That's kind of the contract between workers and their employer. Workers get short-term benefit of not having any risk, and employers take the risk but receive the long-term benefit of capital.
Does this really mean that the amount of profit the employer makes is an adjustment reflecting how little risk the employees take on because they didn't take a chance to start their own companies? With employees whose income they would then hypothetically make the same adjustments to?
It's not obvious that the employer is justified in protecting themselves from their own risks with money skimmed (relative to market value) from consenting employees. These workers allow it because they still prefer it to being unemployed or a contractor themselves. But in theory they are doing worse being a part of the company than if they did the exact same actions (magically disembodied from the company monad) would be compensated, if they took the risk for themselves. The employer's risk is if they don't find enough people who are willing to risk not taking enough risks!
If having more employees means having more sources of income, big employers would not have as pressing a demand to hedge against catastrophic events; so they would recalculate the risk adjustment fee built into wages. Rarely do you see large employers give their employees better wages to reflect the company getting past earlier stages of high risk.
Basically in the long-term the value to being a capitalist is getting others to pay you for working for you. The ability to suppress wages pays for this profit - and that ability comes from contracts being costly to arrange themselves. If we could reduce the cost of contractual work and more people did it, most if not everybody would be better off, except for the people profiting from mutually-satisfactory contracts being difficult to do often.
Taleb gives this a good treatment:
"If the company man is, sort of, gone, he has been replaced by the companies person, thanks to both an expansion of the gender and a generalization of the function. For the person is no longer owned by a company but by something worse: the idea that he needs to be employable
...
A free market is a place where forces act to determine specialization and information travels via price point; but within a firm these market forces are lifted because they cost more to run than the benefits they bring. So the firm will be at the optimal ratio of employees and outside contractors, where having a certain number of employees, even when directly inefficient, is better than having to spend much resources negotiating contracts.
...
an employee is a risk management strategy"
Tools that increase productivity however do come from improvements made by workers - namely the workers whose labor is absorbed into the firm that sells the technology.
The money to pay for the machine comes from revenue earned by workers. The company has to value the employee's work more than their wage for this to happen - so there is selection pressure for willingness to accept low wages; lower wages means greater profit; greater profit means greater precarity... etc. Technology devalues worker specialized knowledge by rendering it less useful, while keeping production at least constant (e.g. being computer-literate is as valuable as it is specialized)
Many fewer people (can) produce a disruptive technology compared to the number of people it displaces, but that does not exempt them from the same selection pressure. A technology firm that depends on outcompeting others in its sector still has to make its own productivity gains. They'd likely prefer to do with one inventor what they could do with two. Maybe very long from now there will be a kind of artificial intelligence/singularity perpetual motion machine that - because of its near-zero cost of tinkering to discover improvements in its self over the competition - no other firm can compete with. A cloud "SaaS-as-a-service" like John Lilly's fears of solid state intelligence (SSI)