Radical Talks

Anna Goldie, CEO and Co-Founder of Ricursive Intelligence, joins Radical Partner Rob Toews to unpack what it takes to build a frontier AI lab for chip design. At Google Brain, Anna co-founded the ML for Systems team and, with longtime collaborator Azalia Mirhoseini, created AlphaChip, an AI system Google has used to design multiple generations of its TPUs. She was also an early researcher at Anthropic.

Now, Anna leads Ricursive Intelligence, which raised $335M within four months of launch to build AI that designs chips, and chips that power AI.

Anna and Rob discuss why Anna chose to build independently rather than inside a hyperscaler, why chips are the fuel for AI progress, and why LLMs alone won’t crack chip design. From choosing investors like a long-term partnership, to deliberately tackling the hardest part of the problem first, this episode offers a candid look at building research-led companies at the frontier.

Mantic co-founders Toby Shevlane and Ben Day recently joined Radical partners Aaron Rosenberg and Rich Kotite to discuss what it takes to build an AI system that forecasts world events better than the best human forecasters.

Tantum Collins, Co-Founder & CEO of Inherent, joins Radical Partners Molly Welch and Aaron Rosenberg to dig into recursive self-improvement (RSI), using AI to build better AI, and why Inherent is betting on RSI at the organizational level, not just the model level. Drawing on his time leading DeepMind's meta-research team and shaping AI policy in the Biden White House, Tantum brings a unique lens to the conversation. He discusses Faraday, Inherent's AI agent, the company's "living within the experiment" culture, and why Inherent is structured as a Public Benefit Corporation (PBC). 

Aaron Rosenberg and Rich Kotite, Partners at Radical Ventures, join Molly Welch to discuss "The Rise of NeoLabs," their report mapping the researcher-led AI startups raising massive capital. Drawing on Radical's track record backing companies like World Labs and Inherent, they dig into the research paradigms attracting top talent, how investors evaluate these early-stage teams, and what it takes to build a category-defining lab.

Yvonne Likomanova, Senior Director of Talent at Radical Ventures, digs into what's really driving the AI talent flows. Drawing on a career that spans Stripe, Cruise, Applied Intuition, Yvonne shares where the world's best researchers and engineers are setting their sights. The conversation covers why elite talent is leaving frontier labs for a new class of research-first institutions, what candidates are really optimizing for beyond compensation, and what founders need to get right today to compete for the people who will define their company's trajectory.

Michael Rubenstein, Co-Founder & Co-CEO of Firsthand, joins Radical's Molly Welch to dig into what's broken in digital advertising — a trillion-dollar-plus industry that AI isn't just improving, it's replacing entirely. Drawing on a career that spans DoubleClick, AppNexus, and now Firsthand, Michael makes the case for why the future belongs to brand agents, not ads. The conversation covers how Firsthand's infrastructure meets consumers at the moment of intent, why the real bottleneck is structural rather than technical, and what has to be true for this new model to go mainstream without leaving brands behind.

Varun Sivaram, Founder & CEO of Emerald AI, joins Radical Partners Molly Welch and Rob Toews to explore what's actually holding back AI infrastructure. While the conversation around AI bottlenecks tends to focus on chips and models, Varun explains why the binding constraint is power, from grid capacity limits and interconnection backlogs to the widening gap between compute demand and energy supply. Drawing on his path from senior energy official in the Biden White House to CTO of India's largest solar company to founding Emerald AI, the discussion unpacks why the fix is software, not just more generation, how Emerald's Conductor platform turns data centers into flexible, grid-responsive assets, and what has to be true for AI infrastructure to scale without breaking the grid.

When Ori Eldarov started Offdeal, he wasn't thinking about AI as a productivity hack. He was thinking about what it would take for a small team to compete with firms that had been doing this for decades. Thirteen people. Nearly $10 million in their first full year. This founder-first take was recorded during our October AI Founders Masterclass, before agentic workflows entered the mainstream conversation. The technology has moved fast but the principles remain the same.

Arsham Ghahramani, CEO and Co-Founder of Ribbon, joins Radical Partner Sanjana Basu to unpack fundraising from the founder’s side of the table. With a PhD in AI focused on cancer research and early experience scaling recommendation systems at Amazon, Arsham brings both technical depth and operating perspective to the conversation. Now he leads Ribbon, an AI hiring platform that has conducted over one million interviews and raised $8M+ to date. They break down what actually drives venture conviction, how founders should think about fund theses and stages, and why investor relationships start long before a formal raise. From the reality of two-seconds-per-slide pitch reviews to navigating process opacity with clarity, this episode offers a pragmatic look at how to approach capital with strategy and confidence.

Jim Fan, Director and Distinguished Research Scientist at NVIDIA, joins Radical Partner Molly Welch to explore what's holding back embodied AI and robotics. While large language models have advanced rapidly, Jim explains why progress in the physical world depends less on ever-larger models and more on solving fundamental problems: exploration-driven learning, scalable data pipelines, and synthetic simulation. Drawing on his journey from OpenAI to leading NVIDIA's Project GR00T, they discuss the robotics data bottleneck, why foundation models matter, and what must happen before robotics reaches its "GPT-4 moment."