The actual manufacture of a robot, depending on its scope of automation, ranges from inexpensive to insane. Most industrial robots are in the upper middle.
Some simple robots are cheap, mostly mechanical.
So, just making one overlaps both our points of view.
Setting mechanical ones aside, the programming, test, safety, and depending on what is needed, interactions between multiple ones is very expensive!
A seat of software can cost a years wage for an assembly line worker, and the people needed to do this work tend to be highly educated and expensive as well.
That can change, and maybe will as tech advances and more advanced computation, sensing and such continue to scale.
But, for the foreseeable future, doing that work is expensive and time consuming. And the product of it tends to be fairly rigid.
Changes are not cheap either, though they can be a smaller fraction of the original build, depending on both the change and how robust the initial build was.
Maintenance is a factor too, as is aging systems and the software footprint required.
I have experience with these industrial machines (FANUC and the like), and they definitely cost less over time than people do, but only at scale and on well defined tasks in well defined conditions.
The initial investment is a lot more than people.
Where there is any "turbulence", task change, part change, condition change, people do prove far more efficient.
As a general rule, robots that are more capable, in the sense of fully replacing skilled people, are a lot more expensive.
Where the tasks require less of people, and conditions are constant, they are much cheaper.
In my manufacturing career, 80's and 90's I spent time automating and training people to exceed automation performance. Have been on both sides of the discussion. Improvement just is not cheap.
Today a lot more is possible, but the basic dynamics remain unchanged. We are far away from being able to generally replace people.
And we flat out are not capable of that where the labor involves taking care of one another and or performing the many basic labors we all depend on far more than we realize.
This is why I make a point of suggesting we value basic labor differently, and in general, higher.
We need it, for one. People, like the HN audience, are not cheap and what we can do is not common. We get valued for that, and there is downward pressure on it all generally.
However, if we suddenly have to perform basics ourselves, our output goes way down, and that too is expensive. Having more of our time available is worth far more than the compensation is for basic labor.
Most people are not like us, and the training needed, assuming the basic skill is present, is expensive as well.
Most jobs are basic labors too. That means most wages are not contributing to demand, and as those numbers rise, with costs and risks, the issues grow more acute.