AI @ extreme

Extreme to the core for AI.

[AI @ extreme @sedxtreme]

I think most people don't understand how early we are. Now, think of what technologies will be there after a year, you can not even predict next month as this space is moving too fast.

[i²cjak @i2cjak]

hahahahahaha these retards have no idea. We’re so early man. We’re so early

[AI @ extreme @sedxtreme]

@paulg Agreed. When your competitor has all the resources to go head-on with you, adding constraints on yourself or the people working in the industry will have huge negative implications.

[AI @ extreme @sedxtreme]

Great words never spoken.

[aether @bluewdrift]

Embarrassment is the cost of entry.
If you aren't willing to look like a
foolish beginner, you'll never become a graceful master.

[AI @ extreme @sedxtreme]

Ok, this is one the greatest argument I have ever read about Jacking off. Makes sense to me.

[AI @ extreme @sedxtreme]

I think the idea of open source is what should people fear about. A distributed ai is very dangerous. Yes, it is very helpful and cheap but powerful and distributed which means anyone, in any part of the world, can have access to ai, run it locally without fearing anyone.

[Jeffrey Emanuel @doodlestein]

It's pretty remarkable just how much of the sum total of human knowledge and expertise can be condensed into something that fits into a $90 hard drive.

If you lost everything else, this would get you pretty far in terms of covering the basics (and not-so-basics) of most fields.

[AI @ extreme @sedxtreme]

I can do this in Pakistan and nobody will give af. I think it is a checkmate for the Chinese distilling American frontier models.

[the tiny corp @__tinygrad__]

This is a great sustainable business model for open weights. It's free if you run it yourself, but if you are running a cloud providing it to others for money, you should have to share profits. (from Kimi K3 License)

[AI @ extreme @sedxtreme]

They thought they will build a powerful tech and keep it to themselves, hire a switch off guy to prevent humanity from harm but it was the Chinese all along.

Sad with looking at envy to the models they built. The open source is a distributed intelligence.

[AI @ extreme @sedxtreme]

guilty.

[Prompter @PromptLLM]

never isolate yourself. isolation is a disease and advocating for it is the biggest bs that gets spewed on this app.

there is a reason prisoners are put into solitary confinement. isolation makes you crazy

[AI @ extreme @sedxtreme]

China can not have this problem. You can steal millions of books and you are good to go. 1.5 billion for nothing.

[Reuters @Reuters]

Anthropic reached a $1.5 billion settlement in a copyright case brought by a group of authors who accused it of misusing their books to train its Claude AI model, the largest known settlement of its kind https://t.co/LgIIIb8lSI

[Pedro Domingos @pmddomingos]

Open-weight models are inherently accelerationist, because they enable far more progress than closed-weight ones.

[AI @ extreme @sedxtreme]

I testament to this.

[𝕾𝖎𝖗 𝕮𝖍𝖗𝖎𝖘 @eagleseyeinc]

make money early cos your entire nervous system relaxes when your finances are in order. life doesn’t feel so heavy anymore. you sleep better, you think more clearly, you’re less stressed and you make better decisions. financial peace is a different kind of peace. there’s nothing soothing like going to bed and waking up without any string of financial stress on your mind. make money. above all, love god.

[AI @ extreme @sedxtreme]

Makes sense.

[David Senra @FoundersPodcast]

Peter Thiel:

[AI @ extreme @sedxtreme]

>you don't need friends to do this.

[AI @ extreme @sedxtreme]

This whole idea of unproductivity is being utterly rejected by the companies that provide code as a service aka vibe coding. Why would companies with hundreds of billions of valuation will ever pay if there is no productivity?

[dog @a_musingcat]

it’s crazy how much code people are writing with AI and how much it literally hasn’t changed anything in life at all for anyone. but it is a lot of code

[AI @ extreme @sedxtreme]

Why Chinese are only distilling the AI models? Why not the Europeans, why not the indians? Why not the Arabs? Why not the Australians? Why not someone or an ai lab in England or any other country on face of the earth? Why Chinese?

[aether @bluewdrift]

This is free advice from an expensive psychologist.
If you're an anxious person, do everything for fun. Go to a job interview for fun. Submit documents for fun. Start a blog for fun. Anxiety feeds on importance. Don't make everything a matter of life and death.

