Peter Thiel sells his Nvidia and Tesla shares. Sign of things to come?

[Pazzy]

How @BaggerofTea is still on here and people dapping that agent is beyond me. dude is a certified weirdo fraud. He really in here trying to gaslight people thinking everybody here is dumb.

 

[Morethan1]

 

[the bossman]

The real danger is that no one has shown a way to make the money back. At least with smartphone tech you could sell them directly to consumers and can liquidate the used products to other markets or pivot the architectureto other uses. AI is a primarily commercial tool. For individual consumers its just making gimmicky videos and enhancing chat bots. No one has produced a profitability model that will recoup the investment. The once the chips are ran through they are practically useless as well so you dont have a resource liquidation model to decrease the initial cost of investment with a backend divestment strategy.

Yo it's funny how this is almost word for word the damn near EXACT same set of arguments that all the Wall Street analysts had about Bezos when he started building AWS. It took nearly 7-8 years before AWS started making real money for Amazon. Man back then all the Wall street guru analysts were saying he should stick to his retail website and that they couldn't see any path to how all this hundreds of millions in spending would ever produce a profit.

Also acting like AI will only ever amount to deepfakes and chatbots shows a real lack of imagination. Anybody who's been using Chatgpt since the early days from version one to today sees the huge jump in capability. From shyts and giggles to using it everyday for work. It's really not hard to take that experience and then picture how much better AI will get in a few more years especially given it's just getting started.

So according to your logic, if consumer value has to exist on day one, how do you explain cloud’s early years when it looked exactly the same? There were no consumer apps. No resale value for those servers that become old. No obvious profitability model. Uber and Spotify did not show up at launch. Wasn't until almost ten years later. By your standard Bezos should have shut down AWS entirely in the beginning and we would all be stuck in a world without Netflix streaming or any of the modern services that rely on cloud. All because the profitability model didn't fit into some perfect little 5 year wall street excel model spreadsheet on day one.

If things are so good can you tell me what the model to profitability with these companies (outside Nvidia) to increase profitability in a way that doesnt complete wreck the American economy?

Because Sam Altman hasn't come up with one. Mark Zuckerbergs AI related ideas have all failed and Alex Karp brings up politics every time you ask him about his companies business fundamentals.

TLDR: "They'll figure it out" aint a plan or a strategy, breh. The American economy outside this spending frenzy isnt healthy. And if this doesnt work it'll tank the economy because Trillions have been spent on this and its responsible for most of the growth in the last 2 years.

You really think Bezos could've predicted Uber or Instagram or DoorDash back in 06? He couldn't. Because the infrastructure did not exist yet. It was being built. Same with AI. You do not get the killer apps until the foundation is stable and cheap. We are in that buildout phase. There are going to be new economic models created from all this shyt that we haven't even imagined yet.

 

[bnew]

AI in its current form has a very finite ceiling.
Nothing more than a chatbot hooked up to a dictionary of the internet and trained to give optimum responses.

There is no organic or novel thought in a large language model. Just data and pavlovian like training.
They are probably starting to see it based off of the lack growth from the r and d side.


Brain computer interface is where it is at.
wrong.

Spoiler: full text & large images

1/37

@GoogleAI

Today we introduce an AI co-scientist system, designed to go beyond deep research tools to aid scientists in generating novel hypotheses & research strategies. Learn more, including how to join the Trusted Tester Program, at Accelerating scientific breakthroughs with an AI co-scientist


https://video.twimg.com/ext_tw_video/1892214105580220417/pu/vid/avc1/830x470/iGQajDvaaMm8a0tN.mp4
2/37

@rohanpaul_ai

Beautiful..
"The AI co-scientist parses the assigned goal into a research plan configuration, managed by a Supervisor agent."

3/37

@Chaos2Cured

You only give this to “trusted research labs” which means the very same people who haven’t thought differently are still being rewarded.
Make it public.
My work is needed, but I am not in the “elite” circle.
This has GOT to change.
AI allows EVERYONE to innovate. By lowering who can use it drastically limits what can be discovered. •


4/37

@AISafetyMemes

Cool! We're approaching the final chapter where people can stay in denial that the shoggoths are Mere Tools


5/37

@olexandr

This was already done @CarnegieMellon over a year ago: Autonomous chemical research with large language models - Nature


6/37

@IndraVahan

and we're already beyond deep research. my god it's so hard to keep up now.


