Amazon’s hardware-as-a-service

The thing I admire the most about Amazon’s business model is the way they sell their core competencies to other companies. The internet allows humans to make something that is (almost) infinitely replicable and available to everyone, and Amazon has taken advantage of that fact to change the economics of multiple industries. This wasn’t possible in the same way before the internet. You couldn’t rent out capacity in your steel mill and continue to use your steel mill for your own work at the same time. Any capacity you lent to someone else took away from what you could produce. On the internet, Amazon can rent out any of their features to other companies who could benefit from it.

Amazon learned how to be an online retailer, and then sold their core function as a service, enabling others who wanted to be online retailers to sell in Amazon’s marketplace (even if they competed with Amazon’s products). They learned how to build scalable, efficient web services, and then sold their web services, enabling others to affordably develop on high quality platforms from the start (even if they competed with Amazon’s web offerings). They learned to get really good at shipping and fulfillment, and then sold those functions to other companies who needed to ship physical goods (even ones that don’t sell in the Amazon marketplace). Amazon’s ability to take over whole markets derives from their ability to simultaneously compete in a market and rent tools in that market. There’s an old saying that the best thing to do in a goldrush is to sell pickaxes, but Amazon can rent out their own pickaxes while simultaneously using them to mine. They have their cake and sell it too.

By itself, Amazon Web Services has had a major transformative effect on the tech sector. It has made it so cheap to start a software startup that it has changed the economics of Silicon Valley, tech startups, and venture capital in general. I just can’t understand why they’re not doing the same thing for hardware startups. (EDIT: It turns out they are doing it, sort of. See the bottom of this article for the update)

Amazon has proven that they have gained the expertise in creating an IoT device that customers will buy and use with the Echo and Echo Dot. They’ve now built the ability to envision a device, source and manufacture the electronics, integrate the hardware and software into a compelling user experience, market, sell, and ship. Granted, they already had some of these competencies, but they were able to leverage the existing expertise to build new ones (another feature of their core business model).

The new competencies Amazon has developed address the major challenges that hardware and IoT startups face. The internet of things has been just around the corner for years, but no one is sure why it can’t quite take off. I think a major contributor is high cost and difficulty of integrating across hardware and software. To build a successful device, you have to design and prototype custom circuit boards, connect input/output devices, build out software to control everything, get drivers to work right (if they exist for your application), and then re-design, re-prototype, and re-program when you find a problem. Creating computers from scratch is hard, and to get IoT to work, you have to build a custom computer to go inside of all of these little devices. On top of that, you have to create a delightful user experience and do it cheaply enough to compete with a non-smart device. And on top of all of that, you need to figure out how to fulfill and ship orders to consumers who expect to get everything almost instantly.

I think that this is too much for most startups to handle, especially when VC cash in Silicon Valley is accustomed to the cost structures of software startups where selling an additional unit has virtually zero marginal cost. In the past, I’ve criticized hardware companies in Silicon Valley for using software strategies. They try to start by selling at a tiny margin (or a loss) hoping that they can grow a large user base, leverage data, and work out the financial aspects of their business model later, but the difference in cost structure often leads VCs to lose patience with a company that burns money at a much faster rate than the software startups that they’re used to.

However, I think that I was wrong about part of it. These companies need to be software companies and hardware companies at the same time. They need an organization that can build ingenious devices that delight customers by solving hard technical and design challenges, but they also need the operational expertise to source, manufacture, fulfill, and ship. Some elements can be outsourced (you can hire Amazon to fulfill and ship), but there is currently no way to outsource the integration across the hardware and software development.

Enter Amazon’s hardware-as-a-service.

Let’s say you’re a product designer that wants to develop compelling projects. You can code, but you don’t know how to build computers from scratch, and you haven’t ever had to scale an electronics manufacturing operation. You just want to use hardware and software to make something great. You already use Amazon Web Services to develop your web apps, and you find out that Amazon is opening a new division called Amazon Hardware Services.

Amazon hardware services sells a small suite of generic circuit boards (think Raspberry Pi) that have processors, memory, storage, power, wifi, bluetooth, and input/outputs. Maybe they sell models that have different sizes, technical specs, or are geared toward different applications like audio, video, voice, etc. You can buy a couple of them to prototype on, or you can order large quantities as you scale up manufacturing.

Amazon can afford to sell the boards cheap because they make money on the developer tools that are fully integrated with the different hardware modules. They remove most of the complexity of integrating the hardware and the software so you’re free to just develop and prototype. Once you create something great, you have a clear view of what it will cost to scale it up on the electronics side, and can focus on the other aspects of the hardware (casings, buttons, and other physical pieces that any physical product needs). If you want, you can also easily integrate your supply chain with Amazon’s e-commerce, fulfillment, and shipping platform, to abstract away even more of the challenging aspects of starting a hardware company.

Strategic importance

Providing hardware developers with tools that are integrated with their hardware and software development environment could significantly lower the barriers to market entry and enable rapid, cheap development of smart products. It places Amazon in a perfect position to do to hardware what it did to software and unlock a wave of innovation in the space by enabling thousands of new entrants. These new entrants would compete with Amazon’s smart home products, but Amazon has always enabled its competitors by selling them the tools they need.

