Is Hardware Still Hard?

Hardware is hard.

If you’ve been working in hardware design or engineering for any amount of time, you’ve probably heard this a lot. If you haven’t yet, you will.

It’s a phrase that is a versatile as it is common, used to warn would-be hardware adventurers, deflect missed deadlines, and shore up one’s own self-importance as a Hardware Person that does Hard Things.

It’s also an aphorism that has been repeated so much and treated as so self-evident that it’s almost lost all meaning, kind of like if you say the word “knee” too much and start to forget what it means. I myself hadn’t thought too hard about what it actually means until I got the idea for this piece and started doing research.

So what do we actually mean when we say hardware is hard?

What’s hard about it? Is it getting harder or easier?

Should we all just pivot to working on software and presumably have an easy ride?

What’s hard about hardware?

Let’s take a look at a typical Internet-of-Things startup today.

While a few software engineers can put together a basic MVP and start selling to users very quickly, hardware tends to require a broad mix of technical experts. This includes designers, mechanical and electrical engineers, antenna specialists, manufacturing experts, and so on. It’s almost impossible to find all this in a small group of founders. So you’ll probably have to hire outside experts.

It takes a lot of capital to deliver a hardware product that is ready to sell. You have to pay for design, development (including fees for all the aforementioned experts), tooling, inventory, shipping, certification, among many other things before you even have a thing to take to market. There are now some shortcuts and ways to de-risk this using crowdfunding, but at the end of the day, you (or your investors) are putting a lot of money on the line before you get any real money coming in.

After all this trouble, your profit margins on hardware also tend to be lower than software products because of all the money you’ve had to invest in making and getting the product in the hands of users.

Not only does developing hardware cost a lot in money, but also in time. It can take 1-2 years to get your first units off the manufacturing line. Feedback loops are long - you won’t really know what the mass market response to your product is until it ships. Mistakes are costly - you can’t ship a patch to a hardware product the day after its release. Supply chain and material delays can throw a giant wrench in your carefully planned schedule.

All the while, your company isn’t making any money (maybe some if you got pre-orders) and there’s a lot of money going out. You’re watching your runway get shorter and shorter and pre-order customers get more and more irate as you miss predicted ship dates.

😰😰😰

That’s why they say, hardware is hard.

Hardware got less hard

After all that, you might be surprised to learn that conventional wisdom in the industry is that hardware actually got less hard over the past few decades.

For one, components like chips and batteries are getting more powerful and cheaper every year, enabling your gadget to do more and cost less.

Additionally, the boom in hardware and IoT development in the last two decades has driven up demand for readily available components, and suppliers are meeting it with all manner of stock components: displays, batteries, microcontrollers, and all manner of sensors.

This has even translated down to the hobbyist hacker level. Take a look on Adafruit and check out all the intricate components that you can buy to integrate with that light-up halloween costume you’re making.

Many commentators also point to widely available, pre-certified radios as a boon for hardware companies, making it easier to clear the technical hurdles and pass FCC certification.

So from a purely nuts-and-bolts (or chips-and-components) standpoint, hardware is now easier than ever. It’s more about putting together the right puzzle pieces and handling development and scaling in the right order rather than doing deep R&D into foundational technology.

There’s also a huge supporting ecosystem that has developed around the industry.

Ben Einstein, founder of Bolt, a venture capital firm that invests at the intersection of physical and digital, describes how since the 90s technical complexity in software development has been largely abstracted away by services (like Amazon Web Services) that popped up to serve the growing demand of the digital economy.

This means it became easy to piggyback off all these existing services and piece together a new piece of software that you could start selling with a very small team and much less technical know-how and startup capital. This led to the rich landscape of SaaS products we see today.

He points to the same thing happening in hardware, with stock components, pre-made wireless modules, contract manufacturers, outsourcing processing to smartphones, and the ease of marketing across social media taking a lot of complexity off would-be hardware founders plates.

This technical abstraction has similarly led to a hardware boom. While there have been some very public hardware company failures, we now live surrounded by smart gadgets and it’s hard not to agree that we’re living squarely in the age of the Internet-of-Things (“ALEXA. TURN ON THE LIVING ROOM LIGHTS”).

