The first or second paragraph should’ve said “ASML generates 250W EUV finally ending up 5W EUV due to reflection, whereas this will end up with 1kW EUV for chip production”. It wasn’t clear at all why they were doing this until 3/4 through the article. And I assume it’s 1kW EUV as the source and with less mirrors possibly 20W or more on the wafer (?).
Maybe I missed it, but it doesn’t mention the amount of energy required to end up with this 20W EUV beam. A particle accelerator probably uses more, but maybe the productivity balances out the efficiency?