TL:DR – RedHat for AI
In an AI market dominated by heavyweight models and billion-dollar APIs, a new contender is quietly redefining the playing field. Thinking Machines Lab (TML)—the brainchild of ex-OpenAI CTO Mira Murati and a formidable roster of co-founders—has burst onto the scene with the largest seed round in Silicon Valley history: $2 billion at a $12 billion valuation. Backed by a blue-chip syndicate including Andreessen Horowitz, Nvidia, Accel, AMD, Cisco, and Jane Street, this public benefit corporation isn’t building another ChatGPT. It’s building the tools and infrastructure for the next generation of AI-native products.
In an AI market dominated by heavyweight models and billion-dollar APIs, a new contender is quietly redefining the playing field. Thinking Machines Lab (TML)—the brainchild of ex-OpenAI CTO Mira Murati and a formidable roster of co-founders—has burst onto the scene with the largest seed round in Silicon Valley history: $2 billion at a $12 billion valuation. Backed by a blue-chip syndicate including Andreessen Horowitz, Nvidia, Accel, AMD, Cisco, and Jane Street, this public benefit corporation isn’t building another ChatGPT. It’s building the tools and infrastructure for the next generation of AI-native products.
How is it different
You could think of it as a “formalized, productized version of open-source AI tooling”—with commercial-grade packaging, high interoperability, and deeper integration support than raw open-source models or repos typically provide.
While rivals like OpenAI, Anthropic, and xAI are racing to scale ever-larger foundational models, TML is carving out a distinct niche. Their goal isn’t to be another black-box API. Instead, they’re designing a modular, developer-friendly ecosystem—where product teams, builders, and researchers can compose, customize, and deploy multimodal AI into their workflows and apps.
This shift from monolithic models to application-first architecture could mirror the evolution from mainframes to cloud platforms—positioning TML not as an AI model provider, but as a foundational layer for AI-native software design.
Premier League Founding Team
1. Elite Founding Team
TML is led by some of the most respected technologists in AI:
- Mira Murati, former CTO of OpenAI
- John Schulman, co-creator of PPO and founding OpenAI scientist
- Barrett Zoph, former Google Brain architect (AutoML)
- Backed by a growing bench of ex-OpenAI, Meta, and Mistral engineers
2. Multimodal from Day One
Unlike incumbents retrofitting vision and speech into text-heavy architectures, TML is building natively multimodal systems—designed to interpret and interact with text, images, interfaces, and state simultaneously.
3. Modularity, Not Monoliths
TML promises tools that allow developers to plug, play, and customize AI components, rather than rely on generalized endpoints like GPT-4o or Claude. Think of it as the difference between renting a tool and owning the factory.
4. Open, But Strategic
While not fully open-source, TML’s commitment to releasing open research components signals an intent to cultivate a transparent, composable ecosystem, similar in spirit to Hugging Face or Databricks—without being fully decentralized.
5. Massive Strategic Support
In addition to its Sand Hill VC base, TML has drawn interest from hardware and enterprise giants:
- Nvidia and AMD (AI compute)
- Cisco and ServiceNow (enterprise integration)
- Jane Street (quantitative edge)
- Even the Albanian government has invested, citing sovereignty and AI participation
How It Compares
TML doesn’t compete head-to-head with OpenAI or Anthropic—not yet. Those platforms are mature, well-integrated, and growing fast (Anthropic in particular has an impressive enterprise roster). But where those firms optimize for scale, control, and safety, TML is targeting customizability, composability, and developer velocity.
If OpenAI is the Apple of LLMs—polished, vertically integrated—TML wants to be Linux meets Unity for AI.
Who Will Choose TML over GPT/Claude
TML will likely appeal to:
- Product teams at AI-native startups (e.g., Figma, Notion, Replit)
- Infra players who want deep AI control (e.g., Databricks, Retool)
- Enterprises building AI into edge applications (e.g., robotics, autonomous vehicles)
- Builders frustrated by closed, one-size-fits-all models
It may not win with the CIO of a Fortune 500 right away—but for the next breakout AI product, TML could be where the magic happens.
Final Word
TML is early, unproven, and undeniably ambitious. It has no product, no revenue, and no customers—yet. But with its technical DNA, strategic capital, and radical product vision, it’s the kind of company investors, developers, and AI strategists should be tracking closely.
In a world of copycats and scale wars, Thinking Machines Lab is one to watch—not because it’s chasing the incumbents, but because it’s trying to build something new.
Discover more from Priory House
Subscribe to get the latest posts sent to your email.