Up until about 12 months ago, we heard every week from new startups focused on big data. Entrepreneurs were patting themselves on the back about catching a huge tailwind of market momentum and seemed surprised that we were worried that the big data market space would become too crowded for most startups to do well. (See here and here for my recent articles on why crowded market spaces hold many underappreciated dangers for startups, and how to navigate them if you choose to go down that path.)

We do not hear so much about big data from new startups anymore. Many big data startups were unable to reach escape velocity, and the number of bona fide successes is still quite small. Big data as The Next Big Thing — in which success was assured because it was The Next Big Thing — has been replaced by the Internet of Things (IoT) in the hype cycle. I want to use the IoT as an example of why Next Big Things can be uniquely dangerous places to focus a startup.

I’ll highlight four key dangers; and while there are probably more, navigating even four simultaneously is the sort of choice an entrepreneur really needs to think about before going into a market segment. This is not to say that no IoT startups will have good exits. Rather, there will be a higher failure rate for startups in IoT than in other, less obvious big market spaces, simply due to the dangers of Next Big Things.

Danger 1: The Law of Large Numbers. We often hear that the IoT will work at a scale never before achieved in human history. It will produce many opportunities because the numbers are so eye-popping big. Everybody quotes a Cisco study purporting to show that a trillion devices will connect to the IoT by the end of the decade. (Coincidentally, when big data was all the rage, the entrepreneurs all quoted a Cisco study showing the amazing amount of data being collected on storage every year. Of course, that study overlooked completely that a very large proportion of those bytes were video pixels, which is not as easily amenable to analytics as web session logs are.)

As VCs, we are very inclined to want to dig into any quoted Large Number, especially when it comes from a firm that is not directly in the mainstream of that number. Do not try to impress us just by throwing a Large Number down on the table during a pitch. You should show us that you really understand why we should believe in that number and that you have an analytical edge that your competitors do not.

Danger 2: Everyone has heard about The Next Big Thing already. Of course, this leads to scores of startups being funded that have all the attendant dangers I cited in my article on the risks of competing in a crowded market. Do you really want to be trying to raise a Series B or C for an IoT startup amidst the bones of tens of startups that have already failed?

Danger 3: The Next Big Thing may not constitute a standalone market. The Internet of Things is a set of communication, computing, and storage protocols being grafted onto a multitudinous range of devices, such as cars, door locks, industrial platforms, factories, and the oft-referenced home sprinkler system, to name just a few. The specific danger here is that all these devices are their own unique markets, each with evolving product roadmaps at completely different rates, for completely different reasons. The networking industry moves faster than most. Grafting IoT networking onto a much slower evolving industry (e.g., home sprinkler systems) will not cause those industries to speed up. They will cause your IoT startup to slow down to a pace that cannot generate the high growth venture capital craves and rewards.

I can think of at least four technology submarkets for IoT. They are the IoT sensors, the microprocessors and commutations chips enabling them to connect, the new software protocols that have to be created to scale to a trillion connected devices, and finally the cloud services to set up, manage, and analyze the data from the IoT. These four submarkets will have hugely different gross margin structures and capital needs. You have to pick one profitable enough to support a startup that deserves venture capital while hoping some other party is happy to take the low margin work. And that someone is happy to fund those development efforts. And you have to wait for the other guys to finish their parts before the whole ecosystem produces big economic rents because it has the “serial circuit” property. That is a hard thing to pull off successfully in the time frame venture capital cares about.

Danger 4: A lack of compelling and convincing use cases. I have no doubt they are out there. If the IoT were working today at scale, we would have known within minutes where MH370 disappeared to in March of 2014. It is my personal opinion that most referenced IoT use cases are overlooking the tracking of high value assets that change position or quality rapidly over time in favor of the oft recycled “smart home” use cases. The former is compelling. The latter has been talked about for 20 years and still has no meaningful traction. The danger is that if IoT is talked about for too long before we see really compelling use cases, it will turn off the investment community.

The emergence of Next Big Things seems to be a long lived fact of life, and so they are a constant danger to aspiring entrepreneurs. To point out two examples, think of Artificial Intelligence in the late 1980s and early 1990s and Solar Energy in the early part of the last decade. Both areas destroyed at least hundreds of millions of venture capital money and produced very few winners.

Before committing several years of your life to a Next Big Thing, make sure you have thought through these dangers and have a clear plan to navigate around them.

Jeff Thermond is a Venture Partner at XSeed Capital and has been involved in Information Technology and computer networking for over 30 years. Prior to XSeed, Jeff was Chief Executive Officer of Woven Systems, Chief Executive Officer of Epigram (which he sold to Broadcom for $498M in 1999), and Vice President and General Manager at 3Com. He currently sits on the Board of Directors at Lex Machina.