The future of the "Internet of Things" (IoT) has been discussed and debated so much over the past two years that you might think that the door has already closed for entrepreneurs thinking about getting into the market. General Electric, Cisco, IBM and others, after all, have already rolled out strategies to colonize large segments of the market while nearly every venture firm has one or more well-funded IoT plays in their portfolios.

And don't forget Google, which bought Nest Labs for $3.2 billion.

Fortunately, that’s not the case. Simply put, there are a lot more devices than people and most of them haven’t been linked up yet. In 2003, the world population was 6.1 billion people and 500 million connected devices, according to various estimates. By 2010, we had 6.9 billion people and 12.5 billion devices. By 2020, we will have 7.6 billion people and 50 billion devices. Put another way, the number of devices will rise from .08 to 6.55 per person.

But to succeed, startups are ultimately going to have to make a very crucial decision: either emulate Apple or emulate Micron.

Entrepreneurs need to either focus on building a somewhat vertically integrated “product” company that sells a complete service under its own brand to consumers and businesses, or focus on being an embedded background player providing services or technology for established brands.

Related: Passwords You Swallow, Sharks That Tweet and 90-MPH Cheeseburgers

At first blush, following the Apple track seems like the obvious choice. Apple is one of the most valuable companies in the world, wielding a market cap that has exceeded the value of ExxonMobil. Public utterances of its CEO and other execs are scrutinized like the Rosetta Stone. Chip companies, screen suppliers and software makers jostle for the company’s business.

Nest and FitBit have done well by going down this path. One of the key attractions of the approach is that you control the customer experience and relationship, potentially leading to increased revenue per customer and recurring subscription opportunities. The most successful companies can additionally open up their platforms to third parties. You can imagine a successful home management company creating an app store.

And Micron? When was the last time the authorized biography of CEO Mark Durcan topped the New York Times bestseller list for several straight weeks? Did you remember the comments on the live webcast when they unveiled their 512 megabyte SLC NAND chip? No? Missed that one?

But when you look a little closer, relative anonymity becomes more attractive. Most technology companies over the past 20 years, in fact, have lived lives of relative obscurity. Demand for flash memory, one of Micron’s main products, is exploding while the number of manufacturers has consolidated. Micron was the one of top performing technology stocks in 2013, according to Yahoo.

Many early IoT markets will likely revolve around integrating networking and intelligence into products without a lot of brand pizzazz: air handlers, cold storage units, warehouse light fixtures. Agriculture -- already taking off as a market -- is interested in how IoT technologies can help manage irrigation systems and pumps.

The technology by its very nature will be embedded and embedded products don’t benefit from glossy advertising campaigns or word-of-mouth. Deals are won or lost through product specifications, field data and a company’s ability to integrate itself into a larger customer’s operations.

Related: 8 Ways the 'Internet of Things' Will Impact Your Everyday Life

Most of the customers -- large industrial conglomerates that have been around for decades -- already have established relationships with manufacturers. A fast food chain isn’t going to switch oven suppliers because a new, untested entrant from Silicon Valley has come up with an internet-enabled one. They are going to wait until their trusted supplier integrates wireless. The opportunity to disrupt the supply chain is nil.

Even many of the IoT companies producing services that will directly impact consumers will end up being sold like industrial products. Utilities, for instance, will likely be one of the biggest customers for smart thermostats: they will buy and install them for free as a way to control peak power. In these deals, there might be five million end-users but only one customer. Again, price, performance and technical specifications will be more important than brand or consumer cachet.

The background route also plays into the strengths of startups. Developing a water dish that tells you when your dog needs a refill is easy. Developing a battery management system and networking stack for a subcutaneous insulin pump that the FCC will also allow patients to wear on airplanes is tough. Most manufacturers don’t have these capabilities in house, but most startups don’t have the channel relationships or technical understanding to build medical devices on their own. By concentrating on a horizontal, embedded technology stack, new companies can compete more effectively and expand the market and the same time.

Ironically, Google’ success with Android has come because it has served as a somewhat even-handed background technology provider to handset makers. How it balances wider ambitions for the connected home with its acquisition of Nest will be closely scrutinized.

Finally, there is the thorny problem of predicting demand. Are smart light bulbs going to be a hit or a flop? Will kids recoil against parental tracking devices? How many IoT products will get absorbed into smart phones? Ultimately, the consumer market, where the end-user product companies will play, will be far less predictable.

Boring may not land you on the cover of Time, but in this case it’s the right move.

Related: Start Preparing Your SEO for the 'Internet of Things'