," he adds, "but as things broaden out, I think bigger ones will do better." Among them: Automatic Data Processing, Cisco, IBM and Microsoft.

Levkovich also suggests Intel as a pure play on semiconductors. But he warns that what looks good now may not look as good in even a few months. "We've been trying to indicate to investors that one may have to rent stocks rather than own them right now," Levkovich says. "I think that the higher beta [more volatile] areas like semiconductors will probably do better initially and then back off next year."

Echoing that sentiment, Milunovich predicts that "semiconductor stocks...[will] probably go up for now because there's low inventory in tech and there is some component build going on." But he cautions, "I would guess that as we get into next year there's going to be increasing risk in the semiconductor names." He also mentions Intel as an innovative company that is benefiting from the pickup in PC demand. Other Milunovich picks: Qualcomm in the wireless space and Mercury Interactive on the software side.

As for tech companies to avoid, Smith mentions Real Networks as a company with too much concept and not enough earnings. "The laggard industry in technology as we see it is telecom equipment, which is dependent on capital expenditures by largely American telephone companies, and that's been pretty slow," he says. "Right now," Smith adds, "it's hard to find [tech stocks] that you'd like to short on a broad basis because you think they'll fall absolutely."

Milunovich says EDS will continue to be vulnerable, as will Infineon, a DRAM [dynamic RAM] maker. "We have some concerns that DRAM prices may begin to soften next year," he explains. And he finds communications equipment company Juniper very expensive now.

Like Smith, Levkovich sees softness in telecom capital spending, so he's cautious about Lucent and JDS Uniphase. He also notes that Sun Microsystems lacks what he calls the Wintel combination: "Windows and Intel--it's just the way most corporations have structured their systems."

While tech-stock gains have been remarkable this year, we are still a long way from the excesses of the late '90s. But the sector's earnings are also a long way from its lofty price gains. Wise investors would do well to remember the primary lesson of the late '90s: Earnings will be the critical and ultimate determinant of price.

Lou Dobbs is the anchor and managing editor of CNN's Lou Dobbs Tonight.