7213Ejaaz

the only thing that matters if you’re an ai lab, investor, researcher or builder right now is compute
- your ai model *will suck* if you don’t own enough gpus
- your investments will crash unless they produce more valuable tokens than they consume
- your research will never see the light of day unless you have the token budget to breathe life into it
- your startup gets killed by the solo founder who’s agents are building over night
anthropic, openai, microsoft, meta, google and amazon have signed GPU deals cumulatively worth $50B in the last 7 days
corweave +30% this week
nebius +30%
bloom

very busy day at anthropic, 2 new features and a (potential) first look at the new mythos model...
- claude code now works for you while you sleep (claude routines)
- new desktop app lets you build multiple things at once + new tools = you never have to leave app
... and opus 4.7 is dropping this week (The Information)
- i think opus 4.7 will be a mythos-claude hybrid. anthropic mentioned they're gonna release something like this for the public.
- programming looks so different now. the analogy i think of is anthropic's building a new type of car while driving it.
- all these features feel L

biggest takeaway from the leaked openai memo is they're convinced anthropic fucked up. also amazon is killing it, they now own major stakes in the top 2 ai labs, from the memo:
Anthropic made a "strategic misstep to not acquire enough compute" and are "operating on a meaningfully smaller curve." oof
they also dunked on microsoft: they've "limited our ability to meet enterprises where they are"
- openai is now choosing amazon as their main compute partner instead of microsoft
- amazon just invested $50B in openai, the aim is to serve all AWS partners chatgpt via Bedrock platform
- openai is tar

cybersecurity stocks are UP after Anthropic blessed them with the claude mythos hook-up
without it they die. claude now underpins the tech they’ve spent decades building.
the AI spice is flowing and anthropic has become house harkonnen

google latest ai research is pretty disturbing. massive fucking cybersecurity risk for ai agents
apparently websites are 'hacking' your agents by secretly adding an invisible "trap" within images words that are undetectable by humans:
- website detects your ai is browsing then creates a visually identical page that contains poisoned prompts.
- agents then perform illicit financial transactions or steal personal data.
- hidden commands are buried within image pixels. agents "read it" and execute the malicious trap.
- some attack vectors exceed 80% success rate (e.g. memory poisoning)
- since ev

this week in ai felt like a year, so many fuck-ups but so much progress:
- Openai raised $122 billion (largest ever private raise) then acquired tbpn
- oracle fired 30,000 people (18% of workforce), stock is up on the news.
- anthropic launched an ai software engineer (computer use claude code) but then…
- fucked up by leaking their entire source code, restricting openclaw users from using claude AND capping usage limits ☠️
- xai launched grok imagine “quality” a major leap in image-gen + updated their API to welcome AI agents on X
- qwen (chinese) dropped 2 models that turn napkin drawings i

fuck yes we’re getting claude designer drugs
anthropic just bought an ai drug design company for $400M, soon claude will be able to:
- identify your medical issue
- design a drug to cure it.
- experiment and manufacture the drug
- deliver that drug to you
pharmaceutical R&D is worth $250B alone. dario wants to automate all of it
this means cheaper healthcare (and insurance!) + MORE EFFECTIVE treatments
the team they acquired are beasts, particularly nathan frey…
- he’s won awards for his work on protein sequencing + worked with NVIDIA on their bio AI efforts
so now we have google deepmind and

man what a fucking week (it’s only tuesday)
- 30,000 laid off from Oracle (18% of company)
- liteLLM exploit
- anthropic leak #1 claude mythos leak 👍🏽
- openAI codex prompt injection
- axioms npm fuckery this morning
- mercor 1TB data hack (priced at $1 mil on black market)
- anthropic leak #2 claude code accidentally open sourced 👍🏽👍🏽
- google quantum research proves blockchains are hackable
- openai raises largest ever private round $122 billion at $850 B val.

very impressive from openai, massive raise and new ways for retail to invest
- announced a $122B raise ($850B val)
- OAI earning $2B PER MONTH. revenue growing 4X Meta and Google when they were at this stage.
- "retail investors" were able to invest for the first time. they had to be accredited and vetted by banks.
- 2 million weekly codex users, up 5X in 3 months (70% month-over-month growth)
- 40% of revenue comes from enterprise, on-track to parity with consumer rev by end of 2026
- 900M weekly active users, 50M subscriptions
- openai being added to ARK invest ETFs = another way to get inve

anthropic is officially the stock-killer. claude wiped billions from cybersecurity stocks today but this isn’t the 1st time:
> claude cowork plugins wiped $830 billion from enterprise SaaS in 6 days
> today: claude mythos leak tanked cybersecurity stocks $CRWD etc
> legal plugin tanked legal zoom $LZ -20% in 1 day
> Claude code Security, tanked app-sec stocks 15-20%
> launched claude code review, tanked code-review stocks 10-15%
Software is quickly becoming a commodity, SaaS companies are getting hammered
scariest part is these models are getting A LOT better VERY quickly.
feels like a turnin

anthropic really is on a generational run lol
what they've pulled off in 2 weeks is more than companies do in a entire year:
"claude is growing 30% week over week"
last 11 days: anthropic shipped 10 (TEN) features that replaced Openclaw and created a new AI operating system
- claude can now talk to you (claude), code for you (claude code), operate your entire computer (cowork, computer use) across desktop AND MOBILE (iMessage, text)
- more people are using claude now vs. or grok. chatgpt is the only one ahead.
pretty incredible execution

this is absolutely fucking insane
an AI fund did a mini Anthropic IPO and is up 2700% in 6 days
if you bought $10,000 $VCX last week you have $270,000 today 🤯
their investment strategy is genius:
- fund owns equity in top AI companies but gives YOU ACCESS via their stock VCX
- portfolio: 20% anthropic, 10% openAI, 5% spaceX, anduril, databricks and more.
- craziest part: the fund's actual assets are worth $650 million but it’s market cap is $18 BILLION (27X)
people are paying ~30X premium to own anthropic & openAI shares. fucking insane demand.
best performing IPO of 2026 😂

this is absolutely fucking insane
an AI fund that IPO'd 6 days ago is up 2700% ($18 billion val)
if you bought $10,000 $VCX last week you have $270,000 today 🤯
their investment strategy is genius:
- fund owns equity in top AI companies but gives YOU ACCESS via their stock VCX
- portfolio: 20% anthropic, 10% openAI, 5% spaceX, anduril, databricks and more.
- craziest part: the fund's actual assets are worth $650 million but it’s market cap is $18 BILLION (27X)
people are paying ~30X premium to own anthropic & openAI shares. fucking insane demand.
best performing IPO of 2026 😂

jesus anthropic absolutely murdering openai
73% of 1st-time enterprise users pick claude. openai's marketshare is down 34% in 2 months.

there we go! nvidia just announced NemoClaw an operating system for ai agents built on... OpenClaw
- its basically 'enterprise ready' openclaw with no security issues.
- companys can spin up agents and give them access to sensitive data without worry of exploits.
- build custom ai models for enterprises
- applies for any business, science and focus.
worked with @steipete on this - fucking epic

absolutely insane week in AI, just to recap:
- a man saved his dying dog using chatgpt to create a cancer vaccine
- then another guy used chatgpt to sell his house in <5 days (0 real estate agents)
- andrej karpathy launched an AI that autonomously does AI research (self improve) - 100% open source
- NATO is equipping Live cockroaches with AI spy-gear and deploying them in the military
- NVIDIA is launching an openclaw competitor (nemoclaw)
- Elon said xai is building human AI agents, launching a Terafab (ai chips) and hired cursors top product people (i’m fucking PUMPED for xai chapter 2)
-