AI Generated by Fortune India
What happens when four of Google’s biggest AI minds leave to build a startup? Discovery Loop is the answerAugust 6, 2026, 14:07 IST
Loading AI Hub...
Disclaimer : Certain content on this page, including summaries, timelines, FAQs, glossaries, highlights, insights, and other supplementary informational features, maybe generated or assisted by artificial intelligence tools. While reasonable efforts are made to review and verify such content, AI generated output may occasionally contain errors, omissions or inconsistencies. Readers are advised to independently verify any information before relying upon them for professional, legal, financial, medical or other decisions. The publisher along with its affiliates and contributors do not warrant accuracy of AI-generated content and disclaim any liability, loss or damage arising from its use.

What happens when four of Google’s biggest AI minds leave to build a startup? Discovery Loop is the answer

/3 min read

ADVERTISEMENT

Four of Google’s top AI architects reunite to launch Discovery Loop, a public benefit startup that uses artificial intelligence to automate the scientific method and accelerate breakthroughs in machine learning, science and engineering.
What happens when four of Goog
(From left) Oriol Vinyals, Sanjay Ghemawat, Jeff Dean, and Quoc Le Credits: Discovery Loop

Google has lost four of the engineers who helped shape its artificial intelligence ambitions over the past two decades.

Jeff Dean, Google’s longtime chief scientist, has teamed up with veteran engineers Sanjay Ghemawat, Oriol Vinyals and Quoc Le to launch Discovery Loop, a public benefit corporation that aims to use artificial intelligence to automate scientific and engineering discoveries. The launch comes as Alphabet undertakes a broader leadership reshuffle within its AI division, with Demis Hassabis becoming Alphabet’s chief scientist and Koray Kavukcuoglu assuming day-to-day leadership of Google DeepMind.

Sign up for Fortune India's ad-free experience
Enjoy uninterrupted access to premium content and insights.

According to the company, Discovery Loop’s mission is “to automate machine learning, science, and engineering to accelerate discoveries and progress.” It says it is building AI systems capable of automating the iterative process of scientific discovery—from generating hypotheses and designing experiments to analysing results and improving subsequent iterations.

The brains behind Google’s AI

The founders are among the most influential engineers to have emerged from Google’s AI organisation.

Jeff Dean joined Google in 1999 and went on to become one of the company’s most respected engineers, helping build its search infrastructure, co-founding Google Brain and later serving as Google’s chief scientist. He announced the venture on X, writing that the four founders had worked together for “between 14 and 30 years” building Google’s AI models, products and computing infrastructure.

Sanjay Ghemawat, another early Google engineer, co-created foundational technologies such as MapReduce and Bigtable, systems that underpin modern distributed computing. Oriol Vinyals most recently served as Vice President of Research at Google DeepMind and was a technical lead for the Gemini models, while Quoc Le co-founded Google Brain and led pioneering work on AutoML, Google’s effort to automate machine learning.

Collectively, the quartet represents decades of Google’s AI and infrastructure expertise, making Discovery Loop one of the highest-profile AI startups to emerge from the company in recent years.

In the recently made website for the company, it stated how the founders’ experience have pioneered massive scale computing and led the creation of critical infrastructure, products, and foundational AI advances, including multiple generations of Google Search, Google Ads, Google News, Google Translate, Google File System, MapReduce, BigTable, Spanner, TensorFlow, Pathways, TPUs, AlphaChip, AlphaStar, AlphaCode, AlphaFold, Gemini, model distillation, mixture-of-experts model architectures, word2vec, sequence-to-sequence models, chain of thought reasoning, neural architecture search, and multiple generations of Large Language Models (LLMs) among others.

“Our relative advantage isn't just our technical ability; it is the unprecedented scale of the systems we have previously built. We possess true full-stack depth that spans chips, hardware infrastructure, software infrastructure, ML models, and products,” the website said.

Automating the scientific method

Unlike most AI startups building chatbots or coding assistants, Discovery Loop is targeting scientific research itself. The company says it wants to build AI systems that can support researchers throughout the entire discovery process. Its initial focus will be on machine learning research before expanding into broader scientific and engineering disciplines.

In a post announcing the company, Dean described Discovery Loop as “a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress.”

The company adds that by automating complete experimental loops at massive computational scale, it hopes to “dramatically accelerate technological and scientific progress.”

Alphabet remains a partner

Despite the high-profile departures, Alphabet is not severing ties with the founders. Discovery Loop’s initial funding round is co-led by Radical Ventures and Khosla Ventures, with participation from Alphabet alongside Lightspeed, Kleiner Perkins and Doerr Capital. The startup has also entered into a long-term partnership with Alphabet for cloud and compute resources.