Complexity and integration requirements. Tanks 70 years ago were not as complex as our current heavy armor by several orders of magnitude.
=== EDIT ===
To clarify: the complexity arises from electronics and electro-mechanical assemblies, the chemistry/metallurgy/materials science from armor/chassis, engine complexity (M1 has a jet turbine), and so on. We have a depth of advancement that is as complex at the knowledge end as it is at the application end.
But isn't part of the problem that very technical complexity? It seems time and time again that military acquisition projects are completely sunk by technical complexity that, when you really look at it, seems wholly unnecessary. Consider the famous story of army radio acquisition (e.g. SINCGARS and JTRS) running fantastically over-budget, over-schedule, and still providing poor reliability, all due to extreme complexity, while the Marine Corps has been very successful using off-the-shelf products made for rugged commercial use. In the case of radio systems, are the needs of the military really that different from the needs of e.g. the oil exploration industry and the freight dispatch industry?
I believe that a fundamental problem is that the defense-industrial complex intentionally exploits the acquisition process to drive up the complexity of every system to tremendous levels. This looks defensible because it's always easy (e.g. for a politician) to say that the military needs the best cutting-edge equipment, but it also contributes enormously to the profitability of defense contractors. This was a pretty swell strategy for the defense contractors but has turned out bad now that (and 'now' here is like the last 30 years) the complexity of many military systems has been increased beyond the ability of the contractors to actually deliver, even at the enormous agreed-upon prices. But it actually still works out fine, because the DoD, congress, etc. have totally failed to put the brakes on well after the complexity trainwreck has begun.
During the Gulf War the Marine Corps went to war with Banyan VINES practically bought out of the local big-box store. Now the AMHS is a fantastically complicated acquisition program that still seems inferior to Isode's "Military XMPP" offering - and then in practice soldiers are using Telegram. DoD tries to push an integration system to relay IMs from one system to another because they can't manage to standardize on any one product anyway. An IRC gateway is still a critical feature because, who knows how many millions of dollars later, IRC is still one of the most usable options available to the warfighter.
How much of this complexity is actually necessary, how much of this complexity is just the contractor's salesmanship?
Not to belabor a point; but the M1 tank you're describing was developed entirely by Chrysler Defense in the '70s before being sold to GD Land Systems. Chrysler had done lots of work with gas-turbine propulsion in cars, and while it didn't lead to a production car, it led to tanks.
I'd just like to know why that happens less and less nowadays...
All of the apologist replies are based on the problem of overoptimizing. Yes, if you pay 100x as much, you can always get something 10% better. Is it worth it? Well, it is 10% better. Depends.
For our military since the 70s, we have not fought a major power, and the main incentives are to keep the contractor money flowing and the body bag count down. So, the system is working to deliver what it is optimized to deliver.
Were we to get into a war with Russia or China though, it's pretty clear we'd get our asses handed to us because we can only make N weapons for Y dollars while they make 10N weapons for the same price, and theirs are only 10% less effective than ours. As it is, ISIS and the Taliban can keep us spinning our wheels for a decade with nothing to show for it. Imagine a foe with an actual military budget…
To be fair, the Russians also occupied Afghanistan with very similar results. I think this is a tenant of colliding ideologies of the occupied people, ideologies of the occupiers, and asymmetrical warfare (see a fantastic book titled "Small Wars").
Complexity in what sense? Level of hardware integration? # of embedded controllers? Amount of time necessary to familiarize oneself with the operation of the system?
Complexity in every sense, at every stage of the design, testing, fabrication, and delivery.
From discerning appropriate tactics for the coming century (which are used to produce design requirements that equipment is designed against) to writing the user's and maintainer's manuals, to the number of people required to get a particular sub-system into production, and so on.
The bottom line is that a modern tank requires more people, more knowledge, more complex (number of parts, functionality, materials selection, fabrication, testing, etc.) and interconnected sub-systems, more complex manufacturing techniques, tighter tolerances (sometimes) (which lead to more complex methods of measure), more exacting specifications, etc., than a modern consumer vehicle.
A standard automobile may have 20 to 50 microcontrollers that talk to each other on one CAN bus. A modern tank has at least 3 independent communication networks, an electrically isolated diagnostics system, distinct optics systems for the commander and pilot and gunner, a fire control sub-system, and so on. Each of these subsystems is immensely complex, designed and manufactured to exacting specifications for durability and maintainability.
The specifications for each are so exact, that any subsystem can be torn out and another designed to the same spec can be swapped in. This is no trivial feat. This isn't like a new set of rims onto your car. It isn't even akin to piggy-backing or changing out the ECU in your car. It is at a whole different level, because each sub-system has to integrate with at least two other sub-systems (power and communications). This is like being able to tear out the autopilot portion of your Tesla and swap in some other company's autopilot, while keeping the rest of the systems intact.
Complexity and integration requirements. Tanks 70 years ago were not as complex as our current heavy armor by several orders of magnitude.
=== EDIT ===
To clarify: the complexity arises from electronics and electro-mechanical assemblies, the chemistry/metallurgy/materials science from armor/chassis, engine complexity (M1 has a jet turbine), and so on. We have a depth of advancement that is as complex at the knowledge end as it is at the application end.