(FORTUNE Magazine) – John Chambers could do no wrong. At least, that was the guiding belief of Cisco's 4.5 million shareholders, who during the Internet bubble seemed to buy stock on the congenitally upbeat CEO's every word. Chambers told them that the Internet was changing the world and that Cisco was the model for what a Net-based corporation would look like. He regaled them with tales of Cisco's paperless business processes. He painted a picture of its real-time access to order flow and its ability to close its books every quarter within 24 hours. And investors believed him. For a few weeks in the spring of 2000 they made Cisco the most valuable company on the planet--worth more than even GE, which had six times Cisco's revenues.

Those days, of course, are long gone. Investors don't have much to show for their faith in Chambers: a stock that at a recent $14 a share is down 83% from its peak; the embarrassment of realizing that a leader they revered is no different from just about every other CEO; and enough agita to send Xanax sales soaring. Chambers may have been better than anyone at touting the Internet and Cisco's role in it; he also turned out to be just as blinded by its hype. When the bubble burst, Chambers had to cut nearly 10,000 employees, write off $2.5 billion in inventory, and restructure the company. Just like everybody else.

He hasn't made it easy for his faithful to digest their disappointment. In November, looking as usual more IBM than Silicon Valley in his suit and tie, he stood before a roomful of big investors and hurried through the problems: Business isn't going to bounce back as quickly as he once thought; the next few quarters are going to be lousy; order backlog is good but not great. The rest of the speech was devoted to--well, exactly the same thing he was talking about during the bubble: The Internet is going to change the world, and the world should look to Cisco as the model for how to make that change happen. "Most customers no longer buy our products for feeds and speeds," he said in his best West Virginia drawl. "They buy it for what productivity advantage we bring to their environment. To do that we have to be the best example, or one of the best, in the industry." As for Cisco's problems? "A hundred-year flood," Chambers declared dismissively. "We got surprised. I wish we had been smart enough to avoid that, but once we were surprised, we executed pretty well against our peers." Translation: Don't blame me. Blame the economy, stupid.

Some investors find it hard to get beyond such remarkable arrogance. "I don't buy anything he says anymore," says Fred Hickey, editor of the High Tech Strategist newsletter and a tech stock investor since the late 1970s. "He's a cheerleader. A hypester." Says another: "He's still telling us about all his meetings with CEOs and government leaders, about how they 'get it.' Like we don't?"

Today, it seems, Chambers can do no right in the eyes of investors, which is too bad, because he's actually doing something interesting at Cisco: turning it around. At a time when the high-tech meltdown is well into its third year--and in some quarters still appears to be getting worse--the $19-billion-a-year Internet switch maker is showing more than just signs of life. Indeed, the last four quarters have been among the most profitable in Cisco's history--beaten only by its performance in the final year of the bubble. Gross margins, the difference between the cost of parts for Cisco equipment and the finished products' selling price, are at a record 69%, up from 65% a year ago. Net income over the past four quarters totaled $2.8 billion, vs. a loss in the previous four. And free cash flow, at about $1 billion per quarter, shows no sign of slowing. The only other company that comes close to Cisco's cash generation is Microsoft, which has amassed a cash hoard of $39 billion. Cisco's comes to $21 billion; if the company were a bank and those were its assets, it would be the nation's 43rd largest.

Most important, Cisco has been using the downturn to seize market share from competitors like Juniper, Nortel, Extreme, and Foundry. During the boom, Cisco and all the networking companies chased the deep-pocketed telcos. When those pockets turned out to be filled with little more than lint, the networking suppliers got trounced. But Cisco's primary business--accounting for about 90% of revenues--was selling routers and switches to the rest of corporate America and to governments. That business, while not growing, hasn't evaporated. Indeed, in some sectors Cisco has actually seen big increases in purchasing. Sales to the U.S. government have doubled in the past year. Add it all up, and today Cisco controls 69% of the market for network switches and 85% of the market for network routers, reflecting gains in each of roughly seven percentage points during the past 18 months.

Which part of Chambers's visionary legacy has led to Cisco's financial resurgence? Well, no part. For all his bluster about the Internet, his cures have been incredibly low tech: He has cut costs. In addition to staff eliminations, he and his team have slashed ten of Cisco's 50-plus product lines, redesigned its routers and switches so that they require fewer and more interchangeable parts, and enforced 40% across-the-board cuts in employee travel expenses. In total, Cisco has managed to cut $1.7 billion, or 17% of its expenses. It turns out that just like every other company during the bubble, Cisco was rushing so fast to get product out the door that it wasn't operating efficiently.

