The Hard Things About Hardware

If there's one thing I've learned from decades building technology in the wireless industry, it's that hardware is…hard. This is, of course, common knowledge or at least commonly assumed. Indeed, much of Silicon Valley has turned away and said collectively, “Why bother? Just do software!”

Why bother after all? Because dedicated, purpose-built hardware is the only way to achieve the highest performance.?

There is a reason that every type of racing car looks radically different from every other type:? top fuel drag racers look different from solar cars, which look different from Formula 1 racers, etc.. And this physical reality is no different in any other type of engineering.? But wait, you object: the world of electronics is different!? Everyone knows that the best thing is to just build a computational engine that can do everything, and then customize it in software!?

And two things are true about this:? 1) that has been the right answer for much of the history of computation;? 2) the reason it has been the right answer is that energy usage, wasted heat, size and weight have not been as critical as they are today. But everywhere you look now, you see NVIDIA's GPUs taking over the world. What is the secret?? Their purpose-built computational hardware enables vastly higher performance than general purpose processors.

Fine. You find yourself in the difficult position of being an entrepreneur whose problem demands a hardware solution. And as every venture capitalist who has run away screaming from your project has told you, “It’s Going to Be Hard.”? Having heard the phrase from countless investors and having spent a career in hardware, I wanted to get concrete about what those hard things are.

Here are the hard things about hardware that as an executive you need to understand, and more importantly, overcome.

1. The Proof-of-Concept or Prototype Trap

With hardware, as with any other type of engineering, it is often necessary and prudent to take baby steps with an ugly prototype to convince yourself that you have something that works. When I left the MIT faculty and started Eta Devices, my co-founders and I knew from our prototypes that we had something special.? However, once we started the company we were utterly blindsided by a real-world performance requirement that, for an awful few months, looked like it would undermine our whole performance advantage. It is amazing to look back at how the team pulled together, solved the problems, and went on to build the highest efficiency wireless transmitters in the world.?

Recommendation: You want to get out of prototype mode and into product design mode as fast as possible.? Note that this particular gem applies to software and hardware equally.? But on software projects, stakeholders don’t take it as hard when a major change in direction needs to be made. (This is largely because people do not always give proper credit to software engineers for the size and complexity of the projects they tackle.) It may be true that it is more expensive to change directions on a hardware project, and it is certainly psychologically overwhelming. As the leader of your team prepare them for this possibility, and budget time and funds to accommodate for surprises. And then it bears repeating: stop designing prototypes and start designing products as soon as you possibly can.

2. Hardware Makes People Afraid to Improve a Successful Product

Think BlackBerry or Tesla. Blackberry rested on its laurels, only to watch its empire crumble in the face of rapid evolution. Nokia had touch screens for their phones before Apple did,? yet that story ended…poorly, for Nokia. The entire car industry sat on the sidelines while Tesla emerged and blew them out of the water. Now everyone is playing catch up. This movie plays itself out over and over again: someone gets a product right, clamps down and leaves innovation aside, only to be suddenly surprised by a new entrant.

What’s going on?? Are these people stupid?? No. However, with hardware products, the up front cost to change your products or your approach is big and loud and in your face, perhaps even to the point of significantly diminishing your short term profits. And there is no surer way to cause a revolt among public shareholders than to lower short term profits. Note again that big software structures are not easy to change either, but the (often enormous) costs of changing tend to be less explicit.?

Recommendation: This is a big one, and obviously easier said than done. As a company, you must not be afraid to improve your hardware products. It occurs to me as I write this that maybe this is impossible for most large public companies. I don’t know.? The glib “Fund the Rebels!” advice coming from business schools ignores the gigantic, real political challenges to overcome to put this into practice.? But they must be overcome. Otherwise, a Tesla-like barbarian storms your gates and you live forever as a Harvard business case cautionary tale.

3. The "Lowest Cost" Trap

Remember the Samsung Galaxy Note 7 recall of over 1 million devices from devices catching on fire? Samsung paid a price far greater than the initial savings on a cheaper component.

It is easy (and fun!) to dunk on Samsung here, but in fairness when you are shipping billions of units every tenth of a penny in the bill of materials does mean big numbers. And customers expect that the price of hardware products will go down over time. (This is a legitimate and confounding difference between hardware and software products, where with the latter consumers expect to pay rising rents more or less in perpetuity.) It is true that the cost pressures become enormous with hardware products.?

