The Rise of the Instant Company

Business 2.0: Commoditization isn’t a curse. For a new wave of entrepreneurs, it’s a blessing. Here’s how they’re using it to make millions.

Krishna “Kittu” Kolluri calls it “a gift from God.”

He’s not talking about the $295 million he recently got for selling his baby-faced little startup, though the sum, particularly since the company is only three years old, qualifies as fairly miraculous given the state of tech lately. He’s talking about what enabled him to create the company in the first place: cheap off-the-shelf parts.

It might seem odd that a tech veteran like Kolluri, who made his bones as one of the early engineers of Silicon Graphics Inc.’s famously high-end computers, would genuflect before the cut-rate processor and the bargain-basement disc drive. But consider: Kolluri’s big score happened because his startup, Neoteris, was able to fashion its flagship networking device out of hardware that’s available at any decent neighborhood computer store. There’s a $900 Intel Xeon x86 chip that’s twice as powerful as an equal-price chip of three years ago. There’s a security chipset that used to cost $200 but now costs $100. The operating system? Linux. Free.

The low-cost parts enabled Kolluri to pour his resources into developing hot-rod software to make his product work better than anything on the market. “That’s what people were going to pay for,” he says. Pay they have: Neoteris’s box sells for $10,000 a pop, and packs far more features than similarly priced products from Symantec and other rivals. Neoteris’s total hardware cost: $2,000. By the time networking gear maker NetScreen announced it would buy Neoteris in October, it was closing in on profitability and had $20 million in the bank.

Kolluri and Neoteris are pioneers in a new era in the tech world. Call it the age of the instant company. Commoditization–the process by which technology products become standardized and their prices are shoved relentlessly downward–has now driven the cost of the building blocks of tech so low that hardware expense has been all but eliminated from the equation of creating new products. Now entrepreneurs and innovators are discovering that they can use cheap standardized parts as basic ingredients, stir in their own flavor of distinctive software, and–voila!–pop out a low-budget company that can swiftly attack a niche market or outmaneuver bigger, bloated rivals. “We are seeing the formation of whole new paths to creating new companies and remaking old ones,” says Michael Raynor, a management expert at think tank Deloitte Research.

Cozying up to commoditization will take some doing for many in the tech world. With the notable exception of Michael Dell, there haven’t been many executives who viewed relentlessly falling prices as anything but a dread disease that has killed countless companies. Some tech managers, in fact, call it “the Big C.”

But the process is unstoppable, and the emerging cadres of people who are learning to turn it to their advantage are likely to reshape the tech business in profound ways. “There are going to be very few new billion-dollar startups in the future,” says Sanjay Subhedar, a VC with Storm Ventures in Palo Alto. Instead, Subhedar and others foresee swarms of smaller outfits that won’t need as much venture money because of their reliance on off-the-shelf hardware. These companies, like Kolluri’s, will tend to cash out not through bubble-era-type IPOs but by selling themselves to larger operators before another ready-mix player springs up to cut into their action. “We’ll see a lot of hundred-million-dollar startups” that have short and sweet life spans, Subhedar says.

The instant-company era may reserve some of its biggest opportunities for entrepreneurs, but established companies are well-advised to heed its implications. Because if they don’t, they risk having life spans that are short–and not sweet at all.

Collapsing prices have always plagued the tech industry. The personal computer, disc drives, memory chips, and low-end servers all had their margins pounded into the ground long ago. What’s different today is the head-snapping speed with which the phenomenon is spreading–and where it’s going. Like a virus rapidly jumping hosts, commoditization is moving into cell phones, DVD players, digital video recorders, and so on. Wi-Fi chipsets that five years ago cost $50 now go for 10 bucks.

The process is invading ever more complex technologies–high-end storage systems, networking gear, even supercomputers. A group of professors and students at Virginia Polytechnic Institute recently cobbled together 1,100 off-the-shelf Apple computers and, over the course of three months and for just $6.7 million, created a supercomputer that appears to be the world’s fourth-fastest. Machines of similar horsepower cost hundreds of millions of dollars and take years to develop.

But why is commoditization accelerating now? For one thing, corporations haven’t been buying technology. That’s forcing the industry to turn to the consumer market, where prices of many technologies are subject to the same margin-crushing forces as, say, paper towels.

