By George Anders, on

Christian Gheorghe hates spending money if he can avoid it. As a teenager in Romania in the 1980s he replaced vacuum tubes to keep his television running and subsisted on a used laptop computer with a busted keyboard. Even after immigrating to the U.S. his frugal habits persisted.

In 2005 Gheorghe seethed at the expenses involved in building OutlookSoft, a business performance company where he was chief technology officer. It cost more than $600,000 to get rolling, buying dozens of servers, ­hiring support staffers to run the ­machines and acquiring costly licenses from the likes of Microsoft and Oracle. “You linger over every single penny,” Gheorghe recalled. “I kept needing to pay for another license. It never stopped.”

Gheorghe isn’t grumbling anymore. He now runs Tidemark Systems, a maker of business-analytics applications that is a penny-pincher’s delight. His corporate e-mail is free, via Google. Other open-source programs take care of databases and word processing. As for computing power, that’s not his headache anymore. He rents access to a constantly changing array of offsite servers, paying as little as 12 cents an hour for such cloud computing.

Silicon Valley is abuzz with ­excitement about low-cost startups. Building a company on open-source software and cloud computing is being hailed as a brilliant fusion of the Valley’s three great virtues—cheaper, faster and better. Take extra whacks at costs via social media marketing, crowdsourced design and offshore engineering, and the perceived gains get even bigger. ­Estimates are that today’s most ­ambitious startups can take shape for $100,000 or less, a mere one-tenth of the cost a decade ago.

“It’s a great time for entrepreneurs,” says Raman Khanna, a venture capitalist at Onset Ventures. Open-source software alternatives keep improving. Cloud computing prices keep dropping every ten weeks or so. Funding an ambitious startup’s first few months ­requires only a ­personal Visa card.

When everyone’s costs plummet simultaneously, though, chaos can ensue. Banks keep proving this point, embarking on ill-advised lending sprees whenever they become awash with cheap funds. Similarly, so many telecom carriers in the 1990s rushed to install money-saving fiber-optic cable that the sector became vastly overbuilt, eventually leading to a slew of bankruptcies.

Hundreds of software companies have formed in the past year, according to TechCrunch’s CrunchBase registry. Many target identical markets. Kleiner Perkins Caufield & Byers is talking about forming a $100 million fund to invest solely in cloud computing startups. Kleiner, you’ll remember, placed a big bet on green tech a few years ago, just as those opportunities were ­peaking.

Cheap startups transform the venture capital game, and not necessarily in a soothing way. Early-stage venture investors like to gain control of tiny companies when they’re in their youngest days, at dirt-cheap prices. But cloud-based startups tend to start looking for venture money only after their businesses have taken off. In such situations venture capitalists end up paying a premium for a smaller stake. Some ultrasuccessful startups such as Facebook or Dropbox can be lucrative for venture capitalists who arrive late, but many more startups get funded at high valuations that eventually prove unwarranted.

Exciting as each business niche seems, it may be difficult for any aspiring entrepreneur to build a “defensible moat” around his or her approach, says Trevor Oelschig, a venture capitalist at Bessemer Venture Partners. If too many hard-to-distinguish rivals are all vying for the same customers, a shakeout is guaranteed.

Still, it’s tough for venture capitalists to walk away. Andreessen Horo­witz (which is backing Tidemark) has invested in cloud computing ideas that range from database management to creating architects’ blueprints. The firm so far has done well enough that its founders, Marc Andreessen and Ben Horowitz, appear high on the FORBES Midas List of top tech investors. “I don’t mind too much if the idea isn’t fully proven,” says Peter Levine, an AH partner. “We’re willing to try some experiments.” (See story on Andreessen Horowitz on page 72.)

Egging on the entrepreneurs is The Lean Startup, a 2011 manifesto portraying agility as the greatest virtue. Author Eric Ries urges companies to build, test and learn as quickly as possible, making lots of mistakes but constantly learning from them. His book has become both battle cry and phrase book for entrepreneurs throughout Silicon Valley. Devotees mimic the book’s gung ho, restless tone, including its advice to change course, or “pivot,” whenever market feedback tells them their original product ideas aren’t working.

At Tidemark the lean-startup ethos is best seen in an abandoned server closet at the company’s headquarters in Redwood City, Calif. There’s nothing inside except some metal casing and lots of cartons of Crystal Springs bottled water. The computer power that used to be in the closet now comes from Amazon Web Services, and Gheorghe, the Tidemark CEO, is glad to say why.

Young companies traditionally have struggled to run their own computer departments, he explains. It’s too hard to staff a round-the-clock information technology department with only three or four people, he says. It’s even tougher to buy the right number of servers. Own too few and the site slows down or crashes. Own too many and expensive computers sit idle. Many startups end up using servers at barely 15% of capacity, fearful of traffic spikes that might crash an insufficiently built-out site.

Cloud computing has offered startups a way out. Vendors such as Amazon, Rackspace and Verizon’s Terremark process data like Cargill handles wheat. These cloud computing specialists ­operate vast squadrons of servers in suburban office parks, allowing customers anywhere to rent as much or as little computer capacity as they need.

Running such data centers can be a pretty good business, even with startup customers enjoying huge discounts over the cost of running servers themselves. The big guys can dial up operating rates substantially higher than a startup’s 15%, safe in the knowledge that one customer’s traffic spikes will overlap with another’s lulls. Add in the rewards of buying servers by the railcar and hiring IT specialists by the hundred, and cloud computing becomes a low-cost utility much like gas or electricity.

