Well.....There is no doubt that the chips are a bit more expensive but then they are literally a system on an chip. I am working on a project there we are upgrading an existing design, changing the PC board and, and adding to the software. The previous design used a Dallas DS80C320 chip plus external 128 K byte FLASH and quite a bunch of external XRAM type peripherals including a 32Kx8 RAM. The previous board was stuffed with parts. By replacing with Cygnal C8051F126 (at about $21 in 250 quanity) we are able to wipe the board off to just a processor, RAM chip and RS232 driver/receiver!! Yes that chip even has a mode where the XRAM bus can run de-multiplexed without the external address latch!! And I get 25 MIPS performance with 1X clocking!! That part also has 2 12 bit DAC's 2 Comparators, 8 bit A/D, 10 or 12 bit A/D, 5 timers, separate PCA, 2 UARTS, temperature sensor, onboard debugging support, Onboard 8K XRAM, onboard clock multiplier PLL, I/Os that can be programmed for active pull-up, etc etc etc...

For sure you would not use this chip in an application that requires an $1 chip. But I have found that even the 'F226 chip with 8K flash, 1K XRAM, and typical 8052 architecture is not so bad on smaller projects. You get WAY more performance in a chip that costs only about 1.5 -> 2X the typical ATMEL or PHILIPS chip. And the things work as advertised too.
Mike Karas