In a 'pure capitalist system' (one in which there were only workers and capitalists, all other classes having been destroyed, and in which the capitalists were forced by competition to behave as the pure embodiment of capital by investing all their surplus value), the mass of surplus value would increase with every cycle of production ad infinitum. The capitalist class would have ever greater quantities of surplus value at its disposal and would be under competitive pressure to invest this in an ever-larger scale of production.

As Michael Kidron has put it, Marx's argument assumed that '... All output flows back into the system as productive consumption. In a closed system like this, allocation would swing progressively in favour of investment.' [24]

That in itself does not automatically mean a rise in the ratio of 'dead labour' to 'living labour'. The investment may be 'capital-saving'. If scientific knowledge is progressing and being applied as new technologies, then some of these technologies may employ less machinery and raw materials per worker than old technologies. To give a relatively recent example, the production of newspapers using phototypesetting and lithopresses is less capital-intensive than the old method using linotype machines and letterpresses.

But that is not the end of the argument. It shows only that at any one time there will be some new technologies that are capital-saving. The important point, however, is: what will be the average result of new technologies? Will they save capital or increase it?

If we take the argument one stage further, it can in fact be shown that if there is a massive amount of profit-seeking investment in the hands of rival capitalists, then the overall tendency will be for the average investment to increase capital, to be capital-intensive.

Firstly, the most competitive capitalists in each line of business will be those who introduce most innovations. At any given level of scientific and technical knowledge some of these may indeed be capital-saving. But when all these have been employed, there will still be other innovations (or at least capitalists will suspect there are other innovations) to be obtained only by increasing the level of investment in means of production.

Secondly, the fact that some technical progress can take place without any rise in the ratio of capital to labour does not mean that all the advantages of technical progress can be gained without such a rise.

The point can be simply illustrated by assuming, for a moment, a state of affairs in which in a given field of production new scientific knowledge is not emerging, and in which all existing techniques possible at a given ratio of capital to labour have been exhausted. In this situation, a capitalist who uses more means of production per worker can expect to get access to improved techniques of production which may have been known about in the past but could not then be used because the ratio of means of production to labour was too small – the capital was not available to develop them. By contrast, a capitalist who does not increase his means of production per worker will be stuck with the existing techniques.

Thirdly, if an individual capitalist can increase the ratio of capital to workers he will be able to invest in innovations that need more capital as well as those that need more labour. If he cannot increase this ratio then he will benefit only from those innovations that need more labour – and he will lose out in competition with those who can.

In the real world, every operating capitalist takes it for granted that the way to gain access to the most advanced technical change is to increase the level of investment in means of production or 'dead labour' (including the 'dead labour' accumulated in the results of past research and development). It is only in the pages of the most esoteric journals of political economy that anyone imagines that the way for the Ford Motor Company to meet competition from General Motors or Toyota is to cut the level of physical investment per worker. The capitalist usually recognises that you cannot get the benefits of innovation without paying for it. His firm may by accident stumble upon a particular innovation that requires less capital per worker, but the only way he can guarantee getting such innovations is to increase his level of investment.

If the capitalist cuts the amount of investment in means of production per worker, he might still stumble upon some innovation unknown to his competitors. But luck such as this is also available to the capitalist who increases his investment in means of production per worker, while he can also match the innovations stumbled on by his competitors and obtain technical advances unreachable by those who cannot afford his 'capital-intensity'. Since, in theory at least, there is no limit to the possible increase in the ratio of means of production to workers, there is no theoretical limit to possible innovation based on this method of competition.

For these reasons, other things being equal, we can expect there to be always more innovations calling for increased capital than those calling for less. The average amount of means of production per worker – Marx's 'technical composition of capital' – will rise.

Only one thing could stop the pressure for this rise: if for some reason there was a shortage of profit-seeking investment. In such a case the capitalists would be forced to forego hopes of achieving the innovations possible through greater investment and settle for those they might stumble upon by accident.

~24. Michael Kidron, Western Capitalism since the War (London 1968), p. 46.~

https://www.marxists.org/archive/harman/1984/explain/01-marx.html