On the long-term strategic level, this move could have another big implication: The hardware makers that Amazon enables could grow to pose a threat to Apple. I’m in the camp that believes that a major source of strength for Apple is the fact that they are integrated across the hardware design and the software development. Because consumers put a premium on user experience, the firm that is integrated across hardware and software will have an advantage in developing new consumer devices. But what if Amazon comes along and makes it much simpler to be integrated across hardware and software? They could commoditize a core source of Apple’s strength.

Now, I don’t think that this would happen in the next year or two. It’s not like these new entrants that Amazon enables will be making products to compete with the iPhone. I don’t believe that any platform like I’m describing could suddenly provide the same advantages as Apple’s three-and-a-half decades of practice. Instead, the platform would follow the pattern of low-end disruption. It would start at the bottom of the market providing a few cheap, generic options for developers of products that aren’t too demanding. It would be just good enough to attract some of the most desperate companies that probably can’t exist without a solution like this (like the multitude of hardware company failures on Kickstarter).

Over time, however, Amazon’s hardware-as-a-service platform could offer more complex tools, a wider variety of compatible boards, sensors, and accessories. Sustaining innovations could make it simpler for developers to keep all these devices secure (which will be an increasingly difficult challenge for IoT developers), increase the energy efficiency of the hardware, get smaller, cheaper, and faster. They could create better and better tools to make development easier. Eventually the service could offer fully customizable circuit boards for highly specialized applications at lower and lower cost, shifting the dynamics of the whole circuit board manufacturing industry.

With all those improvements over time, it’s plausible that the tools could get good enough to threaten parts of Apple’s market. Even if they don’t, when more and more of our physical world is connected to the internet, it may matter much less which brand of phone or watch we use. In the same way that the internet browser made it almost irrelevant which brand of desktop computer you were using, the internet of things could make almost irrelevant what brand of pocket computer (phone) you are using. For a while, we’ll still have a central device that controls all the connected things around us (which we’ll probably wear on [or in] our bodies), but it’s not hard to imagine a future where all other the things we’re connected to have more and more impact on our lives than the rectangle in our pockets or on our wrists.

Whether or not any of these new entrants really threaten Apple, a platform that could simplify hardware development is sorely needed. The market needs a way to make hardware development as cheap as software development. I believe that enabling developers to make smart devices faster and cheaper could have huge impacts on the economy and society, and I think Amazon is in a unique position to be able to make it happen.

Update

Just a few minutes after I posted this, Alexander Wissner-Gross pointed me to Amazon’s “Alexa Development Kits”. This is super interesting. It allows product developers to buy kits to prototype products that use the Alexa Voice Service. Rather than renting out the ability to easily develop general IoT devices, the kit helps developers create new Alexa-based products.

My initial reaction is that this is far too narrow an application to create an IoT boom. Limiting the toolset to only Alexa-based products means that Amazon won’t be able to take as large a share as they could in selling tools to the IoT developer market. My assumption is that there will be lots of smart devices that don’t use voice, and lots of developers who are uninterested in only working on Alexa products. It would be like limiting to AWS to tools for building e-commerce sites. However, there could be a couple of reasons that they are doing this.

Reason #1: They want to boost Alexa over other IoT products

Maybe Amazon believes that most IoT is going to be voice controlled, and they want to push Alexa to rule the space. Inviting other developers to build Alexa products could be a means to strengthen it. The key reason for focusing on this could be that Amazon believes that the data they’re collecting from Alexa is critical enough in enhancing their AI capabilities that it’s worth missing out on other opportunities in order to focus on the race for AI against Google, Facebook, and Apple. If the company with the best AI wins everything in the long-long-term, and Amazon thinks it’s the time to focus on that race, then this could be the time to limit how they enable new entrants to create new products, and only create tools that bolster Alexa. This does, however, run the risk of allowing someone else to develop the tools to make other IoT devices and become the dominant player in the whole IoT space while Amazon’s waiting around to grow AI.

Reason #2: Alexa Developer Kits are a beachhead from which Amazon will launch a broader platform

I hope this is the reason. These kits could just be step one of a plan like the one I suggested in the article. Rather than start with a suite of generic IoT tools like I mentioned, Amazon could be starting with a more limited offering that is focused on Alexa. I said that they had just learned how to make a successful IoT device and they should sell that expertise, but maybe they’re just starting even smaller. In truth, they didn’t learn how to make different kids of IoT devices, they just learned how to make Alexa devices, so they’re selling the tools to make them. Then, as they learn to make other devices, they could sell those tools as well. This strategy allows them to learn more about what developers want and then gradually give them more until it’s a full suite of hardware and software tools that enable all kinds of IoT devices.

Reason #3: Amazon hasn’t recognized the opportunity in this space

It’s always possible they just haven’t thought through all these things yet, but I’m hesitant to assume I have unique insight when I didn’t even know these developer kits existed until today.