I don’t quite understand what you mean by this.
Hardware requires the means of production and a global supply/distribution chain for each unit on the market.
Software is selling infinite copies of your product with no additional cost except marketing and an internet connection.
Modern (not cutting-edge) chipsets used in instruments are in the 100s of MHz and are increasingly using multiple cores.
Think of it this way: 1 MHz ~ 1 million operations (add, subtract are ~1op, multiplication, division, comparison, memory access are more) per second. If your sample rate is 96k, you have about 100 operations to do everything each sample. Oscillators, filters, LFOs, envelopes, effects, etc.
To give you an idea of what a constraint this is, the simplest, most optimised filter you could use is ~25 operations, and that’s for a single static instance. A real filter that saturates and moves around is a fair bit more.
This is very interesting.
Great way of breaking it down. This makes me curious about these chipsets. Are they all custom made per project? Or are there tried and true ones, that most synth manufacturers use?
No. These are industry standard microcontrollers made by the million. Even if you rent the IP for the designs, spinning custom silicon for parts like this is stunningly expensive.
It would be interesting to understand the underlying hardware + software differences between these:
And where do the Monome Norns and Critter & Guitari Organelle fit in?
Like an ‘Explain it like I am 5 years old’.
Funny I was just watching this Octatrack / PerFOURmer video and finding myself envious of this multitimbral setup.
That’s exactly what I like to do with the Model Cycles.
Its different Playmodes make it Super-Multitimbral (monophonic, duophonic or polyphonic; different MIDI channels too).:
Manual - check out page 25.
Fattest PWM I’ve ever heard with it’s synced LFOs slightly out of phase, with slight panning. A4, Tetra, and Tempest couldn’t even get close. Not the most versatile analog though, but I’d get another over any of the Moogs.
The stock Norns runs on a RasPi Compute module. The Shield DIY version uses a RasPi 3B or 3B+. Not sure abut the Organelle offhand.
I think the truth is most people buying hardware at the moment don’t really need their synth to do it all. As a result, multitimbrality isn’t really sought after by most. Especially cause the implementation on many multitimbral synths is very clunky.
The people who are after multitimbral synths are the people currently driving up the price on old Virus and Nords to pair with their digitakts and octatracks 
Shoot, even preset-only FB-01 or MKS-7 have sextupled in price in 10 years!
Yea I was referring to the hybrid stuff like elektron A4 which has a controller and “limited” DSP.
Others have brought up the DAW issue, but I think that’s probably the main reason.
You can have everything you want in a DAW, which makes limitations more valuable.
It’s a bit like the idea of synthesized Scotch. Imagine if we could perfectly synthesize any Scotch. Want to drink (quick Google) 50 year old Balvenie? No problem. The ingredients cost the same as gut-rot whiskey, so punch a few buttons and you’re sipping $50,000/bottle Scotch. In fact, you can make better Scotch. The machine can scan your DNA and synthesize a flight of Scotch that will feel like inappropriate metaphors in your mouth!
Doesn’t seem that great, does it? The minute anyone can have $50,000/bottle Scotch, nobody wants it, and now the “real thing” costs 10x as much, even though it’s completely identical. People will pay so much more for the “experience” of drinking completely identical Scotch out of an “authentic” bottle.
Similarly, I bet part of the reason we’re still playing synths modeled after 80s sounds and guitars designed in the 50s is that for many people the authenticity is worth more than the actual sounds. Because if you want to make new and original sounds, all the tools you need are waiting for you in your DAW.
Maybe, though it still has to calculate 1 vibrato lfo, 1 vibrato ramp, 2 PWM lfos, noise, a noise ramp, 3 envelopes, & 2 audio-rate LFOs per voice. Plus the global FX, which sound pretty expensive.
Diamonds would have been a better comparison since what you’re describing hae actually happened in the diamond Industry. Cheap, great tasting scotch for a fraction of the price would be the best thing ever.
Them youngsters are not a monolith, a few will appreciate a great board but most will not. Just like us old timers.
That is a very colorful analogy and I especially like the idea of a “machine can scan your DNA and synthesize a flight of Scotch that will feel like inappropriate metaphors in [my] mouth.”
I like to use hardware synths because they don’t change very often when I compare them to operating systems and DAWs. They are more of a durable good, and I feel the added connection when I pick up a piece of gear I have had for half my life.
Old personal computers? Not so much. I have a pile of them and I only hesitate to throw them out because I haven’t wiped the hard drives. Those are very expensive disposable goods that have planned obsolescence as part of their business model to feed the treadmill of progress.
Having settled on the OT as quite possibly the last sampler I buy, I would like a good multitimbral synth to go along side it that gives me the same happy vibes. In fact I already have several, the DN, the M:C, and a case of modular. But why let good enough be the enemy of perfection?
Partly yes … partly no … 
There are a couple particular use cases, which are implemented in software, but if I compare them 1:1 with real analogue behaviour, there is quite a difference.
Just as an example let’s assume we check out the filter-modulation by audio rates in a software emulated classic with the real thing. I did this a couple of times. TBH the sound of the software was absolutely okay, but the analogue circuits had a different timbre. Both great but different.
Or let’s compare modular systems, which give us most creative freedom. On my computer is Modular by Softube. Great emulations and also great sound. But there are many more options in the world of Eurorack. Some of the Softube modules are also in my studio as real gear … and there is quite a difference too. But again, both sound great, but it’s not the same timbre.
After this long post I would like to conclude that IMO we have great tools for new and original sounds in the world of DAW and this can be as much as many people need, but there is also much new and original sound outside the box, where DAWs don’t follow