Evidence? Alot of folks through out "it is just like Firewire blah blah" the licence fees are high.
There are licensing issue. The controllers cost money. You loop implied/pragmatic license fees into the cost of the controller but that is like saying there are x86 licensing fees wrapped into Intel CPU's. Not really.
The is an arstechnica article that folks conflate to be licensing costs.
http://arstechnica.com/apple/2013/0...ssues-can-thunderbolt-break-out-of-its-niche/
But is actually says licensing
and costs. It is not an adjective of 'costs', that is a separate thing.
- Certification (much more complicated to hardware developers)
Dual edge sword. There have been zero issues of Thunderbolt interference with other sockets/functionality. 'Race to the bottom' USB 3.0 and Bluetooth can't really say the same. How many 'bad' , as in don't work well, TB systems came to market.
Whether Intel has be too picky, cautious or not is debate, but one thing they have been trying to do is not flood the market with cheap, as is substandard design/engineering, solutions just to hit lower price points.
If trying to establish a standard aimed more at higher end, not ubiquitous , market this has merit. Super cheap vendors with shoddy products is something that USB will survice because it has tons of inertia. If had started that way against somewhat overlapping competitors it would have had problems.
- No add-in cards (hard for early adapters o use it, like old-MacPro users)
Thunderbolt was never primarily targeted at systems with several PCI-e slots. It won't work because those systems don't have the prerequiites. Thunderbolt is more than just "external PCI-e" ( there was already an external PCI-e spec). Thunderbolt is DisplayPort and PCI-e. If don't have both then not Thunderbolt.
That has actually decreased substantially already over time. "has to be just as cheap as USB" isn't really a necessary requirement.
- Arriving late with a lot of competing interfaces (usb3, sata/esata, firewire, SAS, 10Gbit ethernet, PCle, displayport, hdmi, infiniband etc)
USB 3.0 took how long to go mainstream? Almost two years till Intel put it into the core IO chipset. 10GbE? Still no low cost client cards and only recently an affordable unmanaged ports switch. The gap betweeen standard and "lowest price point curve" is often a year or so long than bleeding folks want it to be. The software and product development cycles are no where instantaneous as folks make them out to be.
- Tries to be Jack-of-all-trades but do not really excel in any area, and misses some of the important ones (replace low-end usb devices and highend ultra-performers)
Thunderbolt doesn't try to replace USB so it isn't really missed. The whole "one port to rule them all" was largely made up hand waving created not by Intel but by spin doctors trying to assert how "Apple is going to skip USB 3.0 because Thunderbolt is the future" kool-aid. There were a couple misguided quotes from an Intel rep blow out of proportion but there never was creditible evidence that outside of the hype/smoke selling forums that Thunderbolt was being marketed as a "everything killer".
Apple hyperbole blows crazy talk about everything. iPhone defies physics... not really. Thunderbolt fastest IO ... again not really.
We have still seen very little development in obvious areas like
- Empty JBOD storage boxes ( I guess many want to pick their own disks, not get them preinstalled)
There can be faster ROI on a more complete system with higher margins. Not particularly surprising the first generation has been these more complete subsystems.
The 2013 update to TB controllers (Redwood Ridge) really did nothing for peripherals ( primary aimed at adding host DisplayPort 1.2 capability and basically useless in peripherals ).
- External boxes for highend Display cards, especially for laptops.
Thunderbolt was not really primarily designed for this at all. Part of its primary function is to distribute GPU output to other devices. It is not to remote locate the GPU itself. You can, if the internal GPU is relatively lame, do that but it is not a primary objective. Never was.
- PCs and PC motherboards seem to have very slow adaption of TB
It is not an industry standard. It is an Intel standard. There are lots of folks wary of Intel's long term agenda and not particularly happen with their control of the industry.
Where PC actually means "Box with slots" the slow adoption is not a particular surprise since Thunderbolt was not primarily designed for those legacy form factors anyway.
Some board vendor picked it up because it was the latest techno "shiny object", but yeah it isn't surprising it didn't go much futher than that. The vast majority of the overal PC market isn't interested in bleeding edge shiny objects.
Not to be pessimistic, but it seem like Thunderbolt will face an even harder uphill battle than firewire ( which was aided by a lot of external devices using the interface like videocameras.)
When videocameras have 1TB storage they need to transfer quickly and high end DisplayPort format video they can output then Thunderbolt would be a good fit.
The storage part is coming over time. How DisplayPort does versus HDMI over time is still kind of up in the air, but yes it is behind in deployment.
Thunderbolt as an pragmatic industry standard docking station connector is relatively a far more likely driver though. Not that the PC vendors want to give up their own proprietary docking station connectors. The question is whether customers will eventually push them to something more standard.
The new Macpro will most probably not change much here. It is a small niche product in a niche market.
The Mac Pro could be influential on other PC vendors All-in-one models. Not so much to match the number of TB controllers and ports. But far more in that some devices will trickle down to iMac and PC vendor All-in-ones also.
The new Mac Pro is going to push those were in the OS X PCI-e card market to look harder at wrapping their card with an enclosure and selling it as derivative device with a TB connector. You are somewhat looking in the wrong direction. The former, trapped in small nice of nice (Mac Pro only), vendors can expand into the whole Mac market with Thunderbolt devices.
What will probably happen over time is that the only OS X targeted variant will be the TB model. The PCI-e card form factor version will only be marketed to Windows/Linux folks.
The run rate for Mac is about 10M units a year. Over 3-4 years that is 30-40M systems. Vendors could ignore that market, but something like 2% (of 40M) of that is around 800k. That is going to be sizable units for alot of card vendors.
