For years, hardware makers have eyed the attention and valuation multiples enjoyed by software companies with envy, while employees at hardware companies similarly eyed all the fancy perks their peers were enjoying at software companies, while the hardware teams were lucky to get coffee at work. Software may be eating the world, but does that have to mean only software companies get foosball tables at work?

In recent years, many of the have-not companies have increasingly spoken about their goals of attach software revenue streams to their hardware. This ranges from the reviled, like BMW’s heated seats-as-a-service, to Nvidia’s cloud dreams with its Omniverse.

We were reminded us this recently by a Techcrunch article about car-maker Hyundai’s plans for software profits. Full credit to the Hyundai team, they have some good ideas about ways to squeeze a few software dollars out of their customers. Their vision is to sell ways to customize the interior of the car like sounds and sports team-themed displays. This sounds a lot like how game makers monetize free games – selling character skins and other decorative items. Hyundai hopes to reach $15 per month per customer in such subscription revenue. Hyundai sold almost 4 million cars last year, which works out to $700 million in extra high-margin revenue. Not a bad goal, but set against the $7 billion in profits the company earned, it is not that dramatic.

Hardware companies basically have three ways to sell software:

1. Sell subscriptions for nice-to-have but nonessential features, like custom skins for in-car displays or heated seats
2. Build a software adjacency to chip functionality
3. Sell standalone software

In the context of semis companies all of these are likely to prove challenging.

To date, no semi company we can think of us has created a software business from scratch. They all do a lot of software work, but that is not the same as selling software. A few chip companies have acquired software companies, but those do not provide shining examples for others. Broadcom has now acquired a few software companies, but they are a private equity fund with a lot of semis companies in their portfolio. Intel too has acquired several software companies over the years – McAfee and Wind River, to name two – and mostly ended up exiting them for less than they paid.

The other two situations are equally challenging. Adding additional circuitry to a chip for lowly used features is already something chip companies do. Loosely speaking, this is usually called binning. For instance, a CPU with 50 cores, can be sold at a lower price with 4 cores turned off; or vice versa a chip with 46 working cores that can be upgraded to 50 cores with a firmware update. Either way, it is one chip. The trouble is that features require area on the chip, and that comes at added cost, so it is not always easy to design this in. Regardless, this is not really a software service except insomuch as it is marketed that way.

Alternatively, chip companies can add some kind of add-on side service that is somehow tied to circuits on a chip. This is essentially what Nvidia is trying to achieve with Omniverse which is a cloud GPU service that has all the Nvidia features turned on 100%. But Nvidia is an outlier in that they have a dominant position in AI training right now, which gives them sufficient market power to pull it off.

That being said, this model is not totally outlandish. We think it is something that start-up chip companies should explore. As we have noted previously, the big hurdle to this model is selling it to customers. Most chip companies lack a software sales muscle. Start-ups can build that capability in from day one, which is much easier than bolting on salesforce to a company with thousands of employees already. Even Nvidia, with all its market control, is relying on Microsoft and Google to really sell this service to end customers.

Which gets to the real problem – there is only so much market pie to go around. If Nvidia builds a Cloud, that risks competing with its biggest customers. Chips are built to run software, and someone else is going to design, build and sell that software. Adding a chip vendor into that mix adds cost to the whole system, and this kind of margin stacking ultimately eats into demand. Any company looking to sell software and chips, has to contend with the fact that someone else wants that software business too, and they may be much better at providing it.

At the end of the day, everyone just wants recurring revenue, but that is not always going to be available. These markets are already incredibly competitive. Per the Hyundai example above, that company is going to spend a lot of money building a “software” business that ultimately is only going to gin up profits by less than 10%. For many, probably most, companies, selling chips or hardware is already a pretty good business.