He needs revenue right now, and the other monetization efforts haven't panned out (blue checks, advertising, etc.). So he's trying to make a quick buck from user-generated data now that the LLM rush is at an all time high. For that he needs to limit access to the data in the first place, otherwise nobody would be paying for publicly available data.
This is how death spirals work. Twitter was treading water a year ago. Taking on lots of debt caused mass layoffs, worrying people about site stability and making them investigate other platforms. A lot of bad tweets scared away advertisers, causing more layoffs and desperation. Signed-in users engage more, so make all users signed-in users (no logical fallacies here).
Incidentally, has there ever been a leveraged buyout that wasn’t a disaster? Presumably there must be some which are a success at least for the buyer, or people wouldn’t keep doing it, but it really doesn’t seem to have a great track record for anyone involved, as a strategy.
"As part of the deal, Twitter will add about $13 billion of debt. Analysts estimate, based on terms previously laid out in documents related to the transaction, that Twitter would be on the hook for annual interest payments of more than $1 billion, compared with some $51 million in 2021."
> Why is Twitter forced to hold the debt for being purchased and not Elon?
Just because that's what was written into the contract when the acquisition happened. Something different could have been put in the contract, but this is what was actually written in it and signed off on by all the parties to the deal.
Saying he was "forced to overpay" isn't a fair characterization. As a joke that went too far, he voluntarily offered to overpay at what in hindsight was the peak, then signed a contract making it official. He was only forced to uphold the contract.
Twitter is full of low-quality garbage, partly due to the culture and partly due to the format. I am very doubtful that it has much value for LLM pretraining OR fine-tuning.
That’s a good point. I think (in jest) the data is invaluable for training automated trolls, for which various state actors should be willing to pay good money.
But more seriously, maybe the value is in it being a real-time source of new data, albeit the signal is drowned in noise like a needle in a hay stack.
Twitter is still $20B away from being in the green from even the most generous estimates. (Elon himself)
Currently Twitter is in the red, and he needs it to not be that and to generate a multi-billion surplus to pay back the investors he took up to finance this.
> Twitter is still $20B away from being (profitable)
Can you explain how Twitter manages to spend $20B per year?
Say they have 1000 employees @ $200k/year, that's $200m/year (1/5th of 1B). Where are the remaining $19.8B being spent? That figure doesn't pass the sniff test.
That’s because you’ve misunderstood. Twitter is worth $20B. We don’t know what they spend nor how much they make as a profit because they’re a private company.
What I’m saying is for Elon to even break even on the deal, using the information we have access to, he needs its evaluations to reach at least $40B. Once it reaches that, it actually proves itself as having been at minimum a neutral purchase. Albeit the real numbers it has to reach and how much profit it needs are both currently unknown.
Correlation does not equal causation. Although unprofitable companies might be inclined layoff employees, it's not implied that layoffs are a sign of unprofitability (in fact, the opposite could be true: laying off employees whose marginal value is low could cause a company to be more profitable).
