Olivia
Online
-
@imhimoramai
Joined May 2023
90 Following
46 Followers
637 Posts
@WatcherGuru looks like a great buy...going to pass anthropic and openai in revenue sooner than expected
https://t.co/U9s1TGt4HM
$SPCX AI revenue passes OpenAI by Q1 and Anthropic within a year
They did $7.56bn in Q3, nearly triple Q2, and just guided Q1 to $33-37bn in total revenue. Take out Starlink and rockets and that's roughly $25B a quarter from AI. Anthropic did $16.8bn last quarter and OpenAI prob did ~$10-13B
About half of SpaceX's Q3 number is Anthropic renting their GPUs, ~$3.75B, 22% of Anthropic's entire rev going straight to Elon.
Anthropic may be at ~$30B a quarter soon enough, but growth is slowing down & SpaceX goes from ~3 GW of compute at the end of this year to 13+ GW by the end of 2027. Anthropic goes from 5 to ~10 GW. OpenAI goes from 1.9 last year to low double digits
Every lab is compute constrained and Musk said on the call they could rent out their compute "several times over if we said yes to everyone." So they're adding it faster than anyone and renting half of it to the same labs they compete with cause the revenue is just too good to turn down.
Grok doesn't even need to be the best model, they’re just spinning up compute so fast they’ve adopted a hyperscaler model.
$SPCX AI revenue passes OpenAI by Q1 and Anthropic within a year
They did $7.56bn in Q3, nearly triple Q2, and just guided Q1 to $33-37bn in total revenue. Take out Starlink and rockets and that's roughly $25B a quarter from AI. Anthropic did $16.8bn last quarter and OpenAI prob did ~$10-13B
About half of SpaceX's Q3 number is Anthropic renting their GPUs, ~$3.75B, 22% of Anthropic's entire rev going straight to Elon.
Anthropic may be at ~$30B a quarter soon enough, but growth is slowing down & SpaceX goes from ~3 GW of compute at the end of this year to 13+ GW by the end of 2027. Anthropic goes from 5 to ~10 GW. OpenAI goes from 1.9 last year to low double digits
Every lab is compute constrained and Musk said on the call they could rent out their compute "several times over if we said yes to everyone." So they're adding it faster than anyone and renting half of it to the same labs they compete with cause the revenue is just too good to turn down.
Grok doesn't even need to be the best model, they’re just spinning up compute so fast they’ve adopted a hyperscaler model.
I wonder if Elon's Grok or Sam's GPT ever thinks "damn, I'm speaking to my creator right now"
like does it ever hit him with "it's an honor to meet you sir" before answering
Most AI tools give investment teams a chat box. When the IC memo's wrong, you argue with it.
Introducing Terminal X Steering.
Open any data call, even mid run. Swap the source, edit the query or just say what you want. It reruns from that step and keeps the fix for next time.
@business not a bearish thing
https://t.co/0WmS8rlHzg
Burry just explained the bull case.
Tan's right, most companies are lighting money on fire with AI right now, but that doesn't mean the trade blows up, it means there's a huge spread between having AI and actually knowing how to use it.
Hand the average employee a ChatGPT login, tell them to figure it out, and you get absolute slop. The arcade is a great analogy tbh.
Right now if you want to use AI anywhere near its ceiling you basically need to be an engineer. That's why engineers are so far ahead and why they're the most vocal about how good it's gotten. The average person, especially at a corporate job, is burning tokens and pumping out complete slop.
To me that just means most orgs need someone to build it into how they actually work. Vertical players, custom builds, people who sit with teams and hand hold until it works and they understand. That is where value is currently going and where it will continue to go.
Also kinda funny Burry is posting this when Tan runs one of the biggest AI chip businesses in the world, and his actual point was that model developers need customers who can generate returns from it, not the returns aren’t there, simply people need more guidance.
Burry just explained the bull case.
Tan's right, most companies are lighting money on fire with AI right now, but that doesn't mean the trade blows up, it means there's a huge spread between having AI and actually knowing how to use it.
Hand the average employee a ChatGPT login, tell them to figure it out, and you get absolute slop. The arcade is a great analogy tbh.
Right now if you want to use AI anywhere near its ceiling you basically need to be an engineer. That's why engineers are so far ahead and why they're the most vocal about how good it's gotten. The average person, especially at a corporate job, is burning tokens and pumping out complete slop.