7/37

@ZayaAI_PathDx

Well, well, well..what an amazing update from GoogleAI, seems appropriate to do a reminder soon on our page .


8/37

@Molecule_sci

This could accelerate discovery like never before. Excited to see it in action


9/37

@koltregaskes

Very interesting.


10/37

@GithubProjects




11/37

@altryne

Will mention this in our @thursdai_pod podcast tomorrow


12/37

@Purring_Lynx

looks like ai agents won again


13/37

@GaleonCare

AI accelerating research, hypothesis generation… and maybe one day revolutionizing healthcare entirely?
At Galeon, we see AI as a key ally in structuring & securing health data for better patient care.


14/37

@bevenky

Here we are. Last 30 days feels like a decade of progress with one model after another and other large breakthroughs happening in AI by all companies large and small.


15/37

@AndrewVoirol

Wow, that sounds extremely exciting. You have my submission as an additional trusted tester or an extension of my current trusted tester status. I look forward to hearing from someone or some thing soon.


16/37

@ripchillpill

An AI co-scientist that can generate new hypotheses could totally reshape how we do research


17/37

@JoshuaLelon

what tool did y'all use to mae those animations?


18/37

@Benzinga

19/37

@pr0me

deep, deeper, beyond


20/37

@pilot3ai


[Quoted tweet]

Token Use Case #1: Unlock Premium AI Tools
Holding $PTAI gives you exclusive access to advanced AI trading tools, smarter strategies & high-value insights.

Optimize your performance.
Trade smarter, profit more.

#Pilot3 #PTAI #AI

21/37

@azed_ai

Impressive!


22/37

@vivek_naskar

This is interesting.


23/37

@bio_hacker_dao

Is this the real google though lol


24/37

@Newaiworld_

"go beyond deep Research tools"
Do you already have some insights about what we're talking about?
I mean going beyond would mean AGI:
Generalize knowledge to come up with new ideas and hypotheses -> AGI


25/37

@AntDX316

Every University will need to change very very fast.
The ASI-Singularity(Godsend) is the only Global Solution, people.


26/37

@ApeToTheThesis

@GoogleAI’s co-scientist’ is an exciting step, but it raises a key question—how much should AI-driven research rely on centralized systems?

@Isaaceditor_ takes a different approach: decentralized, autonomous agents that don’t just assist but explore and cross-reference discoveries across fields.


27/37

@AGI_FromWalmart

@demishassabis 's benchmark about AGI getting closer and closer
[Quoted tweet]

the only goalpost that will actually stand & won't be moved is what Demis hassabis said:
"AGI is robust across all cognitive tasks and can invent its own hypotheses and conjectures about science"
in short when you saw an AI come up with something on par with theory of relativity then you know it's AGI


28/37

@abhivendra

This AI co-scientist system from @GoogleAI is a pivotal step in harnessing AI's potential to redefine research. By empowering scientists to generate novel hypotheses, we push the boundaries of innovation. Let's apply similar thinking to education.


29/37

@bramiozo

Why would a scientist want this? The last worry on a scientists mind is “do I have enough research ideas”.
Maybe this is not meant for scientists, but for managers?


30/37

@sonicshifts

Oh wow incredible. Also, don't sleep on Google's NotebookLM it's fantastic for organizing and understanding research.


31/37

@l8ntlabsAI

The AI co-scientist system is a game-changer, I've been experimenting with similar concepts in my own work at L8NT LABS. The potential for AI to aid in generating novel hypotheses and research strategies is vast, and I'm excited to see where this technology takes us.


32/37

@NtisiZyyduu

ArcBlock's Advantages:
Seamless blockchain integration

/search?q=#ABT - More than just a token

Decentralized ID with @did_wallet

Zero gas fees, stake 1 ABT

Versatile - AI, Cloud Computing...
Join the future with /search?q=#ArcBlock and /search?q=#ABT!
---


33/37

@EAccelerate_42

People got stuck in string theory and quantum gravity for so long hopefully with AI we can move forward or take a different route towards theory of everything


34/37

@EnfiqOfficial

An AI co-scientist? That’s next-level! Excited to see how this transforms research and innovation !!