If there's a reason investors aren't cheering Chambers, it's that they're expecting more. Much more. After all, he has long promised a glorious Cisco-driven future. Chambers's mantra has been growth; cutting costs is a game of shrinkage.

Yet Cisco is actually making big moves--though investors won't see the results just yet. As competitors' problems have grown in the past year, Chambers has pushed Cisco hard into new businesses. It has become the No. 1 supplier of wireless-network devices to corporations. It has targeted America's $15-billion-a-year corporate phone business with Net-based phone products. It has improved the integration of firewall and authentication software into its switches and routers, to expand its presence in the security business. And it has set its sights on moving its switches and routers into areas where they used to play only a minor role, such as directing the flow of data for storage networks, and delivering video on demand for cable companies.

Once such sales start rolling in, Chambers's support will come back--albeit perhaps grudgingly. It's already starting to happen. Says analyst Steve Kamman of CIBC Oppenheimer: "Chambers has done a much better job than I expected at dealing with this downturn, and I've been one of his biggest critics."

To get a sense of the scale of Cisco's commitment to its new businesses, check out Cisco TV. In the corner of an office building on its 308-acre campus north of San Jose, Cisco has created a fully functional $3 million cable station. With more than 300 routers, switches, and servers, 12,000 cable modems, and enough fiber and coaxial cable to stretch 300 miles, the station boasts the firepower to serve a community of 250,000 people--rivaling many rural cable operators. Cisco uses the station to transmit video around its campus, but its primary use is to make sure everything Cisco sells to the cable industry works in the field as well as it does in a Cisco sales pitch.

Building a working cable company might seem an extreme way to push routers and switches, but for Cisco the stakes riding on these new ventures are incredibly high. Even the relentlessly upbeat Chambers agrees that his router and switch businesses are so big that they probably won't grow more than 20% a year--and many Wall Street analysts think that's too optimistic. The consensus is that Cisco's core businesses will grow only at about the same rate as overall IT spending. That might turn out to be 10% long term; few think it will hit the 15% rate it achieved during the late 1990s. Chambers says only that he wants to grow revenues in each new business ten percentage points faster than the market average.

In all the markets where Cisco isn't entrenched, it will have to pull out the stops to get noticed. Take wireless Internet delivery, a $2-billion-a-year market growing at some 30% a year: Cisco has to convince corporate IT managers that in the long run it's cheaper to pay five or ten times more for a Cisco wireless router than one available at CompUSA for $100. Similarly, to crack the corporate phone business with its Net-based products, Cisco sales reps must convince customers that it can match the depth of knowledge held by entrenched players like Avaya and Nortel. Even then, the sale won't be easy. Corporate phone networks have long been afterthoughts to companies. Unless the savings are huge--and in the short term they rarely are--why would a CIO want to risk creating problems where none exist?

That's where Charles Giancarlo steps in. The senior vice president for Cisco's switching, voice, and storage business thinks that companies will respond to the beefed-up security in Cisco's wireless routers--and that its crack sales team can persuade businesses to reconfigure their phones completely as they move or build new offices.

All these changes are designed to return Cisco to its growth-stock status. The company is making other tweaks as well, particularly in its R&D labs. While Cisco devotes about 18% of sales to R&D--a mammoth sum of $3.5 billion a year--those funds in the dot-com days were largely aimed at improving existing products. Cisco used its highflying stock as currency to buy some 78 startups since 1993 with promising technology.

Now the company has scrapped its acquire-everything, innovate-nothing strategy. Instead, Chambers has charged his R&D center with developing products from within. To make sure it targets the right goals, he has realigned the company. Before, each product line conducted its own research. Today it's all done under the eye of Mario Mazzola, Cisco's senior vice president of development and one of its longest-tenured executives.

The change is one part necessity, two parts strategy. "There are a lot fewer new companies getting funding right now. That means the likelihood of finding one you want to buy goes down," says Mike Volpi, Cisco's routing chief and its former head of acquisitions. But there is also a hard-edged business reason: The more integrated Cisco's router and switching hardware is with the company's new business offerings, the more entrenched Cisco will become in customers' networks. The more entrenched Cisco becomes, the less danger there is of customers' switching to other suppliers. Cisco is building walls around its territory or, as Chambers euphemistically puts it, "keeping customers from having to make a vendor decision."

