(FORTUNE Magazine) – Serial entrepreneur Mark Goldstein has started at least four companies, the last being Bluelight.com, which he sold to Kmart in 2001--good thing he got out then! Three months ago I asked him to join our venture capital firm, New Enterprise Associates, as an entrepreneur-in-residence. Mark recently wrote me a letter about his effort to find good new businesses to start up in today's Silicon Valley. Here's an edited version.

Dear Stewart:

Could Ollie be right? Ollie, my deck contractor, tells me the real opportunity is in manufacturing plastic sewer pipe. Everyone needs it, and there are only two worldwide suppliers. He asks, Why waste time starting another Internet company? And he is right about one thing--there's no shortage of content for his pipe.

Anyway, for three months now I've been making like a Hollywood producer, except that instead of flipping through scripts I'm checking out business plans. I've been talking to the same folks who helped me build some very cool businesses in the past, to MIT and Caltech professors, former dot-com CEOs, leading tech analysts, and many a savvy Indian engineer. After all that talk, one thing is clear: Something has changed in the past year. It is much harder to find really big opportunities.

These days starting a business is about a lot more than the "coolest" technology, speed to market, or finding a hot market segment. It's about building a solid business that (1) is based on a suite of patents, (2) consumes little cash, and (3) breaks even within 18 months. There simply aren't any more big companies out there looking to pay outrageous premiums to buy little companies. The easiest exit strategies for entrepreneurs are gone--toast.

That said, I've taken some ideas seriously:

Disposable DVDs. A great idea! DVDs that expire after 24 hours are a neat way to distribute movies through places like Wal-Mart. But this idea has no legs. It's just an interim solution--widespread broadband will make this happen.

Integrate my stereo with my PC. Another great idea: Let consumers use the Net to gather and organize songs and movies that they play back on the stereo or TV. Too bad several startups are already attacking this. Too bad Microsoft seems interested. Too bad no investor wants to back a hardware company.

Use electrical wiring or wireless networks in the neighborhood to deliver broadband and Internet access. Broadband communications will be everywhere eventually, so this is a big opportunity. But delivering access on a big-time basis, through any kind of technology, is not all that different from financing and building a new utility. The idea of asking a venture capitalist for $500 million to start a utility seems, well, dated.

Create new application development tools. I found a really interesting Linux-based alternative to Microsoft's tools and applications. But somehow I couldn't get overly excited about going after Microsoft's crown jewels.

Create a new authentication service that gives merchants and consumers an alternative to Microsoft's Passport. See above. Not to mention that AOL and Sun have alternatives. This would be a great way to become the proverbial ant who gets squashed by an elephant.

I'll figure it out eventually, Stewart. Downturns are a terrific time to start new companies, and there are great ideas out there. In the meantime I've developed these principles for any entrepreneur looking to create a major new company today:

Don't plan to spend a lot of money. In fact, I don't think anyone should spend more than $5 million to prove the business concept and sign up the first paying customers. This means doing stuff like buying your office furniture and PCs used on eBay, and outsourcing everything that's not core.

Start with an experienced team. There is a lot of talent available, people with real-life experience. Find them. Use them.

Avoid the sweet spot of the big guys. That's getting hard. Big companies have learned from the Internet. They'll be damned if a Big Idea will surprise them again. They jump fast at new ideas.

Focus on being No 1. Get into the business six to nine months ahead, and stay there by being savvy and paranoid.

Of course, Ollie might just be right. Maybe the big opportunity is in plastic sewer pipe. But I'm not sure I could put together a team of dot-commers who would know what plastic is, much less sewer pipe! So I'll keep trying on the tech front, keep working with the same old geniuses. We'll find something, someday soon.

STEWART ALSOP is a partner with New Enterprise Associates, a venture capital firm. Except as noted, neither he nor his partnership has a financial interest in the companies mentioned. He can be reached at alsop_infotech@fortunemail.com. His column may be bookmarked online at www.fortune.com/technology/alsop.