Recommendation: Prioritize quality alongside maximal cost savings. Consider the longer-term costs of potential recalls or reputation damage. On the positive side, consider the benefits of being the provider of the highest quality solution.

4. Underestimating the Software-Hardware Tango

It is easy for hardware focused people to treat the software and the user interface as an afterthought.? Who has time to worry about the GUI when we’re busy here wrestling with Nature’s Laws?! But your customer does not know and does not care about the physics. If your software, through which they experience your product, is buggy and slow, your product is buggy and slow. The result: poor sales and a tarnished reputation. Also, it is astounding what is possible in a system when the software and hardware are truly made for each other. As a hardware person, I’ve had to be especially careful to take good advice and work with strong software teams, so that the overall product can be 10x better.

Recommendation: Ensure tight collaboration between hardware and software teams. One without the other is an incomplete symphony.

5. Customer Use: Their expectations are not completely under your control

An easy example of failing to set customer expectations is designing a 'water-resistant' gadget, which customers interpret as 'waterproof.' You can guess the chaotic aftermath. Technically, you were correct and honest. But a disappointed customer is still your problem.

Recommendation: Satisfying customers is an art and a science. Respect it accordingly.

6. Overlooking the Power of Aesthetics

It is impossible to overstate the importance of this point. Think about Apple. Their success isn't solely about functionality but also design elegance. Hardware people forget this point easily, because many times just getting the thing to work is a herculean task.? Who cares how it looks, right? Look at the triumph we achieved of getting it to meet spec!

At TalkingHeads Wireless, we had the pleasure of a site visit with a major European cellular carrier. When we asked the installers which equipment they liked best, they sided against every government in North America and Europe, smiled brightly, and said, “Huawei.”? The ergonomically designed, well conceived hardware was easy and a pleasure to install, and the warmth from the guys on the ground suffused widely through the organization.

Recommendation: Invest in design. Aesthetics with a purpose. Remember people don't just buy functions; they buy experiences.

7. Your Customer Cannot Tell You What They Need

I’ll never forget my own reaction when Apple’s iPad was described to me:? “Why do I need that?? I have a Macbook and an iPhone, and you’re going to offer me now something in between?” And then when I had one in my hands, it made sense.

Your customer cannot normally tell you what they need, and telling you is not your customer’s job. If you ask your customer, what they will usually say is, “Give me what I already have, only cheaper and with more features.” A great product designer will listen to the customer, observe the customer, and dream up a solution to the customer’s problem whether they can articulate it or not.

This is a great challenge for all product types, not just hardware. But of course with hardware the stakes feel higher, because if you get it wrong it is not “merely” a change in code to make it right.

Recommendation: Diversifying feedback sources helps. And remember that sometimes, the loudest voices aren't representative of the majority. Learn from feedback but have informed product conviction. There are no hard and fast rules to lean on, here. Product design is a legitimate superpower, and there are people that are good at it. Find those people.

?

8. Scaling Production Isn't Just Multiplication

I’m writing this in the aftermath of a brilliant visit to a high volume manufacturer near Detroit, Michigan.? Here I met people that you don’t often meet in academia or even in the startup space, people who are passionate and brilliant about creating processes that produce millions or billions of hardware units with yields of 95% and more. It was thrilling, humbling, and eye opening to see.

The biggest lesson that I took away is that you do not achieve these yields by putting together an improvised prototype, and then figuring out later how you make things for manufacturability.? And…making things for manufacturability is the only thing that matters, especially if your ambition is to have a global impact. These folks showed me that the earlier you can have true manufacturing expertise involved in your hardware designs, the better your chances are for success.? Again, I would say that hardware and software are not so different in this regard, they just seem different on the surface.

Recommendation: Get serious manufacturing expertise on your side early, and give them a voice in the design process.

As a leader in a hardware focused industry, your role isn't just to make decisions but to anticipate and navigate challenges, especially the hard ones. The path is littered with the siren calls of cost-saving, the deceptive allure of mere prototypes, and, above all, simplistic calls to Listen to the Customer. Yet, as history has shown, those who understand these challenges and anticipate their intricacies not only survive but thrive. By understanding the hard things about hardware, you can not only steer your product and company to success but also redefine the very fabric of the industry.

What do you think are the hardest things about hardware?