The people who built the most consequential big tech companies of the past decade have spent the last few years leaving to build again, on their own terms.
Yann LeCun — AMI Labs
Yann LeCun spent 12 years at Meta. He joined in 2013 to build and lead its AI research division, known as FAIR, and served as the company’s chief AI scientist. For much of that time he had unusual freedom: resources to pursue fundamental research without pressure to ship commercial products. That changed in 2025. After the disappointing reception of Meta’s Llama 4 model in April, Mark Zuckerberg restructured the company’s AI operation and installed Alexandr Wang as chief AI officer with a mandate to compete commercially on large language models.
LeCun announced his departure on LinkedIn on November 18, 2025. His argument against the industry’s dominant direction has been consistent for years: large language models predict words but do not understand the world. His new company, Advanced Machine Intelligence Labs, known as AMI, is building what he calls world models, AI systems that observe and interact with physical environments, maintain persistent memory, and reason and plan, rather than simply predicting text.
LeCun serves as AMI’s executive chairman. The company is led day-to-day by CEO Alexandre LeBrun, a French entrepreneur who previously co-founded and ran Nabla, a medical AI startup. AMI is headquartered in Paris, with offices planned in New York, Montreal, and Singapore. On March 10, 2026, the company announced $1.03 billion in seed funding at a $3.5 billion pre-money valuation, the largest seed round ever raised by a European startup. LeCun had initially sought around €500 million; demand exceeded that significantly. Investors include Cathay Innovation, Greycroft, Hiro Capital, HV Capital, Bezos Expeditions, Nvidia, Toyota, Samsung, and Temasek.
David Silver — Ineffable Intelligence
David Silver spent more than a decade at Google DeepMind, where he led the reinforcement learning team and built some of the most celebrated AI systems of the past decade. AlphaGo, released in 2016, was the first AI system to defeat a world champion Go player without handicap. Its famous Move 37 in Game 2 against Lee Sedol, a move no human had conceived, became the symbolic moment that catalysed the subsequent wave of deep learning investment. AlphaZero followed, mastering chess, Go, and Shogi from scratch through pure self-play with no human game data at all. Silver is also a professor at University College London.
Silver left DeepMind in late 2025. Ineffable Intelligence was incorporated in November 2025 and emerged from stealth on April 27, 2026. The company is London-based and has no product, no revenue, and no public roadmap. What it has is a thesis: that reinforcement learning, not large language models trained on human-generated data, is the path to superintelligence. The company describes its goal as building a superlearner capable of discovering knowledge and skills entirely from its own experience. Silver has pledged 100 percent of his personal equity gains to high-impact charities through Founders Pledge, described as the largest such commitment in the organisation’s history.
On April 27, 2026, Ineffable announced a $1.1 billion seed round co-led by Sequoia Capital and Lightspeed Venture Partners at a $5.1 billion valuation. Investors include Nvidia, Google, Index Ventures, DST Global, and the UK government’s Sovereign AI Fund.
Ilya Sutskever — Safe Superintelligence
Ilya Sutskever co-founded OpenAI and served as its chief scientist, overseeing the research breakthroughs that led to GPT models, ChatGPT, and the reasoning model o1. He left in May 2024 following a board dispute in which he had voted to remove CEO Sam Altman, a decision he publicly regretted within days.
On June 19, 2024, Sutskever co-founded Safe Superintelligence with Daniel Gross, formerly Apple’s head of AI, and Daniel Levy, an AI researcher and former OpenAI employee. The company’s stated mission has a single focus: one goal and one product, a safe superintelligence. It has no commercial products, no public roadmap, and roughly 20 employees. It operates from offices in Palo Alto and Tel Aviv.
In September 2024, SSI raised $1 billion at a $5 billion valuation from Andreessen Horowitz, Sequoia Capital, DST Global, and SV Angel. In March 2025 it raised a further $2 billion led by Greenoaks Capital, with Alphabet and Nvidia also participating, reaching a $32 billion valuation, a sixfold increase in less than a year. Total funding stands at $3 billion. In June 2025, Meta attempted to acquire the company but was rebuffed by Sutskever. Daniel Gross subsequently left for Meta; Sutskever became CEO.
/dev/agents: The Android team builds again
David Singleton spent years as VP of Engineering at Google overseeing Android, then became CTO of Stripe. Hugo Barra was VP of Product Management for Android at Google, ran Xiaomi’s international operations, and then led Oculus as VP at Meta. Ficus Kirkpatrick was an early Android engineer and later VP of Augmented and Virtual Reality at Meta. Nicholas Jitkoff was a principal designer on Google Chrome OS and held senior design roles at Dropbox and Figma. In 2022, all four left their respective roles to co-found /dev/agents.
The founding logic was an analogy to Android. Before Android, developers built for dozens of incompatible mobile environments. Android created a unified platform that unlocked a generation of applications. The /dev/agents team believes the AI agent ecosystem is at the same pre-standardisation point, and is building a cloud-based operating system for AI agents that can run across devices and give developers a unified platform to build on. The company has since rebranded to Dreamer.
In November 2024, the company announced a $56 million seed round co-led by Index Ventures and CapitalG at a $500 million valuation. Notable angel investors include Andrej Karpathy, Alexandr Wang, Android co-founder Andy Rubin, and Palo Alto Networks CEO Nikesh Arora.
Ashish Vaswani — Essential AI
In 2017, Ashish Vaswani was lead author of “Attention Is All You Need,” the Google Brain paper that introduced the transformer architecture. The transformer now underlies virtually every significant AI model in production, including GPT-4, Claude, Gemini, and DALL-E. Vaswani spent over five years at Google Brain in total. Niki Parmar was his close collaborator there and a co-author of the same paper.
Both left Google in 2021 and briefly co-founded Adept, an AI agents startup, before departing in November 2022 to start again. They founded Essential AI in San Francisco in 2023. Vaswani is CEO. The company’s mission is to build full-stack AI products for enterprise workflows, deepening what it describes as the partnership between humans and computers.
In December 2023, Essential AI raised a $56.5 million Series A led by March Capital, with participation from Google, Nvidia, AMD, Thrive Capital, Franklin Venture Partners, and KB Investment. Combined with an earlier $8.3 million seed led by Thrive Capital, total funding stands at approximately $65 million. Vaswani remains CEO.
Discover more founders in various industries:
AI Founders Who Left College to Build Something Bigger
San Francisco Founders Under 30 Who Are Already Reshaping AI
Restaurant Tech Founders Driving Innovation In San Francisco’s Dining Economy
AgTech in the San Francisco Bay Area: The Founders Rewiring Food, Farming, and Biological Production
