Keep on to blur preview images; turn off to show them clearly
RT @doodlestein: @sfmcguire79 You should need to answer 3 of these within 20 seconds at the voting booth. Get it wrong or answer too slowly…
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM
It’s too bad I wasn’t able to monetize these insights better when most people weren’t even thinking about it.
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM
So, MSTR is now trading at a discount to the NAV of its bitcoin holdings, after trading at a massive premium for the last couple of years. The problem is that Saylor is going to need to keep issuing more stock to shore up the balance sheet because of all that debt they’ve taken on. Before, every share sale was anti-dilutive and built up book value while raising liquidity. Now it’s the opposite, and every new share issued is dilutive and destroys book value. The discount is modest now at 1.7%, but all the forces that were helpful before are now working in the opposite direction, so it’s probable that the discount will expand before it contracts. Also, during its peak, MSTR was the only game in town if you wanted a liquid stock that was pursuing this levered bitcoin treasury strategy. But now there are a bunch of copycats competing for the same investors, and those other ones are still in their honeymoon phase of trading above NAV. What Saylor should really do is sell BTC to buy back stock and thereby increase the BTC per share, even if that results in a decrease in the aggregate BTC owned by MSTR. But doing so would likely spook the market (both the BTC market and the MSTR market), potentially erasing any benefit. Tough spot to be in.
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM
This goes back to what @karpathy has said about how increasingly going forward, the audience and user base for software and software documentation is going to consist of AI agents and not people. So companies need to start focusing more on usability and ergonomics for the robots.
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM
The scary part is that currently this kind of knowledge and expertise is rare and expensive. So unless you have state secrets or blueprints to a DRAM fab or the newest GE jet engine, you don’t have to worry too much about it. But AI agents trained on all the latest exploits make this ability much cheaper and more available, so expect to see a lot more of it. Fortunately, the same AI agents can be used to help you lock down and audit and observe your systems and network. Unfortunately, defense is a million times harder than offense! You only need something to work 0.1% of the time offensively to be useful if you’re automatically probing and attacking thousands of machines. But you need your defenses to work 100% of the time to stay secure.
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM
There are a million clever ways an advanced attacker can compromise you for years without you knowing, even if you think you’re checking everything carefully, locking down networks. That’s why most corporate IT security is more about security theater and “CYA” than anything else.
The scary part is that currently this kind of knowledge and expertise is rare and expensive. So unless you have state secrets or blueprints to a DRAM fab or the newest GE jet engine, you don’t have to worry too much about it. But AI agents trained on all the latest exploits make this ability much cheaper and more available, so expect to see a lot more of it. Fortunately, the same AI agents can be used to help you lock down and audit and observe your systems and network. Unfortunately, defense is a million times harder than offense! You only need something to work 0.1% of the time offensively to be useful if you’re automatically probing and attacking thousands of machines. But you need your defenses to work 100% of the time to stay secure.