I am by no means an expert in any part of this but rather than specific startups I would suggest:
- Startups that provide self driving anything will need to step up their game to maintain investments. Either step up to meet and maintain their promises or reduce expectations and limit deployment to well tested scenarios.
- Startups that use ground based robots to deliver anything of value. The robots and the contents will be stolen at greater rates when criminals learn how to manipulate and steel them. They will eventually fail at the same rate as traditional businesses are pulling out of high crime cities. They may limp along on low crime wealthy cities. I expect these to die the slow painful death that occurred in the .com crash.
- Startups that stretch the capabilities of LLM's in ways that are unrealistic and making promises that would otherwise require AGI that does not exist. These startups are entirely depending on marketing that gives CTO's and CIO's FOMO. Once the CIO publications catch up these will quickly fail.
- Startups that puts most of their operational dependency on a third party LLM vs self hosted as future rate limits or restriction changes make them high risk. The big players can put any of the small startups out of business in less than a week. Small players do not have the human networking with government and can not afford long downtime or the army of lawyers required to survive this. Only startups run by golden children will survive this.
- Anything that is not critical or a hard requirement for people in a wartime economy. Exceptions are probably things that provide escapism. Such as entertainment related, especially if they can be enjoyed offline should the internet or global power grids go offline. People with solar and large fuel supplies can enjoy offline content.
- This one will be unpopular and taboo, but EV's HEV's, drive-by-wire vehicles. These were cart before the horse. The power grid is in no way ready for this. If the power grid would substantially tech-refreshed and grown and or if people had mass adopted very large solar deployments on their homes or if mini-micro-mini-micro redundant and load balancing reactors were deployed in every neighborhood then perhaps this would be less of an issue but there are not enough financially stable home owners for solar to be impactful and nuclear has too many unrealistic regulations to be viable in a neighborhood. Drive by wire will be a big problem in wartime as they will all be compromised and people will lose confidence in them. Small numbers of hack fatalities have been suppressed but in high numbers it can not be hidden as there will be massive indirect bystander casualties. I see most drive-by-wire vehicles as weapons of mass destruction that have been the same as websites in terms of security and that is not even remotely acceptable to me. They will need to be hardened to the nth degree.