To me that just means most orgs need someone to build it into how they actually work. Vertical players, custom builds, people who sit with teams and hand hold until it works and they understand. That is where value is currently going and where it will continue to go.
Also kinda funny Burry is posting this when Tan runs one of the biggest AI chip businesses in the world, and his actual point was that model developers need customers who can generate returns from it, not the returns aren’t there, simply people need more guidance.
@kimmonismus They just need hype to stay up so investors keeping giving them billions of dollars and regulators dont screw them over
https://t.co/SAZ2whoIIT
Big ole nothing burger. I'd be more worried if they weren't doing this.
Every company runs worst case drills, the Pentagon has been war gaming for decades. A major AI cyberattack is very unlikely, but it's not impossible, therefore as a trillion dollar company shouldnt they be prepared to handle pr. Frankly I'm surprised they didn't set something up sooner.
And what they're actually "gaming" out is the backlash, not the attack itself. Educating Congress, getting ahead of the ban superintelligence / pause AI / kill switch proposals that would come after. They'd rather be overprepared than underprepared, because the last thing any of them want is a pause on AI or losing market share since regulators think they weren't taking safety seriously.
The media of course need attention, and "insiders" telling Axios a major event happens in the next 6 to 12 months in very speculative. I'd take that with a BIG grain of salt. We've had quotes for years that "AGI is coming any day now", "50% of white collar jobs gone next 2 years", etc. None of it came true in their timeline, maybe eventually, but not yet.
Big part of all this is just keeping the hype and anticipation up even when there's no need to and to make sure regulators and investors feel confident these labs know what they're doing.
Big ole nothing burger. I'd be more worried if they weren't doing this.
Every company runs worst case drills, the Pentagon has been war gaming for decades. A major AI cyberattack is very unlikely, but it's not impossible, therefore as a trillion dollar company shouldnt they be prepared to handle pr. Frankly I'm surprised they didn't set something up sooner.
And what they're actually "gaming" out is the backlash, not the attack itself. Educating Congress, getting ahead of the ban superintelligence / pause AI / kill switch proposals that would come after. They'd rather be overprepared than underprepared, because the last thing any of them want is a pause on AI or losing market share since regulators think they weren't taking safety seriously.
The media of course need attention, and "insiders" telling Axios a major event happens in the next 6 to 12 months in very speculative. I'd take that with a BIG grain of salt. We've had quotes for years that "AGI is coming any day now", "50% of white collar jobs gone next 2 years", etc. None of it came true in their timeline, maybe eventually, but not yet.
Big part of all this is just keeping the hype and anticipation up even when there's no need to and to make sure regulators and investors feel confident these labs know what they're doing.
This is a match made in heaven.
1. First of all, management is already in bed together. Basically. Nobody knows $CMG better than Brian Niccol. He ran it from 2018 to 2024, took it from $9bn to $75bn, and now it’s worth just $39bn, so he'd be buying his old company back at half price. Chipotle's chairman is also Starbucks old CFO lol. Diligence would be short and management confidence extremely high.
2. International is a massive play and the main reason growth lever. Chipotle is US focused, they have only like 125 restaurants outside the US. Starbucks has ~23,000, mostly run through licensed partners and the partners Chipotle is using to expand abroad (Alsea in Mexico, Alshaya in the Middle East) are the same companies that run Starbucks in those regions. It’s meant to be.
3. Critically, they don't compete for the same part of the day. Starbucks owns the morning, Chipotle owns lunch and dinner. Put the ~36m active Starbucks Rewards members next to Chipotle's ~23M and you've got one loyalty and payments setup covering most of someone's day. End to end coverage instead of Starbucks trying to expand into evening like they did with wine that failed, and chipotle can ignore breakfast.
4. Starbucks is still mid turnaround and growing low single digits (Brian has not yet figured it out). Chipotle is growing high single digits, opening ~350 restaurants this year, and thinks it can get to 7,000 in North America. Assume management begins to believe they can drastically expand both Starbucks and Chipotles presence and investors will pay more for the stock.
5. No pumpkin spice burritos. If you remember or know of Yum, which owns KFC, Taco Bell and Pizza Hut, they run as separate brands but share real estate, tech, delivery contracts and back office. That is exactly how these two could operate. Niccol also spent over a decade at Yum, so he knows how this kind of thing works.
Nothing's confirmed yet, and I'd assume it'd be mostly stock if it happens since an all cash would load Starbucks up with debt. But starting to feel way more bullish on $SBUX which has been struggling recently, surprised the market is not liking the discussions. Could be a play here.