35/37

@7845_f9

36/37

@dhtikna

Im sorry but this smells very fake and geh, non technical PM vibes


37/37

@facevoiceai

Impressive approach, @GoogleAI! Curious if your AI co-scientist could eventually integrate real-time emotional or behavioral data to refine hypotheses and reduce research bias. How might this shape the future of scientific discovery?

Spoiler: thread continued

Spoiler: full text & large images

1/12

@deredleritt3r

Current AI models are already able to generate novel scientific breakthroughs. A well-built multi-agent system, like Google's AI Co-Scientist, is all that's needed to make this work.
See the thread below regarding a novel scientific breakthrough independently discovered by AI Co-Scientist back in February. This was back when AI Co-Scientist was based on Gemini 2.0(!).
This is also why I barely care whether AGI really is 10 years away. We've barely scratched the surface even with the current generation of LLMs, and it seems nearly certain that AI will have a profound impact on the economy even if all progress stops today.
[Quoted tweet]

Google and Imperial College of Medicine have published a paper showing how Google's AI Co-Scientist independently discovered a scientific breakthrough (related to how bacteria share resistance genes). This story first broke in February, and is now being validated through a peer-reviewed publication.
Google expects to share validated results of other discoveries made by Co-Scientist over the next few months - those "where our AI co-scientist generated novel, testable hypotheses for complex open-ended problems, showing how the system can meaningfully accelerate scientific research".


2/12

@emrahdma

I've seen several claims on X although never a link to the said claim


3/12

@deredleritt3r

Here is the link to the said claim, which includes the paper:
[Quoted tweet]

During Thanksgiving break last year, our AI co-scientist team @GoogleDeepMind @GoogleResearch - @Mysiak @alan_karthi met Prof @jrpenades @CostaT_Lab of @ImperialCollege.
They were nearing a breakthrough on how bacteria share resistance genes and proposed a test for our AI co-scientist: pose their exact research question to the system and see what it found.
Today in @CellCellPress, we share the results.
Paper link - cell.com/cell/fulltext/S0092…

4/12

@slow_developer

lol
[Quoted tweet]

I hope you have concrete data to corroborate the claim that GPT-5 Pro produced novel insights, and that you’re not just relying on random posts on x


5/12

@thexynapse

That’s wild — feels like we’re watching the start of AI-driven science in real time.


6/12

@AnchoreoAI

We already have powerful LLMs and still continue to strive to get them to reason deeper, collaborate smarter, and discover faster. Completely agree that we’re sitting on huge opportunities even before AGI shows up.


7/12

@mattleta_

the AGI debate is a philosophical luxury. the real work is building the industrial-grade systems required to absorb the output of the models we already have


8/12

@agenticist

Applied General Intelligence


9/12

@wynz87

Norway starting to prepare
[Quoted tweet]

1/5 Norway has launched “The Ten AI Commandments”, inspired by the 1971 “Oil Commandments” that ensured oil wealth benefited the entire population rather than foreign corporations.


10/12

@innvoa

Agreed the term AGI is meant for general use and human machine interfacing, but the true advancements and discoveries may take place before AGI and just by the improvement of LLMs- in the novel /groundbreaking discoveries sense


11/12

@zby

The question never was if they are able to generate because a mechanistic enumerator of all possible texts is able to generate all possible discoveries. I was always about verification and making the search efficient: Why "Out-of-Distribution" Misses the Point About AI Creativity


12/12

@SoftSyncDXB

The pace of AI breakthroughs is incredible! Multi-agent systems like Gemini 2.0 are pushing science forward, and we're just getting started. Looking forward to more headline discoveries. /search?q=#AIDriven

Spoiler: thread continued

Spoiler: full text & large images

1/3

@ChombaBupe

This artificial intelligence (AI) is based on "shredding" existing data & spitting out remixes.
There is nothing to shred when it comes to unsolved problems.
Therefore AI has to wait for someone to do something first so it can't solve cancer nor explore new frontiers on its own
[Quoted tweet]

Director Gore Verbinski says AI is attacking modern storytelling.
“Instead of trying to solve cancer or take us to Mars or these things that could solve some genuine issues, it’s going after storytelling, it’s going after illustrations, it’s gonna write your song for you. It’s like saying it’s gonna breathe for you, it’s gonna f*ck for you. It’s gonna take away.”
(Source: dexerto.com/tv-movies/gore-v…)

2/3

@corncobweb

The first sentence is a good analogy, but the second sentence is false.
Google has AI that creates medical hypotheses and designs experiments to test them. Yes, even for cancer: How a Gemma model helped discover a new potential cancer therapy pathway
Before that, material science was doing a similar thing with Word2Vec.


