The fact that everyone is glazing Wiles so much on her way out, including the guy most sabotaged by her (RFK Jr), indicates that her mob is still effectively running the domestic policy of this country.
Steve Miller isn’t going to last long w/o purging a lot of ppl fast.
PS. Losing Susie Wiles when the most liberal Congress in history is about to sit is very bad news for the US pharmaceutical industry.
ANT could, even today, easily command a $1T public valuation (easily $1.5T if they got rid of their ridiculous supervoting founder LLC).
But OAI, to my naked eye, is the Lyft in this 2-horse race.
They stayed private too long and are still losing way too much money.
If NYT reporting is true ("Softbank to reportedly invest several hundred million dollars into OpenAI"), OpenAI is on death watch, IMO...
OAI has 55-60% of ANT's token spend share (per Ramp) and is only maintaining that via $10B+ cash burn per qtr. If they don't complete their current round, they need to gut subsidies and concede significantly more market share to ANT
If SB won't invest more in them, and ANT's IPO valuation is moving down to $1.25T, OAI basically shouldn't trade above its prior round valuation of $852B
If Dems were smart and went after the genuinely corrupt parts of the Trump Admin, they’d be shocked by the amount of GOP support they’d get
Instead they’ll probably go after the good guys (yes that includes Elon) and rally everyone back to Trump and his Mar a Lago clique
Harmeet @HarmeetKDhillon was one of our best, a disciplined operator who got a lot done while staying in the WH's good graces, and squeaky clean to boot.
Feel like this is the first of a deluge of competent-technocrat exits from the Administration.
Harmeet Dhillon plans to step down as head of the Justice Department's (DOJ) Civil Rights Division, Bloomberg Law reports, citing four people familiar. They say she told allies she would leave after the midterms, and two say her announcement could come as soon as today.
She would be the first senior DOJ official to resign since Democrats’ midterm wins, Bloomberg says, and former officials told it the department is bracing for other top political appointees to leave before the new year.
Under Dhillon the division has focused on religious liberty and created a Second Amendment section. In August she told Bloomberg she planned to return to her California law firm after DOJ.
Source: news.bloomberglaw.com/us-law…
Feeling pretty good about this call.
-25% in 3 weeks. Probably down another 50% by mid-December.
It's never not been a good time to short TRUMP ($1.5 bn unlocked market cap, $2 bn FDV), but the timing now seems too good to be true.
1. 10% of float unlocking in 5 days
2. Remaining tokens unlocking on Dec. 18th
3. 462m tokens (46%) already unlocked, but unsold
4. Midterm elections = Lame duck presidency incoming; NPV of unsold Trump favors will collapse
5. Funding rate extremely normal: somehow this isn't crowded
The other big takeaway is that polling error underestimated Trump disapproval among his own base. It was at 30 before the election. After these results, it could easily be 50.
GOP losing SC means Iowa is gone. Even Nebraska is in play.
8 Dem Senate seats means the GOP won't control the Senate until 2030, federal policy gridlock, and SCOTUS will lose its conservative tilt.
All for the Mar-a-Lago Mafia and their stupid war in Iran
Dems winning TX Gov would signal an absolute GOP bloodbath. Terrible for the AI trade in stock markets tomorrow.
Comprehensive FL data indicates a 13-pt Dem swing, extremely consistent with national polling.
Big night for @NateSilver538 (and anyone else who can read data in a remotely unbiased manner) ahead.
Anthropic has $6.2b of converted note purchase agreements (NPAs) and 15bn of unconverted NPAs with Amazon for compute, and $20b of Google NPAs
The $6.2b converted NPAs should be added to ANT's 2025/26 compute costs, and its unconverted $35b of NPAs should be added to ANT's 2027-30 compute costs at GAAP cost accounting. They also have $5b of AMD NPAs
$40b / 16 quarters = minimum $3.5b / qtr of additional compute costs that are borne by equity/dilution
On my own numbers (ANT just missed my 3Q numbers btw), adjusting for this, ANT is a bit below breakeven on $200b of end-2028 ARR. And I'm not accounting at all for recurring "non-operating costs" like charitable donations
To get juicy 2028+ margins, it seems like you have to just assume that Anthropic's training costs will drop a lot in the future and ANT will be able to coast on what they've built thus far as their compute cost per dollar of revenue re-inflects. What's the steelman counterargument here?