Because LAN (and dedicated servers, by effect) allows people to bypass the publishers servers, and makes it easy to still play multiplayer even though the game is "cracked" (DRM bypass).
Most game publishers do whatever they can to avoid making it feasible to play cracked games, even if some of the techniques they use hurt players who actually bought the game legally.
Also, LAN parties.
AFAIK, anyone with an old disc of Assassin's Creed 2 for Xbox 360 will still be able to play the single-player this Thursday. A disc for The Crew will become just junk when the server is killed.
I don't know what the solution is because a lot of these functions would require complex DRM systems to make a digital product behave anything like a physical one. So legislating to enforce digital consumer rights is effectively legislating for DRM everywhere. The crypto dream of tying every digital good to a token or NFT was effectively an implementation of this kind of DRM using blockchain. However many digital storefronts already implement complex DRM systems, so perhaps at least enforcing some kind of right to return digital goods (including game MTX) would be beneficial.
Ultimately, for business reasons:
1. A good LAN-play UX requires game/host discovery; solutions involving network broadcast will only work in small (i.e. home) LAN environments where everyone is in the same broadcast-domain - but this won't work in larger networks, which then means players will need to manually type-in IP addresses (which is acceptable for IPv4, but less-so in IPv6). Using an internet-based matchmaking and host/game directory service eliminates this problem when, and thesedays it's incredibly unlikely that a user will have a LAN available without Internet access. Internet-disconnected LAN-parties haven't really been a thing since ~2005 (excepting military/naval/antarctic scenarios) so the business-case just isn't there.
2. Providing technical support for games costs money; speaking from my second-hand experience with a major console game publisher: when a (console) game features LAN-play (with discovery/broadcast/manual-IP-addresses) it's inevitable the company will receive support calls/tickets from people having issues with their home networks that the company can't really do anything about; even if there's an explicit warning to users that the game's technical support team won't assist with local network problems they'll still receive those calls and need to triage (and reject) them, which all costs money.
3. The vast majority of a game's players aren't interested in such a feature; so when a game has a dev budget, it's just good business-sense to spend that budget on things that bring the most revenue, such as fixing bugs/glitches (which directly lead to bad IGN review scores and purchases legitimately demanding costly refunds) instead of netplay features (and omitting LAN-play isn't going to lose them that many potential customers; I suspect no-one already in the market for a AAA game is going to not buy it solely for that reason alone).
They couldn't play a multiplayer game because right at release, the servers were overloaded and the new version did not have a LAN mode at launch.
The whole purpose of the event was to have a kind of LAN / multiplayer gaming with the community and stream it.
It's broken and they don't seem to care, which is a huge shame.
There is absolutely zero incentive to add LAN to a game early, mid, or late stage. It does nothing but cost money and reduce profit margins.
As such, you will almost never see it on any large scale service oriented multiplayer game.