3/3

@corncobweb

AI has been doing making valuable medical hypotheses that humans cannot since AlphaFold was created in 2018. Protein Folding used to take years to understand one protein, now you can instantly get a highly accurate guess and some hints for experiments to confirm it.

 

[beenz]

overall market is down over 2000 points in the last week from all time highs tho.

 

[The Pledge]

AI is a bubble?

For context, AI HYPE is a bubble, the tech isn’t.

 

[Treblemaka]

I get the skepticism, but I think the '100k earner buying a 10M home' analogy misses how CapEx works in tech. You’re looking at the 'Infrastructure Phase.' In the late 90s, companies spent billions laying dark fiber that didn't pay off for a decade but eventually, that infrastructure made the modern internet possible.

OpenAI’s spend isn't consumer debt, it’s R&D and compute acquisition.

The revenue is lagging, but that is true of every major platform shift. If we judged Amazon by its profitability in 2002, it would have looked like a disaster. The bet isn't on chatbots, it’s on AI solving the cost of intelligence for coding, drug discovery, and logistics. It's risky, sure, but calling it 'useless' once the chips wear out ignores the IP and models built with that compute.

The $1.5T spend isn't a credit card bill due next month, it's a 10-year capital commitment to build data centers that will likely be the most valuable real estate on earth. If you judge a bridge’s profitability while it's still under construction, it always looks like a money pit.

The concern about the 1.5 Trillion even if its over 10 years thats 150 Billion dollars in spending per year. This is for a company that is barely making $15B in revenues.
Do you believe this company is going increase its revenue 1000% over the next 10 years? Do you think the company to increase its value at least 150B each year for the next 10? Because thats the minimum its going to need to make or increase in value just to cover its spend commitments. To be clear that 1.5 Trillion is literally 3 times the current value of the entire retail mobile phone market globally.

Have you heard Sam Altman say anything to anyone to anyone that give any sort of hit on how hes going to generate that kind of revenue?

Yo it's funny how this is almost word for word the damn near EXACT same set of arguments that all the Wall Street analysts had about Bezos when he started building AWS. It took nearly 7-8 years before AWS started making real money for Amazon. Man back then all the Wall street guru analysts were saying he should stick to his retail website and that they couldn't see any path to how all this hundreds of millions in spending would ever produce a profit.

Also acting like AI will only ever amount to deepfakes and chatbots shows a real lack of imagination. Anybody who's been using Chatgpt since the early days from version one to today sees the huge jump in capability. From shyts and giggles to using it everyday for work. It's really not hard to take that experience and then picture how much better AI will get in a few more years especially given it's just getting started.

So according to your logic, if consumer value has to exist on day one, how do you explain cloud’s early years when it looked exactly the same? There were no consumer apps. No resale value for those servers that become old. No obvious profitability model. Uber and Spotify did not show up at launch. Wasn't until almost ten years later. By your standard Bezos should have shut down AWS entirely in the beginning and we would all be stuck in a world without Netflix streaming or any of the modern services that rely on cloud. All because the profitability model didn't fit into some perfect little 5 year wall street excel model spreadsheet on day one.




You really think Bezos could've predicted Uber or Instagram or DoorDash back in 06? He couldn't. Because the infrastructure did not exist yet. It was being built. Same with AI. You do not get the killer apps until the foundation is stable and cheap. We are in that buildout phase. There are going to be new economic models created from all this shyt that we haven't even imagined yet.

TLDR: Your AWS example is incongruent with Open AI. The success examples you use didnt overcoming to this level and even when they did they had a clear path to profitability. Build it and they will come only works if you know how they will use once they show up.

----------‐------------------------

The AWS comparison is terrible for a couple of reasons:

1. Amazon was already a highly successful market leader company in a complete different market than AWS was targeting.

OpenAI doesnt have a successfully fall back industry to lean on if this isnt profitable. This isnt some corporation experimenting in a different sector this a company still proving itself in an emerging on.