The hardware business is still hard

So from a purely technical standpoint, hardware is actually less hard than it used to be.

What remains hard is the business of hardware.

As mentioned above, hardware by its nature includes challenges around capital requirements, cash flow, supply chain and inventory management, and costly mistakes that are hard to correct.

These factors haven’t changed much and in some cases, like supply chain woes due to COVID, have gotten worse. This leads to the relative unpopularity and lower survival rate of hardware startups.

Software companies have issues of their own - high competition in crowded markets due to low barrier to entry, can be hard to monetize, over-reliance on the platforms they’re built on. But I think the main difference between hardware and software is the number of shots you can take at getting it right.

A software company can constantly user test their product, collect granular data, make changes, even ship two different versions of the same product to different customers and gauge the response. Mistakes can be corrected overnight, and the company lives on to fight another day.

If you are a hardware business, any mistake (e.g. committing to manufacturing 25,000 units of a design that has flaws) or even factors outside your control (say, shipments held up in ports that are closed or sudden unavailability of one of your core components) can be a company killer.

Feedback loops with the mass markets are long, and any course correction is painfully slow. Unless you are extremely well funded, it’s likely that any one of these things that may or may not be within your control can take your runway to 0 very quickly.

So it’s easier than ever to design and build that flying hoverboard you’ve been thinking about. But you better be confident people are going to buy it and love it and plan a big buffer for setbacks (or just pray for none, likely in vain).

Hardware is just hardware

For me, there is nothing quite like the feeling working on physical products - iterating on prototypes until they’re just right, holding something in your hand that was once a sketch on a scrap of paper, and walking into someone’s home to find an object you designed.

Dare I say even the delayed gratification and long feedback loops make the successes more sweet.

And for all its downsides, hardware does also have a lot of upside.

If and when you do make it through the gauntlet of developing and shipping the first batch of product, things get better. You’ll work through the kinks in your manufacturing and supply chain, and the cost of your product can go down over time, increasing the profit margin.

After you do the hard work up front, it can actually get easier. Ben Einstein (Bolt founder) says, “Shipping hardware to your first customer can take 18–24 months, but selling to the next many thousand goes quite quickly. Fitbit scaled from $0 to $1.9B in revenue in 7 years. Very few SaaS companies scale that quickly.”

If your hardware product product is done well and users bite, it can actually be stickier and more profitable than software. Most SaaS software these days sell at a relatively low monthly subscription. The lifetime value of a customer isn’t that high, and they make up for it with volume.

For a product like a laptop, each sale can be in the thousands of dollars, and if the customer ends up investing in the ecosystem, it’s likely they’ll buy another one in a few years. Same goes for smart home products and perhaps even non-tech products like furniture. If you get it right, you’ll be the go-to brand for the customer’s need in that category. SaaS customers are usually more fickle, and switching costs tend to be very low.

Additionally, because it was so hard for you to get to the point where you’re actually shipping product, it becomes very hard for others to imitate you. Yes, there will be copycats, but if you’re differentiating on quality and user experience, it will be hard for them to match. Your difficulties have now become your advantage.

In some ways, “hardware is hard” is a bit of a meaningless statement. People do a lot of hard things. Like exercise. You like to run? Running is hard. I hate running. So I lift weights. It just depends on what challenges you enjoy and fuel your creativity.

If you talk to a SaaS founder, they’ll tell you its hard. If you talk to an architect, they’ll tell you its hard. If you talk to a person trying to start a laundry business, they’ll tell its hard. It’s just hard to make new things of any kind.

And we do need to make new things. Many of our things are old, outdated, inefficient, and based on a model of going from the factory to the landfill with a brief stop in your house. There are new technologies around the corner that need beautiful, user-centric experiences designed around them.

Ben Einstein again: “If we want to live in a world of autonomous cars, consumer space travel, and a green energy grid (I know I do) we have to stop defining hardware by its difficulties.”

💯

Yes, hardware is hard. But its no harder than anything else.

It’s just hard in it’s own unique way.

So to use another meaningless aphorism, “it is what it is”.