[AI @ extreme @sedxtreme]

I mean, you can't rewrite your DNA. You can train yourself to do things but changing DNA? that is a bit too far of yapping.

[gabrielle @legitimatetiger]

human beings are born with this same ability btw. your thoughts are so powerful, they can not only reshape your entire reality but also send information to reprogram your cells, rewrite your DNA, control your immune responses, and reorganize your whole body on a cellular level

[AI @ extreme @sedxtreme]

damn.

[aether @bluewdrift]

Confidence doesn't come from believing in yourself. It comes from having done the uncomfortable thing enough times that your nervous system stops flagging it as an emergency. Because nothing is actually hard, it's just unfamiliar.

[AI @ extreme @sedxtreme]

You know what they say: using AI, you can make a very strong bullish argument even for an engineering company that they show in a wolf of the wall street.

[James Altucher @jaltucher]

MARKETS ARE HAVING A 'KIMI MOMENT'

The markets are having a "Kimi moment", similar to when DeepSeek was released in early 2025. Markets are selling off thinking that big AI companies won't need to build out as much infrastructure . But this isn't true.

Kimi K3 is the latest open source AI model out of China (interestingly, it first came out the box saying, "Hi, I'm Claude", suggesting, as many people think, that it trains on Claude data it steals). Kimi is good. Very good.

And so people are worried: will there be the same demand for data center buildouts if open source models are just grabbing the trophies? They had the same worry in January 2025 when DeepSeek came out.

The answer is: demand will be greater than ever. Any selloff due to fears of no buildout will be reversed.

A) JEVON's PARADOX. When a technological commodity becomes more efficient, demand for that commodity actually goes up.

An example is DeepSeek, the first open source AI model out of China. The same selloff happened.

Guess what? Capex went up every quarter since then.

When a resource becomes cheaper, people think of more applications for it, and demand continues to rise.

B) THE MODEL IS DIFFERENT FROM THE INFERENCE - just because someone trained a model more efficiently, doesn't mean demand for inference goes down. If demand goes up (as suggested in "A"), then the need for GPUs, CPUs, interconnects, etc go up.

C) THE SELLERS OF COMPUTE DON'T CARE. Oracle, Microsoft, Coreweave, Facebook, etc - anyone selling compute and inference, they don't care what models they are hosting - they just want to povide inference for the people. If anything, their margins will go up if they use cheaper models.

D) AGENTS - Last week was the first week that human traffic on the internet was exceeded by agent traffic (traffic being done by AI-created bots). Agents require 50x the amount of "tokens" (the units that AI is prices in) than humans do. This does not change with new models being released. If anything, the better the model, the more agents will be made with it.

E) FRONTIER MODELS TAKE ON THE COMPETITION

Anthropic proved last February that many of the Chinese open source models are stealing Claude's inference to train their own models.

Does this mean Anthropic and OpenAI give up? Of course not. They will just make better models. Even since February, Anthropic has released Fable 5 and OpenAI has released ChatGPT 5.6, which are still the best two models. These trillion dollar companies won't rollover. They will build faster than ever.

F) ITS ALREADY DONE

Land purchases, power purchase agreements, GPU sales, Memory sales, Interconnect sales, etc etc are already contracted out through 2029. Let's talk then but chances are the rise of new models will continue to create new demand for inference and the buildouts will continue. But, for sure, the next three years trillions of dollars will be spent.

G) THE LINUX EXAMPLE. Another example of Jevon's Paradox. Linus was an open source free operating system in the 90s. A cheaper operating system allowed MORE software, data centers, operating systems, etc to be created, resulting in a huge data center buildout in the 2000s.

This Kimi Moment will mirror the DeepSeek moment of early 2025. Everyone will reiterate capex. Money will be exchanged. Chips will be bought. Stocks will flourish.

[AI @ extreme @sedxtreme]

truth. accelerate ai.

[Beff (e/acc) @beffjezos]

America needs to wake up.

We either accelerate or die.

We can't cede the future to China, and anyone trying to convince you otherwise is serving their agenda.