2. AWS was attempting to be a disruptor in a proven market that already had stable publically traded market leaders (Microsoft, Cisco, Oracle, etc.)

Open AI part of a market that has no definitions yet, is very unstable, and has no proven models.


I dont know if AI will amount to deep fakes and chat bots but thats what the market is using it for.
As a comparison when most of America got the internet they started using it to host or participate in multimedia platforms, digital commerce hubs, social gathering sites, video game hubs, etc.

These are things the market started using it for on its own. I use AI at work everyday. Its a slow personal secretary at best that helps me find things faster and generate content more quickly. Thats it. Right now AI is an enhanced version of Clippy for most of its commercial applications. For consumers its a creative tool. Thats it.

What is the model that is going to generate 1.5 Trillion dollars? They are spending money to build it but that dont know what they are going to use it for.

As for the cloud, it was being supported by IT companies that could literally afford to lose the entire business and could still have positive quarterly earnings.
Can Open AI say the same?

I used to work in a Datacenter. Those old servers often took 5 years to be old and 10 to be obsolete. We had Mainframes that were 10 years old with no expectation to replace them anytime soon. Even if you replaced them at 5 years. You could sell them to another company who would resell them to a different buyer in the global market. Can't do that with GPUs. And because of the way they use these GPUs they have a shorter shelf life.

Heres the other problem. If you have a fleet of Trucks and Ford believes they'll be inoperable in 3 years and you think they wont break for 6. Youre going to incur a massive unforseen spend when you have to replace them early, especially if you're red-lining them the entire time.

Last thing. This isnt like Uber or Doordash where youre competing against the other market players. Open AI is competing with the Chinese government. They have infinitely more capital than he can acquire, with a cheaper labor force, more market leverage and better connections.

You guys keep saying the market will find a way to use this but show me the last company that did business that wasn't already successful.

 

[IIVI]

The concern about the 1.5 Trillion even if its over 10 years thats 150 Billion dollars in spending per year. This is for a company that is barely making $15B in revenues.
Do you believe this company is going increase its revenue 1000% over the next 10 years? Do you think the company to increase its value at least 150B each year for the next 10? Because thats the minimum its going to need to make or increase in value just to cover its spend commitments. To be clear that 1.5 Trillion is literally 3 times the current value of the entire retail mobile phone market globally.

Have you heard Sam Altman say anything to anyone to anyone that give any sort of hit on how hes going to generate that kind of revenue?




TLDR: Your AWS example is incongruent with Open AI. The success examples you use didnt overcoming to this level and even when they did they had a clear path to profitability. Build it and they will come only works if you know how they will use once they show up.

----------‐------------------------

The AWS comparison is terrible for a couple of reasons:

1. Amazon was already a highly successful market leader company in a complete different market than AWS was targeting.

OpenAI doesnt have a successfully fall back industry to lean on if this isnt profitable. This isnt some corporation experimenting in a different sector this a company still proving itself in an emerging on.

2. AWS was attempting to be a disruptor in a proven market that already had stable publically traded market leaders (Microsoft, Cisco, Oracle, etc.)

Open AI part of a market that has no definitions yet, is very unstable, and has no proven models.


I dont know if AI will amount to deep fakes and chat bots but thats what the market is using it for.
As a comparison when most of America got the internet they started using it to host or participate in multimedia platforms, digital commerce hubs, social gathering sites, video game hubs, etc.

These are things the market started using it for on its own. I use AI at work everyday. Its a slow personal secretary at best that helps me find things faster and generate content more quickly. Thats it. Right now AI is an enhanced version of Clippy for most of its commercial applications. For consumers its a creative tool. Thats it.

What is the model that is going to generate 1.5 Trillion dollars? They are spending money to build it but that dont know what they are going to use it for.

As for the cloud, it was being supported by IT companies that could literally afford to lose the entire business and could still have positive quarterly earnings.
Can Open AI say the same?

I used to work in a Datacenter. Those old servers often took 5 years to be old and 10 to be obsolete. We had Mainframes that were 10 years old with no expectation to replace them anytime soon. Even if you replaced them at 5 years. You could sell them to another company who would resell them to a different buyer in the global market. Can't do that with GPUs. And because of the way they use these GPUs they have a shorter shelf life.

