> one of the places that a lot of old 8-bit chip inventory ended up
Those 8-bit CPUs didn't end up in pinball machines because they were old or cheap, they got there back when they were brand new, bleeding edge and very expensive. The history behind how pinball machines became CPU powered in the mid-70s is surprisingly interesting. Starting in the 1920s as simple boards, with pins and balls which incrementally evolved over decades to include mechanical score-keeping mechanisms and then electro-mechanical bells, lights and coils, pinball had never been a "tech" business. Yet the need to attract players and keep them pumping quarters had gradually turned pins into fantastically complex electro-mechanical computers. They were essentially state machines with layers of conditional branching logic and computed score tallies which made them increasingly hard to design, manufacture and repair, causing prices to spiral out of control and margins to shrink.
By the early 1970s this arms race was already a serious problem on its way to being an existential threat. Dave Nutting, a consulting engineer and erstwhile entrepreneur with an interest in transistors had been trying to apply discrete solid state parts to make non-pinball amusement machines with marginal success when he saw an ad for the first commercial microprocessor, Intel's 4-bit 4004, and immediately ordered one of the first development kits. He thought the integrated processor would be much better than stringing together dozens of separate logic chips and maybe powerful enough to replace the state machine logic in a pinball machine. He successfully pitched Bally on funding a proof of concept based on an existing Bally EM machine. Although it did eventually (mostly) work, there were teething pains and it was unreliable. Being unfamiliar with bleeding edge R&D, Bally quickly lost interest but never asked for their EM 'donor' machines back and Nutting used them to try to interest other pinball manufacturers, sending one to Mirco Amusements in Arizona. Mirco had just a hired few engineers away from a nearby Motorola office and on seeing the Nutting prototype, those former Motorola engineers realized the new Moto 6800 8-bit CPU might be even better and quickly produced a 6800 CPU version of a Mirco EM machine which Mirco showed at the 1975 Amusements Expo. Although Mirco did manage to ship some CPU-based units of that EM, they too were notoriously unreliable, making "Spirit of 76" CPU machines extremely rare today.
While the first two attempts at CPU pinball weren't successful, others in the industry including Williams saw the 6800-based boards in the CPU few units Mirco shipped and quickly set to copying the design (copying designs was rampant in pinball), then other companies copied William's copies. And that's how almost the entire pinball industry ended up using the Motorola 6800 family of CPUs for the next 30 years. Interestingly, the only major manufacturer who didn't was Gottlieb, who instead adopted a strange, little-known Rockwell CPU largely because Rockwell created a pinball CPU board based on their dev kit and licensed it to Gottlieb for free to get the design win (a common practice in the defense industry).
It's a testament to the power of inertia that for over 30 years CPU boards and code were incrementally improved the minimum amount necessary and incessantly copied back and forth between manufacturers. Stern made the last pins based on the Motorola 6809 around 2007 and they only stopped then because the chips were finally discontinued and couldn't be ordered anymore. Weirdly, Motorola's 6800 CPUs weren't chosen because they were cheap since they were actually the most expensive 8-bit CPUs when they launched (about 8x more than the later 6502 which powered the Apple II, C64 and Ataris). It was just that the 6800-based design existed and it was easier to copy something that worked than hire engineers to design something new. It's fortunate that the 6809, a much expanded and improved 6800 which launched in 1978, turned out to be a CPU now widely regarded by experts as the fastest, most powerful 8-bit CPU of the 70s or 80s (largely because is was partly 16-bit). That plus the fact pinball machine logic doesn't really need more than 64K of RAM or 16-bit processing meant manufacturers had little reason to develop something new until they were finally forced to.