When you look at the costs for mining PoS vs mining PoW coins as a percentage of the value of the total coins then PoW is much more efficient. PoS actively discourages people from spending their coins therefore the economy will never mature and thus the value of the coins will always be small. This will result in the PoS coin always having a smaller potential market cap then any PoW coin to the point that the cost per dollar of market cap is higher for a PoS coin
This is not true as there are versions do PoS that it doesn't matter how long you hold the coins. Delegated proof of stake for instance destroys fees and that is how the companies dividends are paid, thus it doesn't matter how long you've held the coins or if you buy something with them. BitsharesX is one of the very few coins that is truly deflationary at this point in time. A lot of coins claim to be deflationary... cough... Bitcoin... but they will be inflationary during our life time.
I am 100% with the OP, PoW is not the best solution to decentralized consensus any longer due to numerous reasons.
I would argue that there are simply too many ways to manipulate PoS crypto coins for any of them to be secure. When you have a PoW coin you must invest in some kind of machine in order to mine and it will take some amount of time of using the machine to secure the network before you will have earned your initial investment back. This gives the miners an incentive not to attack a PoW network even if they have the capacity to do so.
This is one thigh that annoys me about the PoW vs PoS debate. People saying PoW is so much more secure than PoS, because it is not entirely true as both have their strengths and weaknesses. People have been claiming PoS is insecure for years, yet no one has been able to successfully attack it. A lot of people don't realize that there are many different versions of PoS also, some more secure or better than others. There have been something like 10 PoS algos released (or in development) this year alone. Older versions were less secure than newer versions, but that is just natural as people improve and expand upon the original PoS innovation.
You are right that people investing in mining hardware are unlikely to attack PoW. I think that is an unlikely attack vector. A more likely scenario would be a government or large global bank pay for someone to root the discus fish and/or bitfury mining pools, or build a large farm themselves. Due to economies of scale that ASICs provide, it is economically feasible that over 100 countries in the world could attack Bitcoin with no support from any other country. Furthermore, it is mathematically likely to be able to attack the Bitcoin network with as little as 30% of the hashing power with a reasonable success rate. I think most people read 51% attack and assume because of the name that someone would need 51% of the network, but mathematics and probability will show that is untrue. A prolonged double spend attack would crush the confidence in Bitcoin, brining it to its knees.