Not unless you're trying to compete with Intel. Consider the various bitcoin ASIC miners: beat all other technologies for solving that problem for an investment of a few million.
However, algorithms are so rarely the stumbling block for applications; data storage, management and communications are. The major exception is graphics, hence the GPU.
a bitcoin miner isn't a isn't a general processing unit.It does something specific that's very very easy to optimize and implement, and you don't use latest manufacturing process but tools like easic , hence the small investment.
I don't think he was suggesting that developing a general process cpu that is competitive with Intel's offering costs only 10k. He said that getting your ASIC design, i.e. application-specific integrated circuit, etched into silicon is cheaper than is used to be and that's true. I know guys who implemented things like novel power converters on an IC as their masters projects, using only the budget of, well, master projects.
The money is a huge barrier to entry for hobbyist types, but if we're talking about commercial stuff, it'll probably cost you a lot more to pay for the engineers who are competent enough to implement something that will actually work than it will to do the fabrication.
Hence my comment about it being cheap for very simple applications. If you want to compete with market leading vendors then yeah of course it's going to be hideously expensive.
However, algorithms are so rarely the stumbling block for applications; data storage, management and communications are. The major exception is graphics, hence the GPU.