• The Deep View
  • Posts
  • 23-year-old fired from OpenAI now manages $1.5 billion

23-year-old fired from OpenAI now manages $1.5 billion

Welcome back.

IN TODAY’S NEWSLETTER

1. 23-year-old fired from OpenAI now manages $1.5 billion

2. xAI makes Grok 4 free globally days after GPT-5 launch

3. New model helps robots predict falling boxes and crosswalk dangers

FUNDING

23-year-old fired from OpenAI now manages $1.5 billion

Leopold Aschenbrenner got fired from OpenAI for raising security concerns, had zero professional investing experience, and somehow convinced people to hand him $1.5 billion anyway.

The 23-year-old former researcher's long/short hedge fund, “Situational Awareness”, (yes, he named it after his own manifesto) reportedly gained 47% in the first half of 2025 while the S&P 500 managed just 5.5%. Not bad for a kid who got pushed out for raising security concerns.

Aschenbrenner is the latest example of what we might call the "AI Mafia" — the growing network of former lab employees who've parlayed their insider knowledge into new ventures, investment funds and startup empires. Think PayPal Mafia, but with neural networks and a lot more existential dread.

The pattern is everywhere once you start looking:

  • Ilya Sutskever left OpenAI and launched Safe Superintelligence, which hit a $32 billion valuation before it even had a product

  • Daniel Gross and Nat Friedman — fresh from their Meta's superintelligence recruitment — are backing Aschenbrenner's fund

  • Anthropic itself was founded by former OpenAI executives who saw the writing on the wall

These people have seen the technology's development from the inside. Aschenbrenner's strategy exemplifies this insider advantage. He's betting long on semiconductor companies, infrastructure plays and power providers while shorting industries that could get left behind. His backers include the Collison brothers from Stripe, Graham Duncan who organizes the Sohn Investment Conference and Carl Shulman, another AI intellectual who used to work at Peter Thiel's macro hedge fund.

Apollo's chief economist recently warned that AI mania is "worse than 1999's tech bubble," pointing to price-to-earnings ratios that have eclipsed dot-com peak levels. In the late 1990s, venture capitalists admitted they didn't care if companies were profitable because they just needed successful IPOs to cash out.

The AI Mafia is making money off the disruption they helped create. Aschenbrenner literally named his fund after his own manifesto about AI superintelligence. Sam Altman pushes generative AI with OpenAI while simultaneously founding WorldCoin to solve the problems that same AI creates.

When the people building the future leave to get rich off predicting it, their forecasts become suspiciously self-serving. AI might indeed reshape everything, but that doesn't mean giving a 23-year-old $1.5 billion based on six months at OpenAI is smart money.

TOGETHER WITH MODE MOBILE

Apple’s Starlink Update Sparks Huge Earning Opportunity

Apple just secretly added Starlink satellite support to iPhones through iOS 18.3.

One of the biggest potential winners? Mode Mobile.

Mode’s EarnPhone already reaches 50M+ users that have earned over $325M, and that’s before global satellite coverage. With SpaceX eliminating "dead zones" worldwide, Mode's earning technology can now reach billions more in unbanked and rural populations worldwide.

Their global expansion is perfectly timed, and you still have a chance to invest in their pre-IPO offering at just $0.30/share. Final allocations are disappearing quickly.

Mode’s recent 32,481% revenue growth and their newly reserved Nasdaq ticker $MODE puts them one step closer to a potential IPO.

MODEL WARS

xAI makes Grok 4 free globally days after GPT-5 launch

Elon Musk's xAI announced Sunday that its flagship AI model Grok 4 is now available to all users worldwide for free, marking a major shift from the paid-only access since its July launch. The move comes just days after OpenAI released GPT-5 to all registered users.

Free users can access Grok 4 through two options:

  • Auto mode, which automatically routes complex queries to the advanced model

  • Expert mode, which gives direct access to Grok 4's full capabilities for every query

The most powerful version, Grok 4 Heavy, remains exclusive to SuperGrok Heavy subscribers at $300 per month.

xAI is offering "generous usage limits" for a limited time, though exact quotas remain unclear. Some reports suggest limits around five queries per 12 hours, while others indicate more generous temporary allowances. Users must sign in to access Grok 4 as staying logged out restricts access to the older, faster Grok 3.

