Since 2000, it would seem that software companies have been the only technology startups disrupting markets, that is until recently in the last few years with a small uprising of hardware startups. The hardware industry has started to gain momentum thanks to success stories such as Nest, Dropcam and Oculus being acquired. Valued together, these companies have a total worth of more than $10 billion. Despite these laudable gains, recent events have nonetheless reconfirmed the risk and difficulty associated with sustaining hardware startups. Recent events such as GoPro’s layoffs, and Pebble, Lily and Skully shutting down operations further substantiate the trend of corporations dominating hardware, and startups focusing on software.
This reality becomes even more evident in new markets such as Virtual Reality/Augmented Reality, Robotics and AI. Zooming into VR/AR, prominent companies include Sony, Samsung, HTC, Facebook and Microsoft. Given the fact that these companies have more resources such as capital, people, manufacturing and even distribution, their prominence makes sense. If their initial product version does not become a hit or the market takes longer to rise, these companies nonetheless have the endurance to continue beyond just a couple of years.
How can any startup compete when the stakes are so high and chances so low?
Like many entrepreneurs, I was optimistic when I decided to take my chances at building a VR camera startup from the ground up. Fast forward several years, and looking back, I can definitely say that the challenges from lack of capital to limitations with manufacturing made the experience seem all the more insurmountable. Yet the approach that my team and I took along our journey not only helped my company to launch, but also allowed us to continue advancing in spite of the odds stacked against building a successful hardware startup in VR.
Capital, capital, and more capital
Software startups build multiple prototype versions, improve on features and gain customers and traction with a couple hundred thousand dollars and three developers in a garage. Hardware startups, however, have to raise millions just to manufacture a small batch of their first version device. And as promising as crowdfunding sounds in 2017, the more units a startup sells, the more likely it will fail to deliver. Why? The odds of failure increase because the investment needed to ramp up manufacturing, inventory, headcount and distribution is not covered by the backers.
Non-stop fundraising and building products for a hot market such as VR/AR becomes almost inevitable. Yet even with the initial success of a product, a VR/AR startup is still confronted with the reality that 90 percent of investors do no invest in hardware or want to wait until the sector actually will take off.
In order to survive these circumstances, we developed more effective ways of running a hardware startup. First, we flipped every penny twice. Every expense, even if just a couple of dollars, was based on thorough calculations with an evaluation of the opportunity costs and return on investment. Second, we were extremely disciplined and designed metrics to assess our activities. As a result, marketing activities required traction goals and grading to be repeated. We shifted from pumping thousands of dollars into social media to word of mouth campaigns, since on-the-field testing by having individuals experience VR in-person was a much more powerful and cost efficient form of marketing.
Finally we developed methods for generating revenue beyond just hardware, from licensing our software to white labeling our product to building software services into the offering. Through this painstaking process, we learned that even a hardware startup can make money without its hardware, which completely changed the way we looked at our business and allowed us to build a product from nothing to shipping out with less than $1 million.
No grey hair, no hardware
We have seen many successful software startups built by 20 year-olds in the past, but rarely do we see the same for hardware startups. The advantage of quick pivots and improvements based on customer feedback at almost no cost allows developers with minimum experience to learn from their mistakes and become better within seconds. However, the quality control for hardware undergoes a more complicated process. It is not uncommon for hardware engineers to design a product, send it to Asia for assembly, and after a few weeks, receive a prototype which does not even work. Not to mention, the error will cost the hardware startup thousands of dollars.
And that is just the beginning. When moving towards manufacturing, the cost of tooling, assembly, and inventory can carry a price tag in the tens of thousands of dollars. Also, even the smallest mistake can ruin your investment. Therefore, people in the hardware business often have a decade’s worth of experience, at least, and even then, these experienced professionals still run into unforeseen challenges. Ramping up manufacturing for the exact same product twice might have completely different outcomes, even though it’s the same product and same process, for example.
We were a young team with little experience in hardware startups, but three things helped us overcome what we lacked in years of experience. Due to the complexity of manufacturing hardware and the number of steps needed, attention to detail is key. Moving too fast on a product and feature level will often lead to omitting the actual quality of the result. In addition, team members can easily miss the problem because they have been working on a given product for years.
To maintain our consistency in quality and attention to detail, we turned to social media, focus groups and backers to test our camera output. We realized the feedback from these resources early in our development cycle would make a huge difference in the extent to which our product would succeed. We were also “paranoid” about making mistakes, which worked to our advantage in drastically reducing the number of technical errors. For example, even if a product worked during the first or second test, we continued to test an additional 10 to 100 times in order to ensure that the hardware worked properly.
Because of the lack of time available to test thousands of times, we also regularly assessed the consequences a potential error may have to recognize the risk which we were taking to release that product, and how to best solve the problem before it occurred.
Questioning the status quo
As software development has evolved significantly over decades, hardware building and manufacturing has been practiced for centuries to a degree of perfection, and art. The Toyota production systems and lean manufacturing approach in the 1950s, which allowed Toyota to legitimately compete against American and European automobile manufacturers, represent one example of the evolutionary path that hardware building has traveled.
Unfortunately, the developments within this industry have left hardware manufacturers wary of change, and committed to the status quo. This status quo has been reinforced by those within the industry who prefer to run a well-oiled machine rather than innovate and develop entirely new processes, as their key performance indicators have been cost reduction and productivity. Imagine the difficulty for a hardware startup to convince contract manufacturers to build a completely new product with a completely new process—the attempt would be a hassle for both parties.
The solution we developed was based on the understanding of our own strengths and weaknesses. At the end of the day, we had only three engineers in our team. We decided early on that we would focus on our strengths, and work with others who excelled in areas where we were not as strong. Therefore, we partnered with one of the top 5 ODMs (Original Design Manufacturers) in Taiwan to build the product together. In this way, we could focus on our expertise in real-time 3D imaging and computer vision, while the Taiwanese ODM helped us manufacture the product.
This partnership was beneficial in many ways. In addition to what we learned over the course of building a hardware startup, we gained additional insight into the processes and limitations of hardware manufacturing. We also contributed to the partnership with our questions and ideas, which, in turn, impacted the way in which this ODM operated.
If we were told something was impossible, naturally, we questioned it and worked to find solutions and ways to work around the supposed impossibility. And this ODM kept us on schedule, providing information about trade-offs to make the right decisions to move forward. The partnership was a team effort, and required that both parties be willing to work together to reach the intended goal. It also required that our small engineering team of three consistently outperform itself. Given the sheer sweat equity and attained metrics, we worked as if we were a team eight times our size. The Spartans would be proud.
Hardware is not impossible
In retrospect, many things fell in place because we continuously challenged the traditional ways of running a hardware startup by running our company like a lean software startup. While capital, years of experience and excellent production processes are the core to hardware, I believe that with the new generation of software companies doing more hardware such as Facebook with Oculus and Snapchat with Spectacles, the past will soon be disrupted by innovative approaches of launching hardware. A perfect way of building and scaling a hardware startup may not exist, but as long as you are willing to push the boundaries and see beyond what is possible, you will always find better approaches to succeed in this industry.