Welcome, AI Innovators!
While AI labs chase scale, governments are laying down steel and concrete. Mira Murati now leads a $12B lab without a product. And Trump’s $100B AI-industrial push brings Google and Westinghouse into the trenches.
📌 In today’s Generative AI Newsletter:
Murati’s AI lab raises $2B before launch, aims to rival OpenAI
Trump backs $100B AI-energy blitz to outpace China
Runway’s Act-Two enters Hollywood production pipelines
Former OpenAI engineer exposes internal chaos and code sprawl
Special highlight from our network
AI isn’t as independent as you think. IBM explains why that matters.
A lot of people think AI can just learn and make decisions on its own.
That’s not how it works.
In this interview from Gitex Europe in Berlin, we sit down with Christian Metz, AI Engineer at IBM, to clear up the confusion.
He breaks down what’s really going on behind the scenes of enterprise AI:
✔️ Why AI still needs human direction
✔️ How IBM builds trust and control into every system
✔️ Why company data is the new competitive edge
✔️ What “Agentic AI” is and why it’s coming fast
✔️ How IBM’s hybrid, open approach sets it apart from startups
Christian also shares what IBM’s focused on now, what excites him about GenAI, and where he sees AI going in the next 24 months.
🎰 Mira Murati Raises $2B Seed Round, Valuing Her AI Lab at $12B
Image Credit: : Newscom | JOHN ANGELILLO
Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, just closed a $2B seed round led by a16z, valuing the secretive AI company at $12B. Despite being under a year old with no product yet launched, the deal reflects staggering investor confidence in Murati’s next act.
Here’s what we know so far:
Investor roster: Andreessen Horowitz, Nvidia, Accel, Cisco, AMD, ServiceNow, and Jane Street joined the round.
Talent magnet: The team includes former OpenAI researchers like John Schulman, Luke Metz, and Barrett Zoph.
Product hints: Murati teased a launch in “the next couple months” with a “significant open source offering.”
Infra deal: Thinking Machines Lab will train its models on Google Cloud, fueling speculation of frontier-scale ambitions.
The funding size puts Thinking Machines in rare air, rivaling Anthropic’s early raises and outpacing most startups before shipping a product. With major ex-OpenAI talent, billions in backing, and growing whispers of Meta’s failed acquisition bid, the lab is shaping up to be the boldest new contender in the AI arms race.
💲Trump Unveils $100B Plan to Cement US Tech Supremacy
Image Source: AP Photo/Andrew Harnik
In a fiery speech at Carnegie Mellon, Donald Trump announced over $100B in AI and energy investments aimed at making the US the global tech leader. The initiative includes new AI data centers, power infrastructure, and regulatory fast-tracking to supercharge development on home soil.
Here are some examples:
$25B from Google: Building large-scale AI data centers in Pennsylvania and beyond
$15B in grid upgrades: Federal funding to modernize and expand the national power grid
$6B in nuclear energy: A Westinghouse-led push to supply stable, AI-grade electricity
New executive orders: Fast-tracked permitting and deregulation for AI infrastructure projects
In a post-event briefing, Trump added, “Whether it’s robots or AI, we need somebody to take care of it. Behind the bravado is a deeper signal: America’s bet on AI won’t be built in the cloud alone. It’s being hardwired into the land.
🎬 Runway’s Act-Two Levels Up AI Character Animation
Image Source: Runway
Runway has launched Act-Two, a major upgrade to its motion capture tech that brings static characters to life using just a single video performance and reference image. It captures everything from hand gestures to facial nuance and it’s already being folded into real Hollywood pipelines.
What’s now in motion:
Single input, full output: One driving performance video and one image are all it needs to animate full-body motion, speech, and expressions
Style-preserving generation: Maintains original art style, background, and aesthetic of the reference image
Major fidelity upgrade: Outperforms Act-One in movement consistency and realism, according to Runway
Hollywood buy-in: Lionsgate and AMC Networks are already onboard, exploring the model for future production workflows
AI may be taboo in public writers’ rooms, but behind closed doors, tools like Act-Two are being tested, tweaked, and rapidly integrated. As production costs climb and timelines shrink, the creative pipeline is shifting and Runway is betting it will soon run straight through AI.
🗣️ Inside OpenAI: An Ex-Engineer Spills the Beans on Chaos and Code
Image Source: The New York Times
Former OpenAI engineer Calvin French-Owen just published a rare inside look at what it’s actually like to build at the center of the AI storm. After helping ship the Codex coding agent in seven sleepless weeks, French-Owen walked away not because of scandal, but to start building again.
What it’s like to build inside OpenAI:
Explosive scale: Headcount jumped from 1,000 to 3,000 in a year, breaking hiring, reporting, and shipping processes
Move-fast chaos: Despite its size, OpenAI still runs entirely on Slack and feels like early Facebook, with no real guardrails
Codebase mess: Duplicate libraries, inconsistent quality, and a sprawling Python monolith create daily friction for engineers
Launch velocity: Codex shipped with a 15-person team in 7 weeks and exploded in usage just by showing up in the ChatGPT sidebar
Concrete safety focus: Internally, safety efforts zero in on prompt abuse, bio-weapon generation, and political manipulation and not abstract existential threats
For all its global stature, OpenAI still acts like a startup wired on adrenaline and Twitter feedback. It may be in a fishbowl, but the build-fast energy is real and French-Owen’s account shows just how fast the gears are turning behind the scenes.
🚀 Boost your business with us. Advertise where 12M+ AI leaders engage
🌟 Sign up for the first AI Hub in the world.
