It's all about the economics. The guy who builds the thing will be aware of how many Watts of rating-per-dollar a tube offers. In 1975, the 6MJ6 tube would provide roughly 50 Watts plate rating (intermittent) for six bucks. That's eight and a third Watts per dollar. The 3-500Z tube cost around $150 at the same time, offering three and a third Watts per dollar. This is why the Phantom has ten or twelve 6MJ6/6LQ6 instead of one 3-500Z, more or less.
Best example of this I ever saw was in QST 35 or so years back. A military-surplus tube, the '1625' was a quarter or so each if you knew where to buy. It's about the same rating as a 6146. The fella wired thirty (30) of them in parallel. Ran it on 80 meters, so all that parallel capacitance didn't kill him. Used a half-dozen big power transformers from old tube-type console televisons for the H.V. The trick would never have worked at a much higher frequency, but a 30-Watt tube for a quarter gave him a ratio of 120 Watts per dollar more or less. Too bad it would never work on 27 MHz.
A few years back there was a rash of amplifiers built with the 250-Watt ceramic tubes built near here. Anywhere from one to seven of those tubes in them. The biggest reason was cost. He was getting surplus tubes meant for aircraft use. The filament (heater) in them ran from 26.5 Volts. Not one commercial base amplifier was ever sold that used those. Nobody wanted them, making the price as low as $30 each. Didn't matter that the filament ran at an oddball voltage. A 26.5-Volt tranformer cost him about the same as a 6-Volt transformer. That tube offered him the same eight and a third Watts per dollar as the 1975 sweep-tube comparison. Using any other kind of tube would have doubled or quadrupled his cost to make them.
Didn't make it a good idea, just made it cheap to build. Never mind about the cost of ownership. The more tubes, the more it matters that they share the load equally. If they don't balance the load, one tube out of two can only get pushed twice as hard as it should. Not good, but not fatal if you unkey soon enough. And don't do it too often.
If one tube out of six is "stronger" than the rest, it could get much more than just twice the power on it that it's rated for. The more tubes you've got, the more you'll buy to keep them matched, balancing the load evenly. If two tubes out of six fail, you may very well have to buy all six. Just to get six tubes that match evenly enough. If one of them is much stronger than the rest, a six-to-one overload will "POOF" a tube a LOT sooner than just two-to-one. Combining new, strong tubes with old ones usually will poof the newest, strongest one out of six. It will "hog" the load while the other tubes in parallel with it just "loaf", pulling less than their share.
A bigger tube may cost you more per Watt, and usually does. If the design is good, and the operator is sensible, it saves you money in the long run, in cost of ownership.
Another one of those "Pay it now, or pay it later" propositions.
Oh, and the broadcast equipment contains separate bias adjustments for each tube, so that they will "balance" properly. This also affects how cleanly the tubes will run. A really BIG deal with a broadcast transmitter. Purchase price is usually much less important than the cost of operating a transmitter for broadcasters. They'll pay more for a tranmsitter up front if it will make more money for them in the long run. That sentiment is usually not found in the CB/Ham market. Not often, anyway.
Strategies that reduce purchase price nearly always boost the cost of ownership in the long run.
73