The bigger impetus comes from the x86 factor. The Intel chip architecture was first introduced in the 1970s and was a race car in its day. In time it too became commoditized and, as the brains of most of the world’s PCs, so common that it’s somewhat taken for granted. But many generations of Moore’s Law later, new x86-based chips are not just cheap but tremendously powerful. Since 1998 the price of a typical x86 chip has fallen roughly 30 percent to $140 as its speed has jumped 12-fold. “The x86 has brought PC-type economics to markets that have never seen them before,” says Steve Milunovich, Merrill Lynch’s head of technology research.

One of those markets is data storage, heretofore dominated by the likes of EMC, which became a powerhouse by building machines with proprietary chips and other custom parts to warehouse corporate data. Cliff Miller has come up with another approach. A veteran engineer, Miller zeroed in on so-called network-attached storage–using off-the-shelf x86 chips and other cheap parts to build PC-style storage servers for networks. Not that he was interested in putting the boxes together himself. “I wouldn’t touch hardware with a 10-foot pole,” he says. “I would have been crushed.”

Instead, Miller and his new startup, Mountain View Data, focused on creating specialized software for managing network-attached storage systems. Miller, 46, originally funded his company with $3 million of his own money. He took advantage of another tech field that’s rapidly being commoditized–talent–by setting up a development shop in China, where top-notch Linux programmers work for a fifth of what their U.S. counterparts command.

Because of the Chinese farm-out and the use of cheap hardware, boxes equipped with Mountain View’s software cost about one-fifth the price of some competing products. Miller says that in the two years since Mountain View’s software hit the market, the company has rung up several million dollars in sales, mostly in Asia, where manufacturers slap it into x86-powered machines. He expects his startup to become profitable next year and predicts sales of around $10 million for 2004. Many experts believe that Miller has found one of tech’s next sweet spots: Commodity-based storage is up to 70 percent cheaper than today’s high-end storage systems and packs more than enough power for many companies’ needs. “We’re in the twilight of the hardware-centric age of storage,” says Peter Gerr, an analyst with Enterprise Storage Group. “The value used to be in the hardware. Now it’s in the software.”

The same software-is-supreme phenomenon is making many other tech markets fertile ground for instant companies. Sean Patterson and Tim Cutting spent a collective 10 years as executives at Sun Microsystems, which is locked in a struggle against commodity servers powered by Microsoft or Linux software and Intel chips. In January, with $40,000 in seed money–and while still employed at Sun–they co-founded a company to fix a problem that had bedeviled both men for years: It was a hassle to get their VCRs to record Japanese animation aired on PBS, and they didn’t want to pay for the easy-to-use digital video recorder TiVo. “We’re cheap,” Patterson, 37, explains.

In August their four-person startup, Niveus Media, released the OneBox, a device that combines a DVR, a DVD player, an MP3 jukebox, and a PC running Windows XP–all for $899. (Any TV serves as the monitor.) Almost everything in the OneBox is a commoditized component; the magic is in the software created by Cutting and Patterson to make it all work together. “We don’t have to incur any traditional research and manufacturing costs, since we can build this off commodity products,” says Cutting, 30. Five years ago, a DVD player alone would have cost more than the OneBox.

Now the OneBox has become a hot commodity: Niveus, based in Los Gatos, Calif., has sold nearly 1,000 of them so far. Cutting says the company will break even next summer and will be selling 1,000 OneBoxes a month by the end of 2004.

One of the challenges of dealing with the commoditized world, even for those trying to capitalize on it, is that there are ever fewer barriers to entry. Just about anyone with a good idea for software can afford to build a machine to run it on. And the home entertainment market where Niveus Media is making its play is already crowded with other startups jostling to turn commoditization to their advantage.

In a San Francisco office crowded with dismembered PCs, plasma-screen TVs, and two very large dogs, Mediabolic CEO Dan Putterman and his team are working on their version of the home entertainment operating system that ties DVD players, CD jukeboxes, and other devices into computer networks. Unlike Niveus, Mediabolic’s software isn’t aimed at a single all-encompassing device like the OneBox; Mediabolic’s software can run on anything that uses standard–and, yes, commoditized–audiovisual chips made by Texas Instruments and many others. Mediabolic’s customers already include Denon, Fujitsu, Hewlett-Packard, and Pioneer; all are using Mediabolic software in their own quests to fuse home entertainment and the PC. Putterman says Mediabolic will have “a few million” in sales in 2003 and will become profitable in 2004.