What Cisco doesn't need to change is its relationships with customers. While CIOs are practically in open revolt against many suppliers--which, they feel, gouged them during the boom--they rarely have a mean word when it comes to Cisco. "Their products, service, and support are so good that if they moved into new markets, we'd be inclined to look at whatever they did very favorably," says Mike Prince, CIO of Burlington Coat Factory in New Jersey.

Cisco works at customer relations relentlessly, says analyst Mark Fabbi at Gartner Group. It's not just the fact that Chambers meets with some 30 CEOs a month. It's that Cisco's training courses have become so rigorous and sought-after that getting certified can significantly boost a technician's pay. And it's that whenever there is a problem big or small, the folks running the networks in corporations know they can call Cisco. As with IBM in the mainframe era, the mantra among CIOs today is "No one ever got fired for using Cisco."

Cooing CIOs alone won't be Enough to make Cisco thrive. Even as Chambers widens the company's lead over smaller rivals and knocks on new doors, other competitors smell blood--most notably Dell, which has already sold more than 100,000 low-end switches. Even more fearsome than having Dell as a competitor is the change its arrival signals: Commoditization has come to the networking market. For now, Dell's market share is so small that it doesn't even register in surveys, but Cisco is worried enough that this past summer it banned Dell from selling Cisco products. Huawei, a $3-billion-a-year, privately owned Chinese company, is causing a similar stir in Cisco's China operation by developing and selling its own switches.

Cisco likely has some time before Dell-ification starts eating at its margins. But the mere mention of the Austin PC giant sends Volpi to the whiteboard in his office to scribble an illustration of why Dell won't succeed. His argument is that, unlike players in the markets for PCs and servers, Cisco controls the router and switch game from start to finish. In computers, Intel makes the chips, companies like IBM, HP, and Dell assemble the machines, Microsoft makes the operating system, and hundreds of companies make applications. Industry standards help the products interact, but often force companies to compete on price alone. In networking, by contrast, Cisco owns the hardware, the operating system, and the applications. Any company that wants to compete has to start anew. "For anything to become a commodity, it has to be sufficiently simple that you could send your mom out to fix it," says Jim Reese, who runs the network at Google, the search-engine company. "Networks just aren't there yet." For its part, Dell says that commoditization is coming faster than Cisco believes.

Cisco also could eventually face challenges on the high end of its product line. Procket Networks, founded by Juniper and Cisco veteran Tony Li, and Caspian Networks, founded by Larry Roberts, the inventor of the wide area network, believe that by starting fresh with new software and hardware technologies, they can build better Internet equipment for less money. Caspian is making what it calls a superswitch that will collapse the functions of three high-end switches into one. Procket will say only that it is developing a next-generation router and switch, but it has generated buzz thanks to its $272 million venture capital hoard and the pedigree of its founder.

These startups are on to something. Cisco's Achilles' heel is IOS, the operating system it uses in its routers and switches. The program was developed soon after Cisco's founding in 1985--making it as old as Methuselah in tech terms. Over the years, Cisco has updated and patched the software, but techies gripe about its creakiness. Why not just scrap IOS and start over? Cisco's customers remain loyal in part because they already have so much invested in IOS. Any new operating system could weaken that bond. Volpi, for one, thinks IOS's maturity will help stave off the upstarts: "Companies like Procket and Caspian can mount a challenge, but it's going to take them time. We have 1,200 people working on IOS. They have to start from scratch."

What does all this mean for Cisco's giant pack of disillusioned investors? There's hope. Okay, maybe that's not true for those who bought at the top when Cisco was trading at $82 a share. The chances of the stock getting that high anytime soon are remote. But those with more modest goals can think of it this way: Even if Chambers misses on his new targets and Cisco grows at, say, a mere 12% a year through this decade, it'll still likely be a $35 stock by 2010. That may not get prime ministers flying to San Jose just to hear Chambers's pearls of wisdom, but as stock market returns go, it's nothing to whine about--and a far more credible goal.

FEEDBACK fvogelstein@fortunemail.com