Heres the other problem. If you have a fleet of Trucks and Ford believes they'll be inoperable in 3 years and you think they wont break for 6. Youre going to incur a massive unforseen spend when you have to replace them early, especially if you're red-lining them the entire time.

Last thing. This isnt like Uber or Doordash where youre competing against the other market players. Open AI is competing with the Chinese government. They have infinitely more capital than he can acquire, with a cheaper labor force, more market leverage and better connections.

You guys keep saying the market will find a way to use this but show me the last company that did business that wasn't already successful.

On the highest levels, a trillion dollars isn’t even that much money anymore breh. In 10 years these’ll be joke numbers.

Apple was worth $500 billion dollars ten years ago as the #1 company. Now companies are worth over $5 Trillion, which is nearly two trillion dollars more than Apple (the largest company) was worth last year.

shyt, we’ll have trillionaires soon. The U.S Debt is over $40 trillion.

You also have to be aware of margins. Exxon Mobil and large companies previously had a lot of CapEx and Operating expenses because they had to manage a whole lot of facilities.

 

[Treblemaka]

On the highest levels, a trillion dollars isn’t even that much money anymore breh. In 10 years these’ll be joke numbers.

Apple was worth $500 billion dollars ten years ago as the #1 company. Now companies are worth over $5 Trillion, which is nearly two trillion dollars more than Apple (the largest company) was worth last year.

shyt, we’ll have trillionaires soon. The U.S Debt is over $40 trillion.

You also have to be aware of margins. Exxon Mobil and large companies previously had a lot of CapEx and Operating expenses because they had to manage a whole lot of facilities.

Youre right a Trillion isnt as much in capital as it used to be but when you have 1500 times your income in financial commitments its practically the sword of Damocles. Especially when you have no strategy to generate revenue.

Again youre comparing OpenAi to companies that spend more on failed projects every year than Open AI makes in revenue.

Apple made over 90M it net income last year. Open AI isnt even profitable yet.

Apple has multiple profitable tangible product streams.
OpenAI has zero

Apple despite having 6x times OpenAi's revenue in profit has never entered a new market thats wasn't already proven without already having been successful in another market.

Name another company that has been successful when it initial product has no clear path to profitability.

Amazon, Microsoft, Nvidia, Amazon, Meta etc. None of them made a bet that was 1500x their total revenue while they were

 

[IIVI]

Youre right a Trillion isnt as much in capital as it used to be but when you have 1500 times your income in financial commitments its practically the sword of Damocles. Especially when you have no strategy to generate revenue.

Again youre comparing OpenAi to companies that spend more on failed projects every year than Open AI makes in revenue.

Apple made over 90M it net income last year. Open AI isnt even profitable yet.

Apple has multiple profitable tangible product streams.
OpenAI has zero

Apple despite having 6x times OpenAi's revenue in profit has never entered a new market thats wasn't already proven without already having been successful in another market.

Name another company that has been successful when it initial product has no clear path to profitability.

Amazon, Microsoft, Nvidia, Amazon, Meta etc. None of them made a bet that was 1500x their total revenue while they were

Remember, software and software services can be installed into many things. They need much much more flexible than hardware.

Think about recommendation systems on Spotify for example. They can connect to an A.I to analyze someone’s preferences so they listen to more music. Think about sensor in an air conditioner system finding the optimal time to turn itself on for someone based on multiple parameters, think about software that can be put in cars, etc.

Think of all the software that accesses the internet for the user. Additionally, think of all the smaller models of A.I that can be installed on things locally.

A.I can basically be installed on everything today and these companies would subscribe to these companies or build their own (which maybe they don’t want to due to high expenses all to underperform a specific A.I another major A.I company has). It’s all connected and A.I is simply software, software is malleable, intangible and highly versatile with the ability to access things seemingly unrelated if the programmer wants it to.

Plus it’s contagious: someone or a company gets ahead of someone else because they used A.I. That’s going to eventually turn into multiple people using A.I.

 

[Gloxina]

@28min

Check NVDIA sales SEC Filings for last 3 months, ending Oct 29.


He somehow timed the top.


And “he” isn’t Thiel.


Although Thiel basically timed the top, too