Knowhow is quite expensive.
When's the last time that you hired an engineer able to design an imaging system? Or designed one yourself?
Knowhow costs millions to tens of millions. Many companies can afford that.
The ability to actually build high-end sensors on a commercial scale costs multiple billions. Few companies can afford that, and Nikon isn't one of them.
Many companies buy components. Integrating them into a working camera should not be underestimated.
Dedicated stills cameras are a shrinking market. The bottom end is collapsing due to phones and GoPros. High-end action bodies are soon to be under attack by 8k-capable video cameras, and will likely be gone within 10-15 years. Aside from optics, Nikon is good at doing something which is becoming less and less required. That's not a winning strategy.
Furthermore, Nikon's biggest saving grace is that they're using the best sensor on the market. Imagine the D800/D810 with the 5D3 sensor (which came out at the same time). It wouldn't have been competitive at all. But basing yourself on using the best-available sensor, which you can't build yourself, leaves you beholden to the manufacturer of that sensor. Who just happens to also make cameras and other imaging products and is expanding aggressively in your direction.
Canon's and Sony's R&D and manufacturing capabilities all tie into expanding markets. Both design and make sensors - although the camera market is shrinking, the sensor market is expanding rapidly. Their work in mirrorless technology is also applicable to video, security systems, AI and many areas other than stills cameras. In contrast, Nikon's capabilities (other than optics, which is restricted because it only makes components for Nikon products) are all tied to SLR cameras. Outside of an SLR, their developments in AF and other supporting components in a camera (i.e. everything but the lens and sensor) are essentially useless, with little commercial application. Aside from optics, they can't produce or sell anything that is of use outside an SLR camera - a market that's slowly dying.
Sigma is the perfect example of a company trying to do it all and coming short in the user experience. Foveon is a nice if specialized sensor. The lenses on the DP2M 30mm and 50mm cameras are spectacular. Ergonomics - nope.
Sigma is in a much stronger position than Nikon. Although it doesn't have to be that way, if Nikon would just unleash their optics division to make lenses and other optical components for everyone and everything, rather than just Nikon-branded equipment.
They hardly pretend that Foveon and their camera business is anything other than an interesting side-activity. Their main business is optics, and they know it.
They design and make lenses for everything and everyone - cameras, medical equipment, anything that requires a lens. They don't seem to care if it carries the Sigma name or not. Sigma (and Zeiss, and Tamron) are to lenses what Sony are to sensors - a supplier in a market that, overall (including all uses of lenses, not just dedicated stills cameras) is expanding.
The poor ergonomics and lack of support in their cameras matters little to their bottom line. The only people it bothers is the photographers who use them, which is a tiny market that Sigma doesn't seem to bother trying to expand, or even keep, because it's not their core business and is a shrinking segment.
Sony has been progressing a lot and I like many of their products and own more than a few.
But today I feel very limited appeal in their FF offering for the various types of shooting I do. I will believe in a possible demise of Nikon when I'll be convinced that Sony helps my photography more than what I currently own. Today it's not even close.
Sony - and Canon, and others - doesn't care what type of shooting you do. Sony, Canon and others are there to make money. It just so happens that the technology they're developing helps with photography. That's all well and good for photographers, but advancing photography is not the fundamental reason they're developing it, and the technology has many applications far outside photography. Sony, especially, appears to be trying to dominate the sensor market and make inroads into the non-photographic camera market (everything from security, to video, to medical, to telecommunications applications), and its photography developments are a useful marketing tool to showcase all their developments. It's the same reason Ferrari or MacLaren participate in Formula 1 racing - not because making race cars is profitable, but because it lifts the status of their brand.
Given the advances made between the A7r and the A7r2, the A9 (or whatever their likely 2017 high-end release is called) is likely to have a much more sophisticated AF system and other non-sensor components, and may well match the capabilities of non-action-specialised SLRs, either in this generation or the next. But Sony isn't developing this technology to displace SLRs. It's developing this technology because it's applicable in many different, expanding fields that rely heavily on imaging sensors (think driverless cars and other 'intelligent' robotic systems), of which still photography is just a small, but highly-visible one. Just look at 'eye focus' AF. It's a small tool for a photographer, and hardly makes or breaks a camera. Yet it costs millions of dollars to develop. The return on investment, if it were just for stills cameras, would be abysmal. But look at it another way - it's a rudimentary form of subject recognition and AI. Something that will be invaluable for smart security systems, automation and future, non-photographic applications, whose capabilities will improve as processing power and battery power/efficiency improve. Today, it's press a button to focus on the human's eye. Tomorrow, it's 'track that leopard's head'. In five years' time, you might tell it (verbally) 'Take the drone, fly it six metres into the air and take an aerial group shot. And focus on the dog'. In ten years' time, it might be a driverless car that can read and follow street signs (in twenty different languages). And in fifteen years time, you might tell a network of cameras (both static and drone-mounted), all hooked up to a computer, to 'find this guy somewhere in the city, track him with a drone, keep me updated on his location and send off an alert if he pulls out a weapon, starts speaking in French or mentions the word 'bomb''. It's all the same, expanding technology, with applications far beyond photography, all of which tie into having high-resolution, high-dynamic-range sensors and an intelligent AI system which can interpret the data generated by the sensor - sensors which Sony can make, and an AI system which Sony is developing. Mirrorless cameras for still photography are just a small part of it - a valuable showcase, and, unlike driverless cars or crimefighting drones, something which can be demonstrated now, even at this rudimentary stage of AI development.
Canon has also demonstrated capability in mirrorless cameras, although, thus far, they haven't made them in anything but consumer-level, budget models. But the technology and manufacturing capability are there, and it doesn't take much to apply them to a higher-end model, or to non-photographic applications (which Canon has already done).