Want to store 100 gigabytes of data, equivalent to all the books in a good-size bookstore? That will cost you about $10 a month in the cloud. Would you rather crunch through an extra-large database for hours on end? Be prepared to spend 97 cents an hour for the privilege. Entrepreneurs find those prices laughably

appealing, and the terms keep getting better. In the past four years, says Amazon Web Services’ chief, Andy Jassy, his company has cut prices 17 times.

Typical among the thrifty startups is San Francisco’s Reppify. The 12-person firm is building software to help analyze job candidates’ profiles on social sites such as Twitter, Facebook and LinkedIn. Sometimes Reppify needs to crunch through a torrent of résumés in a hurry; other times the flow abates.

Such zigzags are fine with Maneesh Raswan, Reppify’s chief technologist. He’s hardly a key account for Amazon, spending only $1,000 a month. Even so, Raswan can get extra computer capacity within minutes and then shut it down when demand abates. “If we owned our own servers, it would take me a week or two to get a new one provisioned,” Raswan says. “And then I’d be stuck with it forever.”

The rise of open-source development and productivity tools has also lowered startups’ costs tremendously. At Tidemark Gheorghe hunts for low-cost tools with the intensity of a coupon-clipper at a supermarket. Tide­mark’s internal letters and spreadsheets are created with Google Docs rather than the costlier Microsoft alternatives. Phones are hooked up via voice-over-Internet technology rather than the telephone company wiring. “I rarely think about licenses now,” ­Gheorghe says. “Being able to pay as you go lets you focus on what matters most: solving customers’ problems.”

Another zone of vanishing costs: market intelligence. Old-time focus groups or clipboard-toting experts have given way to dirt-cheap online research. When Anne Kallus last autumn wanted to build an online wedding-dress store, it took her just $300 of ­online-ad experiments to gain key ­insights into shoppers’ tastes. All she needed to do was show a variety of sales pitches to 50,000 newly engaged women on Facebook—and then tally up which ad pitches generated the most clicks for her site,

Such quick surveys aren’t the ultimate in scientific precision, startup CEOs acknowledge. But so what? Small companies would rather try lots of these cheap experiments—known as A/B tests—until they find some insight that’s explosively powerful, rather than burn through cash by seeking deeper insights via old-school approaches.

“Fundamentally, social networks are changing the ways we all discover things,” observes Ryan Sweeney, a venture capitalist at Accel Partners. The payoff for startups includes cheaper and faster publicity, brand building and sales strategies.

Young companies that might have hired p.r. advisors for $10,000 a month in the 1990s now find that they can attract attention just as well, if not better, by putting the energy into social media. The most ambitious newcomers may still throw big launch parties, woo journalists and advertise on radio, television or Silicon Valley’s most traveled highways. For others, Twitter or chat boards can generate similar buzz at hardly any cost at all.

Access to low-budget expertise around the world is also more readily available than ever. Give credit to chat boards, online ratings systems and word of mouth among startup executives. Companies that might have seen only two options a decade ago—hiring U.S. engineers or shipping the job to India for a fraction of the cost—can now align their needs with the global talent pool more ingeniously.

Catchy logos needn’t come from artists within nearby Zip codes, for ­example. Instead, sites such as ­ provide quick online access to thousands of potential ­designers around the world. Projects typically ­attract six to ten designers competing for a $300 job, with each entrant typically submitting a handful of designs. Recent winners have included designers from New Zealand, Italy, Indonesia and ­Slovakia. Customers pay only for the winning design.

Gordon Tucker, who ran in the late 1990s, watches the rise of such services with admiration. Back then he hired swarms of greeting-card designers in San Francisco at $60,000 a year each. “If I were doing that now,” Tucker says, “I’d crowd-source some of it instead.” By seeking bids from all over the world, he says, he probably could have cut his design costs by two-thirds.

Eve Blossom, another veteran of early Internet companies, is applying that lesson to her latest startup, ­, which will, when it goes live later this year, sell artisanal goods from Cambodia, Mexico and other emerging markets. To build the new website, she put a technologist in the U.S. in charge of the project but gave him free rein to hire coders from Argentina and Romania, sending their work to testers in the Philippines for quality checks.

Thanks to Google’s Picasa photo-sharing service and Skype’s free phone service, Blossom says, she could run meetings smoothly from California with partners more than 5,000 miles away.

At Tidemark, founder Christian Gheorghe has installed the first versions of its business analytics software with pilot customers. Now he must ensure those creations work as promised. After that it will be time to hire a sizable sales force and seek enterprise customers worldwide. Software sales specialists aren’t cheap; full compensation packages can run at an annual rate of $200,000 or more per person.

Tidemark has raised more than $35 million of venture capital, with most of that money coming in January. Even on the sales side, Gheorghe thinks he can stretch a dollar further than most. He explains that traditional software systems can get caught up in a 6- to 12-month purchasing cycle because ­customers must test them extensively before installing them on in-house servers. Because Tidemark software operates in the cloud, he contends, it can become operational with clients much more quickly.

Tidemark has three main corporate customers right now, led by U.S. Sugar. If Gheorghe’s company can attract wider interest, venture capitalists will likely end up with an investment winner. If not, at least Tidemark won’t have to sell its servers and other equipment at fire-sale prices.