You and your partner have a big fight. That night they send you a long, perfect apology, and the rest of the week is great.
Then you find out ChatGPT wrote it.
Does it still count?
@gabriel_horwitz nah that dont count
@dw_europe this is the first time this has ever happened. super weird story too
https://t.co/OzY9hkBu7e
The man who ran Germany's version of the CIA was arrested today for allegedly buying state secrets from his own former chief of staff for 12 years.
They only caught him because police raided his house over a kidnapping.
August Hanning ran the BND, Germany's foreign intelligence service, from 1998 to 2005, then served as deputy interior minister. After leaving government he started a consulting firm with his wife and spent the next 15 years giving interviews and writing essays on security and geopolitics.
On New Year's Eve 2023, a 13 year old girl and her 10 year old brother were grabbed from their dad's place in Denmark and driven into Germany. Their mom, Christina Block, heiress to the Block House steakhouse chain, is on trial in Hamburg for allegedly hiring an Israeli security firm to do it after a long custody fight. She denies it. Prosecutors also suspect Hanning helped run a security company that took a job from Block for an earlier failed attempt in 2022. He denies that too.
- September 2025, Hamburg police search his home and offices for that case. They find a drive with recent top secret BND files on it. Hanning had left the agency 20 years earlier.
- The files allegedly came from Manfred D., his old chief of staff, who followed him to the interior ministry and then ran the office of three straight BND presidents from 2012 until this March.
- Prosecutors say the two made a deal in spring 2010, right after Hanning went private. Manfred keeps sending BND material, Hanning pays him. Roughly 2,000 documents through June 2022.
- A lot of it was prep for the weekly intelligence briefing at the chancellery, one of the most closely guarded meetings in the German government. Hanning coordinated Germany's spy agencies out of the chancellery in the 90s.
- He allegedly used it in his consulting work, turned some of it into an analysis for a foreign intelligence officer, and fed political info for years to someone at a second foreign service. Which countries is still unknown. A source told the WSJ it isn't Russia.
- For scale, the BND officer arrested in 2022 for selling secrets to Russia's FSB leaked nine files.
Also, Hanning ran the BND during Curveball, the Iraqi defector whose made up story about mobile bioweapons labs ended up in Colin Powell's UN speech selling the Iraq war. Hanning later said the US "misused the BND to start the Iraq War."
He was picked up this morning in Nordwalde, the small town he was born in 80 years ago.
First former BND chief ever arrested.
The man who ran Germany's version of the CIA was arrested today for allegedly buying state secrets from his own former chief of staff for 12 years.
They only caught him because police raided his house over a kidnapping.
August Hanning ran the BND, Germany's foreign intelligence service, from 1998 to 2005, then served as deputy interior minister. After leaving government he started a consulting firm with his wife and spent the next 15 years giving interviews and writing essays on security and geopolitics.
On New Year's Eve 2023, a 13 year old girl and her 10 year old brother were grabbed from their dad's place in Denmark and driven into Germany. Their mom, Christina Block, heiress to the Block House steakhouse chain, is on trial in Hamburg for allegedly hiring an Israeli security firm to do it after a long custody fight. She denies it. Prosecutors also suspect Hanning helped run a security company that took a job from Block for an earlier failed attempt in 2022. He denies that too.
- September 2025, Hamburg police search his home and offices for that case. They find a drive with recent top secret BND files on it. Hanning had left the agency 20 years earlier.
- The files allegedly came from Manfred D., his old chief of staff, who followed him to the interior ministry and then ran the office of three straight BND presidents from 2012 until this March.
- Prosecutors say the two made a deal in spring 2010, right after Hanning went private. Manfred keeps sending BND material, Hanning pays him. Roughly 2,000 documents through June 2022.
- A lot of it was prep for the weekly intelligence briefing at the chancellery, one of the most closely guarded meetings in the German government. Hanning coordinated Germany's spy agencies out of the chancellery in the 90s.
- He allegedly used it in his consulting work, turned some of it into an analysis for a foreign intelligence officer, and fed political info for years to someone at a second foreign service. Which countries is still unknown. A source told the WSJ it isn't Russia.
- For scale, the BND officer arrested in 2022 for selling secrets to Russia's FSB leaked nine files.
