Fifteen years ago, my company, Danger, spent $240 million to develop a production-ready smartphone. Today, a similar effort would cost just over $3 million.

That’s within budget for an A-round startup, and it’s one of the reasons the pendulum of innovation is shifting from software back to hardware or, more specifically, hardware that uses software to do interesting things.

Historically, teams I've worked with have spent an enormous amount of time and energy building the software stacks that enabled their hardware ideas. From operating systems to cloud services, these software stacks were often coded from scratch specifically to power the devices coming from companies such as General Magic, WebTV, Danger, etc.

Inevitably—software schedules being what they are—the hardware was often ready before the software, and the product would have to wait at the factory until the software was ready to be flashed into ROM. Only then could we ship the final product to customers. I've seen unspeakable tricks to streamline this process. And some of those tricks create horrendous user experiences. I know companies that ship products into the retail channel with unfinished software. They count on over-the-air updates to deliver the final software just in time—as the user unboxes his new purchase, only to power it on to a long and tedious update process.

Open source software solves some of these complexities and unknowns

The sea change began when open software went mainstream. Gone are the days when proprietary software was re-invented each time an OEM built a new product. These days, software developers can take many core functions from vast libraries of established open-source code and combine it with their magical ideas to create something entirely new. And they can focus the majority of their time and talent on the magical new thing.

Open software is battle-tested. Chances are, your team isn't the first team to ship the code you inherit from open source. Android is based on Linux which was shipping in a wide variety of consumer products way before Android was even conceived. It gave us incredible comfort knowing that we didn't have to worry about process scheduling, memory management, file systems, etc. We could focus our talents on what mattered at the time: building a great phone.

Now, Android is powering over 1.4 billion devices, which means Linux is powering many, many more—the scale is astounding. You'd be hard pressed today to find a digital consumer product that doesn't include some piece of open-source software.

At Playground, we think this opens up interesting new market opportunities. Entrepreneurs will leverage the modular nature of combining the right amount of open software with their ideas and technology to build all kinds of fun and useful devices. And, in the process, they’ll lay the groundwork for what we believe will be the next major computing wave.

Many things about hardware are still, for lack of a better word, hard. The proverbial two guys working on a device in a garage need to figure out everything from how they’ll power it to how they’ll package it. To turn it into a product, they need to line up manufacturers, retailers and much more.

But a maturing hardware supply chain is reducing the number of components a startup needs to build from scratch. At Danger, we built custom LCDs, as well as our own operating system and RF front-ends to enable our phones to run on different carriers’ networks. Today, Sharp or JDI will sell you a display, MediaTek or Qualcomm will supply you with SoC that include all the peripheral devices needed to build a phone. Want an operating system? You can use Android for free.

Hardware makers 10 years ago struggled to synchronize their work with that of software developers. Unless their releases were perfectly coordinated, one would always feel out of date.

Not only is it easier to write software for devices now, but with cloud-enabled software updates, hardware makers can better focus on getting the product right. With 3D printing, they can quickly produce prototypes. And given the sophistication of manufacturers in China, they can accurately and efficiently produce small runs of products without much lead time.

These things can still trip up an entrepreneur. But there’s a method to it.

The case for Playground

When we started Playground, we asked ourselves: What could we do to make it even easier to bring innovative hardware to market?

It starts with capital, of course. We’ve raised $300 million to invest in hardware-focused startups.

It continues with space and tools. Playground’s studio is large enough to host 30 startups. We have thermoplastic and multimaterial printers, laser cutters, CNC machines, Metal sintering 3D printers, spectrum analyzers, network test equipment, RF chambers, Optics labs and pretty much any piece of equipment a company making stuff could need.

Most consumer hardware startups hand a portion of their first check to outsourced industrial-design houses. We have award winning in-house industrial designers who work with Playground companies for free, sitting a couple of desks away from your team. We find that this close proximity and collaboration makes design iteration a lot faster, with better results. We also have electrical and mechanical engineers, optics designers, program managers, product designers, software developers, and other specialists whose only job is to apply their expertise to your products.

We have an experienced operations team whose job it is to smoothly transfer designs to our manufacturing partners. We have people in China to babysit the production process. We also have relationships with retailers, and we hope to sell every product developed at Playground in our online store. (At first it will just be T-shirts and coffee mugs, but we have broader ambitions.)

The next wave

So what’s it all building toward?

Every 10 to 15 years, the dominant computing platform shifts. We’ve gone from mainframes to minicomputers to PCs, then to the Internet, and we’re now smack in the middle of the mobile era.

When we ask ourselves what’s next, we see devices that sense, compute, and act on information, connected to a cloud that has mature deep-learning capabilities.

For example, your security system might disarm itself when you get home from work, or your Nest thermostat could adjust the temperature based on whether or not you're home. You can do these things today, but they require using an app or some other human trigger.

In the future, devices relying on the cloud’s ability to interpret data and recognize patterns may be able to determine these things on their own.

It might take 20 years for this vision to come true, but we’re starting down the path today. We’re training the deep-learning algorithms every time we search Google, upload a photo, or use a speech recognition program.

Getting the rest of the way there will require hundreds of millions of data points—more data than humans at keyboards can possibly provide. The various devices you interact with daily will free the cloud from it's proverbial "brain in a jar" jail and allow it to interact with us in our environment—our homes, cars and offices. And it will give those devices, in turn, more purpose.

So we’re aiming big and thinking decades into the future. In the meantime, we’re going to make some truly great things.