Not counting fortifications, roads, related logistics, opportunity cost and lost production from harming civilian enterprise, treaty and agreement expenses, losses due to enemy action and threat, environmental degradation, economic inefficiencies from security provisions, political compromises driven by military necessity, and so forth (you can read all that in the links above). And taking no account of the actually available GDP - the kinds of surplus available for military siphoning without destruction of the society being defended. (Sparta's millitary, for example, did not have access to that fraction of its GDP involved in supporting its agricultural and manufacturing labor (most of its GDP), because they were slaves of a kind (quasi serfs) who could be neither safely killed or safely armed. So the small percentage of its GDP that it devoted to its military (your figure of nearly 100% obviously hallucinatory) was nowhere near the burden of that military - too many dead men in which large investment had been made, for starters, eventually proved irreplaceable from its relatively small military class). So the actual monetary burden of the military, the money only, would be a much larger absolute amount forming in comparison a fraction of a much smaller available GDP. 2.5% of GDP, in other words, was nowhere near the burden of the Roman military, even in money only. And likewise with the US - as we can see by the strain put on the US economy and society by the Iraq War: a total victory of the least costly kind imaginable, initially, but we will be lucky to ever get out from under it without more and more serious economic damage. And that is only a fraction of the burden of the US military overall.