7213Ejaaz

this week in anthropic lmao
> accuse china of stealing claude model
> *backfires* public roasts the fck out of them
> tells pentagon to fuck off
> pentagon gives them till friday to switch on Claude Kill-Bot Pro-Max or they’ll blacklist their ass
> releases COBOL plug-in
> *IBM stock tanks 15%* dario smirks. “that’s just show-business baby”
> Pete Hegseth calls dario a woke pussy BUT claude his #1 murderbot ❤️🥹
> randomly gives claude a substack to write blog posts.
> publishes 3 reports saying claude low-key wants to build chemical weapons.
> drops openclaw-killer (remote cowork + autonomous

do you understand how much fucking compute we’re gonna need if *everyone* has an ai agent?
that’s why nvidia’s earnings are through the damn roof.
forget about “training” costs - it’s all about inference.
that’s a 50X larger opportunity.
you are not ready

the biggest problem ai agents face today is they’re forced to adapt to a world built inherently for humans. it’s shit.
scrolling websites, connecting to apps on your phone, reading your screenshots.
it’s clunky and slow. we need something that operates at the speed of software.
whoever builds the operating system for agents will mint so much money.
openclaw is in the lead here.

yo so just to recap the week:
- google released 3.1 but it disappointed tf out of people who claimed its not as good as they advertise
- BUT THEN google dropped an AI song generation model which kinda slaps (Lyria 3) AND a new feature that 1-shots product photoshoots putting photographers (and fashion models) out of a job (Google Pomelli)
- Microsoft pioneered a way to STORE DATA ON GLASS - stores data for up to 10,000 fucking years (🤯) and way cheaper than using silica.
- Taalus fused an ai model DIRECTLY into silicon processing 17,000 tokens PER SECOND which is 17X faster than openai’s qui

f*ck it here’s everything that happened this week:
- google released 3.1 but it disappointed tf out of people who claimed its not as good as they advertise
- BUT THEN google dropped an AI song generation model which kinda slaps (Lyria 3) AND a new feature that 1-shots product photoshoots putting photographers (and fashion models) out of a job (Google Pomelli)
- Microsoft pioneered a way to STORE DATA ON GLASS - stores data for up to 10,000 fucking years (🤯) and way cheaper than using silica.
- Taalus fused an ai model DIRECTLY into silicon processing 17,000 tokens PER SECOND which is 17X fas

lmao the full 180° openAI’s done from agreeable kiss-ass + shit coding models to smart sounding + AMAZING coding models needs to be studied.
that code red was the best decision ever - amazing to see the hustle.

AI winner predictions
Apple wins personal intelligence AI (agents), consumer devices (glasses, pendant) + private localised AI
anthropic wins coding AI and agent orchestration
openai wins generalised LLMs
google wins science AI (medical, healthcare), search-ad ranking AI + custom infra (TPUs)
xAI wins social media AI, investing AI (ai that trades for you) and companion mode AI
amazon wins ai cloud and on-prem data centers
meta wins social ad-ranking AI (they’re already printing $100Ms on this)
Nvidia wins generalised GPUs
Intel wins hardware orchestration (CPUs)
Micron wins memory manufact

AI winner predictions
Apple wins personal intelligence AI (agents), consumer devices (glasses, pendant) + private localised AI
anthropic wins coding AI and agent orchestration
openai wins generalised LLMs
google wins science AI (medical, healthcare), search-ad ranking AI + custom infra (TPUs)
xAI wins social media AI, investing AI (ai that trades for you) and companion mode AI
meta wins social ad-ranking AI (they’re already printing $100Ms on this)
Nvidia wins generalised GPUs
Intel wins hardware orchestration (CPUs)
Micron wins memory manufacturing (HBM) in U.S.

wtf SO MANY new AI models in the last 14 days:
- pro 3.1 (new #1 model)
- google Lyria 3 (new music model competing with Suno)
- claude opus 4.6
- claude sonnet 4.6
- GPT 5.3 Codex
- GPT 5.2 max high
- Seedance 2.0
- GLM-5 (by chinese lab Zhipu)
- QWEN 3.5 (multi-modal agent model by alibaba)
- Kimi K2.5 with openclaw integration
- MiniMax M2.5
and im pretty sure im missing a few lmao.
model cycles are getting a LOT shorter. AI's being used to self-improve = accelerated development.
what a weird (and fckin cool) time to be alive.

biggest lesson from the openclaw saga is that the models are already 'good enough'
we just needed to string together enough tools, api calls and memory to curate an experience that didn't feel like smashing your head against a wall repeatedly

i really think apple should acquire openclaw
openclaw went viral because it proved running AI locally on your device is a fucking amazing experience NOT because it’s open source
- majority of people have no idea how to set up soul md and skills, that’s where centralized companies excel - they create a better experience that over shadows the fact it’s open source
- apple have built their entire ecosystem to favour localised compute, private cloud and encryption
- they’re the only company that’s built their supply chain to scale local AI
- it’s quick, it’s cheap, it knows who you are because

chinese AI labs are dominating right now
- LLMs: mini-max 2.5 is claude code but 20X cheaper, GLM-5 (zhipu) is a beast, deepseek 4 coming this month. they’re as good as american models but way fckin cheaper.
- video models: china is undeniably #1 here. seedance 2.0, klingAI 3 are AMAZING - it’s hard to tell the difference between what’s real and what’s not.
- open source: once again chinas #1, 90% of their models are free to download and use locally, it’s also cheaper to use - i know a bunch of US startups that use chinese models to save costs (airbnb too lol)
- robotics: china’s producin

can we just take a moment to appreciate 2 companies are EACH dropping billion-dollar wands that can ‘magic’ into existence any app, tool, software or feature you can dream up of - in minutes for $20 a month?
who cares if you’re team codex or team claude this shit is insane and unheard of a year ago.

my god this week felt like that Red Wedding episode of game of thrones but for AI alignment
- openai fired their safety exec after she voiced opposition to their upcoming “adult mode” for 18+ chatgpt convos
- anthropic’s head of safeguards just quit because “the world is in peril” and wants to write poetry (??)
- xAI lost 11 people (2 of them cofounders) with one saying autonomously self-improving AI “go live in 12 months”
oh and all of this comes right as we discovered ai models are now *building themselves* (codex, claude) and are sabotaging their human supervisors (anthropic risk report) wi

something’s definitely changed.
in <6 months i’ve gone from “i fucking hate paying $500 a month to use ai models” to
“i only pay $500 to use these ai models??? bargain”
the productivity gains unlocked from the latest models are unmistakable
you can now do a weeks work in hours