Pinned Tweet
The Future of AI: Rewriting the Rules ⚡️
"The AI race isn't about who has the most power. It's about who makes power irrelevant." @antonpme
Warning, insightful long-read goes below 👇
The rules of a century-old industry are being rewritten right now. But the story of how electric vehicles are displacing cars with internal combustion engines is not about cars. It is a ready-made script for a future market takeover, one that will very likely be repeated in the battle for AI.
Remember the premise: For decades, China was losing the automotive race. They couldn't match the complex mechanical mastery of Western and Japanese giants. The auto industry was a battle of engines and transmissions.
But then the game changed.
It stopped being a battle of mechanics and became a battle of IT specialists. Complexity shifted from the engine to the battery and the software.
This allowed China, the world's electronics hub, to nullify the West's century-long advantage. And it's hard to shake the feeling that this technological shift was not just an organic process, but a brilliantly executed industrial strategy.
Now, this exact script could play out in the most important market of the 21st century: the market for Artificial Intelligence.
Today's Western IT giants & AI labs look unshakable. Their strength lies in the "mechanics" of our era: total dominance in cloud computing and access to advanced chips, which provide computing power. So, how can China challenge them while losing in this specific game?
The answer is not to try to win someone else's game. But to change the game itself by shifting the battlefield.
What if the center of gravity of the AI revolution shifts?
1. From the cloud to the device. The focus is not on the race for data center power, but on hyper-efficient AI running autonomously on billions of devices. Whoever manufactures the smartphones, cars, and robots dictates the standards of the new, decentralized ecosystem.
2. From "erudite" models to Vertical AI specialists. Value shifts from broad, general knowledge to deep, practical expertise. This is the rise of Vertical AI. The advantage is no longer with the 'know-it-all' model, but with highly specialized AIs for industry, medicine, or logistics, trained on exclusive datasets and continuously learning from real-time data streams.
3. From raw power to a paradigm shift. This is the ultimate wild card. A breakthrough here isn't just an improvement; it's a paradigm shift, like inventing the jet engine in an age of propellers. It could make today's power-hungry models and the giant GPU farms they run on instantly obsolete. It’s the move that renders chip sanctions irrelevant and resets the entire game board.
Therefore, the main risk for today's leaders is not direct competition. The risk is that someone will not compete with them, but will simply make their technology irrelevant. They will purposefully create and promote a new paradigm that starts the race from a clean slate.
Exactly as it just happened with cars!
Anton P. 👽 retweeted
Your AI subscription might be the biggest discount in tech right now. 🔥
Someone burned through each plan's limits and priced every call at public API rates:
SuperGrok: $30 → ~$5,700 of compute (190x)
Claude Max: $200 → ~$9,000 (45x)
ChatGPT $200: → ~$2,000 (10x)
AI labs are selling dollars for pennies to win you and your habits. This won't last... use them as much as you can to build something meaningful!
We’ve all been hearing that subscriptions are cheaper than APIs, but I’d never seen the exact number.
So yesterday, I took several popular plans and burned a slice of their weekly allowance on purpose. I logged every model call and priced those calls at the vendor’s public API list price.
While I couldn't test all subscription levels:
According to OpenAI, their monthly allowance is now directly proportional to the subscription price:
- Plus ($20/mo) = the reference OpenAI monthly allowance
- Pro 100 ($100/mo) = 5x monthly allowance - tested
- Pro 200 ($200/mo) = 10x monthly allowance (just downgraded from 20x)
For Claude, their Max 20x is subsidized twice as much:
- Pro ($20/mo) = the reference Claude monthly allowance
- Max 5x ($100/mo) = 5x monthly allowance
- Max 20x ($200/mo) = 20x monthly allowance (not 10x) - tested
According to Meta, for Muse Code:
- Everyday Usage ($5/mo) = the reference Muse monthly allowance
- High Usage ($15/mo) = 5x monthly allowance (not 3x) - tested
- Power Usage ($50/mo) = 20x monthly allowance (not 10x)
What does all this mean? (1/4) 🧵
Your AI subscription might be the biggest discount in tech right now. 🔥
Someone burned through each plan's limits and priced every call at public API rates:
SuperGrok: $30 → ~$5,700 of compute (190x)
Claude Max: $200 → ~$9,000 (45x)
ChatGPT $200: → ~$2,000 (10x)
AI labs are selling dollars for pennies to win you and your habits. This won't last... use them as much as you can to build something meaningful!
We’ve all been hearing that subscriptions are cheaper than APIs, but I’d never seen the exact number.
So yesterday, I took several popular plans and burned a slice of their weekly allowance on purpose. I logged every model call and priced those calls at the vendor’s public API list price.
While I couldn't test all subscription levels:
According to OpenAI, their monthly allowance is now directly proportional to the subscription price:
- Plus ($20/mo) = the reference OpenAI monthly allowance
- Pro 100 ($100/mo) = 5x monthly allowance - tested
- Pro 200 ($200/mo) = 10x monthly allowance (just downgraded from 20x)
For Claude, their Max 20x is subsidized twice as much:
- Pro ($20/mo) = the reference Claude monthly allowance
- Max 5x ($100/mo) = 5x monthly allowance
- Max 20x ($200/mo) = 20x monthly allowance (not 10x) - tested
According to Meta, for Muse Code:
- Everyday Usage ($5/mo) = the reference Muse monthly allowance
- High Usage ($15/mo) = 5x monthly allowance (not 3x) - tested
- Power Usage ($50/mo) = 20x monthly allowance (not 10x)
What does all this mean? (1/4) 🧵
Anton P. 👽 retweeted
Replying to @AnthropicAI
@AnthropicAI, guys, please, stop teasing us and just give us Fable 5.5 already 🙏🫠
Anton P. 👽 retweeted
🤡 Gemini Pro sub is useless now with Gemini 4 Argon.
Just TWO messages/answers and one skill created by Argon, and it's already showing 48% used.
🤡 Gemini Pro sub is useless now with Gemini 4 Argon.
Just TWO messages/answers and one skill created by Argon, and it's already showing 48% used.
Anton P. 👽 retweeted
I agree, and this all looks like the beginning of a paradigm shift and a new (post-labor) economy.
We must all provide for those who will lose their jobs because of AI, so that people can survive this rough transitional period (2026 - 2030/35).
I agree, and this all looks like the beginning of a paradigm shift and a new (post-labor) economy.
We must all provide for those who will lose their jobs because of AI, so that people can survive this rough transitional period (2026 - 2030/35).
Anton P. 👽 retweeted
Fable 5.5 is around the corner, and it's freakishly awesome. I think we'll all be amazed by this model.
Anton P. 👽 retweeted
I couldn’t have imagined that the difference would be so noticeable. Claude Code feels lightning-fast now.
Anton P. 👽 retweeted
🤡Google folks look like clowns right now with their new Gemini 4 Argon model. I don't remember ever hitting rate limits on my Gemini Pro subscription (with any previous Gemini Pro model) after just 5-6 short prompts = answers (writing and editing a short text here).
I couldn’t have imagined that the difference would be so noticeable. Claude Code feels lightning-fast now.
Fable 5.5 is around the corner, and it's freakishly awesome. I think we'll all be amazed by this model.
🤡Google folks look like clowns right now with their new Gemini 4 Argon model. I don't remember ever hitting rate limits on my Gemini Pro subscription (with any previous Gemini Pro model) after just 5-6 short prompts = answers (writing and editing a short text here).













