It's pretty much right. Dig into what it takes to run a social network in most European countries and you'll hit at minimum the following problems:
• Lack of a DMCA equivalent. DMCA lays out a lightweight process for platforms to process copyright disputes which if they follow it will avoid legal liability, which is needed on any platform that hosts user generated content. The EU Copyright acts require platforms themselves to enforce copyright and prevent users violating it. This is a gigantic technical implementation problem all by itself. Also, the US has the legal concept of fair use but that's not a concept in much of Europe, so people posting parodies etc thinking it's OK can still create liability problems.
• No equivalent of Section 230. Many new laws that specifically criminalize the hosting of illegal speech, and which don't give any credit for effort. As what's illegal is vague and political in nature you can't make automated systems or even human-driven systems that reliably handle it, so the legal risks are large even with a good faith effort to comply.
• GDPR, "right to be forgotten" and NetzDG style laws have large fixed costs associated with compliance which established companies can absorb but startups can't. For instance it's common for EU lawmakers to demand 24 hour turnaround times, which you can't reliably comply with if you're a one man startup.
• Algorithmic transparency laws, which mean you can't obtain any competitive advantage by better ranking (being good at this is how TikTok got so big), and which can threaten your ability to clear spam or use ML.
• Laws around targeted advertising mean you can't generate revenue comparable to what the US based firms can do, so you can't be competitive and your users will be annoyed by low quality barrel scraping ads for casinos after they click "No" on a consent screen without reading it.
There's probably more. For example, running a commercial search engine or training AI models on the internet is illegal in the UK, because UK copyright law only allows "data mining" for research purposes. There's no way to argue it's fair use like they do in the US. Just one of many such problems off the top of my head.
