Hindsight is 20/20.
IDK if selling nokia was a bad move. I think buying it was the bad move. The decision ended up costing so much money, that shareholders couldn't bare it, and became risk averse.
But MSFT inherited a lot of problems, like windows 8, SE, RT. It had some major pivots, and is still essentially pivoting. I think overall they have done very well to adapt, and that has had some casualities.
Could they have done better in the phone market? Perhaps. With all the game studios they own, they could have marketed a gaming phone. They have been late to some parties and early to others.
Interesting times lay ahead though. Premium phone markets have peaked. Its shifting year by year into low margin, low cost. Something that will not be without its noteable marketplace casualties. The shifting favours of chipset wars still are less stable that people assume, with background major legal cases mounting between apple and QUALCOMM, and intel quietly working on its arm chipsets. Desktops may now shoot leaps and bounds beyond what phones are capable of, due to the ability to circumvent failing moore's law, via bigger buses, faster drive speeds, caching and especially multi-processing on the GPU.
And what has been a focus on mobility may also shift, as technologies like AR, VR and proper neural net AI emerge, and the chipset technologies that make them possible, while networking technologies approach technical limit bottlenecks in their minimisation of latency.
Lastly, what has been, one of the most amazing, and fast adopted consumer driven technologies of our time, the smartphone, will not likely be followed by technologies that are as quickly adopted, nor consumer driven.
Its likely high end AR, VR, IoT and AI, as well as graphene based computing and display, is better driven by higher margin enterprise and niche markets, with bigger pockets for more specialised applications, before they reach the point of mainstream adoption, and technical completeness. The consumer is less demanding, has smaller pockets, and has simple needs. Like the smartphone before it, the mobile phone too, the early steps may have very little to do with every day consumers.
Take something like IoT, for an example. While a business can save millions with the right application, a farmer can manage their crops or herds, the everyday person can turn on a lightbulb with their cellphone. Until the AI/IoT can do more, its not worth a lot to people. It's however worth a great deal to business.
The same might be said of AR. The everyday person, might one day, love augmented reality entertainment, and functionality. But they aren't going to pay thousands for the gear, or want to walk around wearing a visor from star trek. Whereas surgeons, engineers can use these tools to do things that would ordinarily cost them a great deal more.
The example you bring of apps is also a good one. While people might use their phones more than their PC, they spend money on their PC software, and very little on their phone. Its worth it, for top ten developers, with those microtransactions, but quite unsatisfying to those trying to break through.
We sit, at the very crest of an era of a phase of consumer based technology, that according to all market analysis, should slowly shrink in volume, until every major player involved, not selling budget devices, starts, or already has started, to look elsewhere for the next cash cow. But history teaches us, that ,most phases of technology, are not end-state situations like the smartphone, or the laptop. Most such technologies require time to evolve, improve, and have manufacturing processes evolve, to the point where they are something everyone wants.
So what happens from now, I think will be a surprise to people with such expectations, of forever consumer driven progress and sales.
Basically everyone that wants a smartphone has one. The hardware is not vastly improving and is already satisfactory for most peoples needs. It's also deeply limited by chipset processes, and the size and battery tech constraints of current technological processes. The software, in terms of apps - its as fully funded as it will get. Most advances from here will be attempts to drive new use cases, like AR on phones, which I don't think will be hugely popular, and software - cloud based AI, that will quickly reach the limits of network latency and server load - as well as the free software model.
As hardware profits, over the following years begin to dip, growth going negative, such efforts to keep the fire alight will begin to be perceived by shareholders as wasteful. First party software being offered free, because of hardware sales, will be increasingly less justifiable. And then the issue of what's next becomes a real concern - nothing in recent history has ever been as fast adopted and as profitable as smartphones, and its unlikely future fields will evolve fast enough to yield the same results from consumers.
Moats like apples iPhone, and googles search are not invulnerable. Apple for example has a "dark data" ML program, and access to IBMs query matching ML, Watson. The two together might be a better search engine that google, if put together right. Indeed "autoML" has the potential to out perform any human designed software with a simple and singular application.
Equally, apples baby, the iPhone - there is basically no way, no possibility profits will not shrink. Its not impossible that google and apple might both have to do their own pivot.
As we moved from beta-max, to streaming, and telegrams to smartphones, the shifting favours of technology are never as permanent as they might appear to people living with them. Fortune favours the bold.