I do not think an answer can be provided, or at least would be quite hard, for the question as stated. I do not know what is a current popular belief about prices and can not rate the level of misconception. Furthermore, different CubeSat projects can cost very differently. My answer will focus on prices, not beliefs of prices.
Distinction has to be made between the cost of a launch, the cost for building of the satellite and the cost of a whole mission. Launch costs are surely significantly lower than for large satellites, starting from a price you would pay for a nice new car to put a satellite into a LEO. However mass grows quickly and price with it, 2U is already as twice as heavy as 1U. Satellite building prices vary widely. You could assemble basic 1U with COTS for under 100k, or go with custom platform with novel and unique payload for several millions. How long will the satellite be operated? Someone needs to be payed to carry on with the operations. Do you build your own ground station or rent one? Maybe another million or two for missions longer than 5 years.
That sounds quite expensive, sure. Though how it compares with larger satellite classes? Right now I am working on a software for a large satellite, a bit above 2 tons, and costs for only developing of the software is already in the range of the costs of the whole CubeSat missions I have been working on previously. Also, it is worth noticing that there are a lot of Universities and startups that are building CubeSats, while none (as far as I can say) is building medium and larger satellites (sure, there are collaborations and consortiums that include them, but no independent projects). Add to this much faster development cycles and possibility to reuse developed modules and designs for CubeSats, especially if constellations are planned, and you can get significant savings there as well over time.
Estimating a return on investment is another comparison metric. I will not go into this, as it is quite hard to compare commercial and scientific missions. However, one can expect that the CubeSats are less reliable and prone to premature failures. And it should be obvious that they can not carry large instruments, have limited power generation abilities and can be limited with thermal capacities as well, reducing possible returns, whatever they might be.