Startups aren’t the only companies that can apply ready-mix techniques. In March, Matt Medeiros was named CEO of troubled SonicWall, a network security gear maker. SonicWall has been around since the early ’90s and, after a boom-era IPO, was worth $3.8 billion at one point. But it relied on proprietary chips that were taking as long as two years to develop and were costing the company nearly $20 million a year. Competing with deep-pocketed rivals, mainly Cisco Systems, that approach was proving ruinous: SonicWall lost $94 million on sales of $103 million in 2002.

Medeiros is a veteran of the chip industry and has his own scars from that business’s brutal price collapses, but he saw commoditization as SonicWall’s only shot. Custom chips were siphoning so many resources that “we couldn’t focus on our software, a business where we have 90 percent margins.” So Medeiros sold off SonicWall’s chip business and retooled its product to run on an x86 Xeon chip. The moves have enabled SonicWall to cut its annual R&D budget more than a quarter, to $22 million, and freed up money to hire more engineers to perfect its software.

“Somebody can reverse-engineer our box in a few weeks, but they can’t do that with the software,” Medeiros says. SonicWall’s new strategy is just taking hold, but its main product now sells for about half the cost of a rival Cisco box. Pacific Growth Equities analyst Erik Suppiger predicts that the company will lose only $8.6 million this year and turn profitable in 2004. “Commoditization saved my company,” Medeiros says.

Network Appliance also found unlikely salvation in the Big C. The storage company was founded in 1992, but its performance skyrocketed after it shifted to the commodity-based model about four years ago. Today its boxes are snapped together from x86 processors and other cheap parts. “Eighty percent of our engineers are software engineers, because we’re a software company,” says Daniel Warmenhoven, Network Appliance’s CEO. That, analysts say, is one reason the company managed sales of $798 million in 2002, up 38 percent from two years ago; it has remained profitable throughout the tech bust. Over the same period, storage market leader EMC’s sales dropped more than a third to $5.4 billion; it lost $118.7 million in 2002.

EMC itself is borrowing from the instant-company playbook to stanch the bleeding. In the past six months, it has spent more than $3 billion to buy makers of specialized software that runs on standardized hardware. Mark Lewis, the company’s CTO, says the moves show that EMC has fully embraced commoditization. “Better to do it yourself,” Lewis says, “than to have someone do it for you.”

Beating the scourge of collapsing prices doesn’t guarantee lifetime immunity. Indeed, the instant companies are helping to create an environment that may prove as Darwinian as any the tech business has seen. Subhedar, the Storm Ventures VC, thinks the swiftness with which companies now can be formed–and re-formed–will accelerate the metabolism of the entire industry and force everyone to run faster or be crushed. “You’ll need to be profitable on less than $25 million of capital,” he says; otherwise, investors will look for someone else who’s using cheap parts and premium software to attack the same problem. Historically, VCs have given startups five years to take a product from idea to market. Soon, Subhedar says, companies may have no more than 24 months to prove themselves.

For those that do, there’s always the threat that Dell will someday come after them, and no one can make anything cheaper than Dell can. “We can’t play the price game forever because Dell is going to eventually kill us,” Cutting says. (Indeed, Dell is already plowing into home entertainment markets like Niveus Media’s.)

So the only option for the instant companies is near-instant improvement–constantly making software better and adding even more cut-rate hardware features. Niveus Media’s next OneBox will have audiophile-quality sound and a better DVD player. Likewise, Miller is already expanding Mountain View’s software lineup to manage more network functions, not just storage.

As for Neoteris’s Kolluri, he’s not wasting any time counting that $295 million. Now a general manager at NetScreen, he’s eyeing the new 64-bit chips and overhauling his software, and he thinks he can double the power of his networking devices without denting his margins. In the era of the instant company, “there are new ways to make a ton of money,” Kolluri says. “But if you slow down, you’re dead.”

Om Malik ([email protected]) is a senior writer at Business 2.0.