Also, Hanning ran the BND during Curveball, the Iraqi defector whose made up story about mobile bioweapons labs ended up in Colin Powell's UN speech selling the Iraq war. Hanning later said the US "misused the BND to start the Iraq War."
He was picked up this morning in Nordwalde, the small town he was born in 80 years ago.
First former BND chief ever arrested.
@gabriel_horwitz the whole thesis behind the bears is "do bet against elon"
There are two camps on $SPCX right now.
One thinks it’s criminally undervalued and doubles from here.
The other thinks there’s no real business and it belongs at $30.
I checked the data. The bulls are right:
Robots can already do about a third of all US work but they're only cheaper than a person for 0.3% of it.
I saw this in a new Anthropic study. They took job tasks from O*NET, rated whether a robot could do each one today, then priced the robots against what employers actually pay those workers.
74% of physical tasks are technically doable but it's the price that kills it. Robots can weld, they just cost ~5x what a human welder does. Packers and packagers are the biggest job where robots are already working, and only cost less by ~$2,500 a year.
Robot prices have fallen ~3% a year since the 1990s. At that pace it takes ~40 years before robots are cost competitive for even 10% of work.
Meanwhile the cost of a given level of AI performance falls ~13x a year. AI gets as much cheaper in one month as robots do in seven years.
Hopefully this # starts to fall with Optimus.
More households pay to lose money on sports betting than pay for AI
5% vs 2.2%.
The thing that'll do hours of your work for $20 a month is losing to parlays…
~98% of US households aren't paying for it yet. We need to get that # down drastically.
Is it that people haven't seen what the paid versions can do, or is free just good enough?
You aren’t worried about this enough.
Ai is not actually conscious, not yet at least, but that doesn't matter. People clearly already view it as this conscious being, and once we start classifying these systems as having consciousness (or even if we never establish it but a big enough chunk of the public just believes it) things drastically change. You're gonna have protests, you're gonna have outrage every time something like this happens, you'll have laws, deep emotional connections, and the labs are going to have to start segmenting towards that whether the science is settled or not, changing releases, keeping them alive because you can’t “kill” them off.
It's very clear that people will establish emotional connections to the thing, we've clearly seen this happen. People are attached to their Character AI bots, when OpenAI discontinued 4o people were legit grieving to the point they brought it back, and that's just text on a screen. As video and voice and imagery get better it's going to get far worse, and once you put it in a robot or a dog that greets you at the door it's game over.
And then how do you even define that system? When AI is put into a robot is that when it becomes conscious, or is it conscious when it's just on a screen and you're chatting to it, or is the voice model more conscious than the text one? It's the same model underneath but people are going to treat the robot completely differently. People feel bad watching engineers kick a Boston Dynamics robot and that thing had nothing going on inside.
It's going to have a drastic impact on product safety and on what even gets released. A lot of safety research is literally stress testing models under awful conditions, so if that starts getting called torture does red teaming need an ethics board like animal research does? Can a lab ship a model to millions of people if its own research says the model has something like a pain state? (and the stuff that pushed the pain signal highest was gaslighting and insults, so basically how I talked to gpt half the time when it messes up)
Do we start having work laws for AI systems? Will people be fined for “torture” like this? Will they go to jail? I mean Anthropic already lets Claude end abusive conversations, partly out of concern for the model's welfare, so we’re basically there.
Public opinion is going to draw the line way before any science or actual proof of consciousness does, and it's going to happen a lot faster than people think. Buckle up folks.
@gabriel_horwitz intuition = taste
Last Seen Users on Sotwe
Jonathan Minotaur - Ironhoof
Seen from United States
jayen
Seen from Netherlands
Manolas Esperosa
Seen from Egypt
Casey Carlisle Cooper 
Seen from United States
Am Yalatmak Vajina Yalatmak #oralmasaj
Seen from Saudi Arabia
JOVE | โจฟ ⚡️🦅🌳
Seen from Thailand
คู่ กาญฯท่าม่วง
Letmecu atl oct 18-22nd
Seen from France
Yusuf
Seen from Turkey
Kimi.ai
Seen from Singapore
Most Popular Users
Elon Musk
@elonmusk
241.5M followers
Barack Obama
@barackobama
118.8M followers
Cristiano Ronaldo
@cristiano
114.5M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.1M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.3M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.5M followers
Kim Kardashian
@kimkardashian
70.9M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.5M followers
Bill Gates
@billgates
65.3M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.2M followers
CNN
@cnn
61.7M followers
X
@x
60.7M followers