The expansion also includes free access to Grok Imagine, xAI's image-to-video generation tool, though only for US users initially.

Musk previously indicated plans to integrate advertisements into Grok to help cover the high operational costs of running advanced AI models. The company says the free access will help expand its user base and gather data for future improvements.

TOGETHER WITH MOMENTUM

Introducing SmartClips: TikTok+Opus Clip Meets Revenue teams

Stop spending hours manually reviewing calls for best practices. SmartClips revolutionizes enablement with AI-powered video intelligence that automatically transforms your customer conversations into shareable mini-clips.

Our breakthrough technology uses Signal AI Agents to instantly identify and extract key moments from objection handling, winning talk tracks, product feedback all without human intervention. Clips distribute automatically across Slack, Teams, and the Momentum platform within minutes of call completion.

Finally, enablement teams focus on strategy, not data detective work.

SmartClips leads our biggest product launch ever, featuring 5 game-changing innovations: Account Briefs, Coaching AI Agent Dashboard Upgrade, Microsoft Integration, and Signals to Email Expansion.

Ready to eliminate manual work and unleash your team's potential?

ROBOTICS

New model helps robots predict falling boxes and crosswalk dangers

Nvidia announced Monday at SIGGRAPH a fresh batch of AI models for its Cosmos platform, headlined by Cosmos Reason, a 7-billion-parameter "reasoning" vision language model designed for physical AI applications and robotics.

The announcement builds on Nvidia's world foundation model ecosystem that was first launched at CES in January. While the original Cosmos models focused on generating synthetic video data, the new Cosmos Reason takes a different approach — it's designed to actually understand what's happening in physical spaces and plan accordingly.

The latest releases include Cosmos Transfer-2 for faster synthetic data generation and a distilled version optimized for speed. But Cosmos Reason is the standout, promising to help robots and AI agents think through spatial problems like predicting when "a person stepping into a crosswalk or a box falling from a shelf" might happen.

This represents Nvidia's continued push into what it calls "physical AI" where they are trying to bridge the gap between AI that works well with text and images, and AI that can actually navigate and manipulate the real world. Robotics companies have been struggling with the expensive process of collecting enough real-world training data to make their systems reliable.

Companies like 1X, Skild AI, and others are already testing Cosmos models, suggesting there's real demand for tools that can generate physics-aware synthetic data rather than forcing developers to film thousands of hours of robot footage.

The models are available through Nvidia's API catalog and can be downloaded from Hugging Face, continuing the company's strategy of making advanced AI infrastructure accessible while positioning itself as the essential platform for the next wave of robotics development.

LINKS

  • Speak: A new way to learn languages, powered by AI

  • Anyscale: Platform to run and scale all your ML and AI workloads

  • Jasper: AI that helps marketing teams perform better

  • Zero Email: AI-native email client that manages your inbox, so you don't have to

GAMES

Which image is real?

Login or Subscribe to participate in polls.

POLL RESULTS

Do you think entry-level jobs will disappear?

  • No they’re crucial for culture (15%)

  • Yes, companies are greedy (52%)

  • Need more info / I’m writing in (6%)

  • I hope not… (27%)

“The sky color in [the other image] looks way off.”

“The building seemed too unusual to be an AI response to a basic prompt. Also the shadows in the other pic seemed off. ”

“I was convinced by the obviously repainted patch across the bottom front of the picture.”

“[The other option] did not look real. The concrete of the building was too perfect and only the baseboard near the chair looked real.”

“I knew it was trick and should have chosen the other image, but I just couldn’t bring myself to do it. That one appears so flat, the water lifeless, and the distance between the staircases doesn’t make sense. ”

The Deep View is written by Faris Kojok and The Deep View crew. Please reply with any feedback.

Thanks for reading today’s edition of The Deep View! We’ll see you in the next one.

Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning.

If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

*Indicates sponsored content

*Mode Mobile Disclaimer: Mode Mobile recently received their ticker reservation with Nasdaq ($MODE), indicating an intent to IPO in the next 24 months. An intent to IPO is no guarantee that an actual IPO will occur. The Deloitte rankings are based on submitted applications and public company database research, with winners selected based on their fiscal-year revenue growth percentage over a three-year period. Please read the offering circular and related risks at invest.modemobile.com.