FWIW when I was in business school, one professor strongly argued (based on an Aswath Damodaran blog/video - he is a business school "celebrity") that R&D should generally be considered a capital expense in terms of valuing companies even when it can be considered an operational loss on an income statement. Technically, this (likely catastrophic) accounting change may be kind of correct in a pure sense.
However, the "asset" associated with software development seems pretty uniquely hard to transfer. Almost every other intangible asset has a strong marketable component (eg a patent on a widget, a copyright on a song), whereas software just doesn't have one. Licensing doesn't count here - that's not a transfer of the asset.
> However, the "asset" associated with software development seems pretty uniquely hard to transfer
Isn't it what's happening when a company gets acquired? Buyer pays for a software and a team (more for the latter usually, including business process as a whole), not for laptops or servers. It's not "equal" to salaries but you can't create the former without the latter. So in a sense when you pay salaries your create the above-mentioned "asset".
Yes, but that's basically the only mechanism for this kind of IP transfer. Other types of IP are a lot more transferable and don't have to be tied to a team. Think about book/song copyrights or widget patents.
The rights to software get transferred all the time though. It doesn't have to be highly marketable to be an asset.
For sure, lots of software is pretty tied up with the day to day operations of a particular company, but so are things like a specialized manufacturing line or whatever.
Not really trying to argue how that should impact the accounting, just arguing the specific point.
Not disagreeing with you at all, but I want to suggest that we look at it another way - which is that accounting principles as a whole are really not built for technology companies, and we are straining them right now to fit ourselves in.
When you build a manufacturing line, you usually buy your equipment off-the-shelf (capex, with good resale value) and hook it up in a semi-custom way with cheap, movable conveyors (also capex, with resale value). This means that you can make a "stuff" company with practically no R&D. In that world, you only do significant R&D of any kind once you are big enough that your chance of going bankrupt before the end of even a 15-year amortization period is pretty small.
Even in those companies, the work product of R&D groups is often relatively modularizable and transferable. For example, your R&D group may modify a machine to produce coke cans 10% faster. That improvement is likely transferable to pepsi. You may also patent the improvement, but that's not required for it to transfer well. Your R&D can also be hard to transfer, like specific factory layouts to make $widget that only you would ever want to make, but you're only doing that research when you are an established manufacturer, which makes the depreciation not a big deal (you probably prefer to capitalize that expense anyway).
The fact that the work product of R&D is somewhat transferable at a price near how much it cost to produce is what allows you to get loans backed by R&D work (technologies, patents). Without the secondary market value, you don't really have an asset.
In contrast, companies whose primary work product is technology (software companies and some digital hardware companies) do significant R&D from day one, both before proving the value of the R&D work and in a way that it usually isn't sellable or modularizable.
Like the coke cans, there's probably a market of people who would buy the product "patch that makes redis 10% faster," but the tech transfer mechanisms we have today don't support that market well (if at all), and most startups don't waste time improving redis. There is almost certainly no market that would buy "efficient database schema for a SaaS for gravediggers," which actually is what occupies your time if you're building a SaaS for gravediggers.
That suggests, using the "capitalized R&D" and "software development is R&D" ideas, that the first few years of a software company will pretty much always involve vastly overpaying for an asset with likely no real resale value. That is very different than how a traditional capex works: usually the vast majority of the price you pay on a capex goes into the value of the asset.
So that leaves us in a weird sort of void. I assume the very-long-term solution is to figure out how to make an IP rights system that actually works for software (eg a new kind of patent), so you can finance your software R&D the way you do a house.
However, the "asset" associated with software development seems pretty uniquely hard to transfer. Almost every other intangible asset has a strong marketable component (eg a patent on a widget, a copyright on a song), whereas software just doesn't have one. Licensing doesn't count here - that's not a transfer of the asset.