> [H]ow much of the cost of a good is the energy, how much labor and raw materials?
For physical goods, most of it. Raw materials in particular aren't expensive except in terms of the energy required to extract them and process them (modulo supply and demand).
For labor it's more complex, but in general falling energy costs drive capital investment in automation (displacing and lowering demand for less-skilled labor) to lower price and capture greater market share (in order to get a return on the investment of capital), lowering labor's share of the final price, and the Jevons paradox drives increased consumption of the lower priced goods (as well as the lower priced energy). Software has been driving the same sort of dynamic for a while, though the way it affects industries seems to be more contingent, 'lumpier', and less predictable (eg. in hindsight, the disappearance of the travel agent as a result of online booking was somewhat obvious, but AirBnB effectively adding a lot more inventory to the market wasn't).
Not sure what happens when software and energy both more directly affect each others' supply and consumption, but the effect on the rest of the economy is going to be... interesting. Smarter energy grids that shift the economics of energy plants and sources at different scales; vehicles, homes, factories, & data centers that can all adjust their energy use/storage to take advantage of spot pricing, energy costs driving the ROI of training machine learning models etc. and thus changing the return on investment in compute capacity (and in software efficiency), and so on. There are a lot of feedback loops, and as "software eats the world", more (unanticipated) feedback loops will be created.
For physical goods, most of it. Raw materials in particular aren't expensive except in terms of the energy required to extract them and process them (modulo supply and demand).
For labor it's more complex, but in general falling energy costs drive capital investment in automation (displacing and lowering demand for less-skilled labor) to lower price and capture greater market share (in order to get a return on the investment of capital), lowering labor's share of the final price, and the Jevons paradox drives increased consumption of the lower priced goods (as well as the lower priced energy). Software has been driving the same sort of dynamic for a while, though the way it affects industries seems to be more contingent, 'lumpier', and less predictable (eg. in hindsight, the disappearance of the travel agent as a result of online booking was somewhat obvious, but AirBnB effectively adding a lot more inventory to the market wasn't).
Not sure what happens when software and energy both more directly affect each others' supply and consumption, but the effect on the rest of the economy is going to be... interesting. Smarter energy grids that shift the economics of energy plants and sources at different scales; vehicles, homes, factories, & data centers that can all adjust their energy use/storage to take advantage of spot pricing, energy costs driving the ROI of training machine learning models etc. and thus changing the return on investment in compute capacity (and in software efficiency), and so on. There are a lot of feedback loops, and as "software eats the world", more (unanticipated) feedback loops will be created.