Sequoia Capital
Showing 1–15 of 190 transcripts.
- 1h 1m
Crusoe CEO Chase Lochmiller: Turning Stranded Energy Into Intelligence
Crusoe, a $31 billion vertically integrated AI infrastructure platform founded by CEO Chase Lockmiller, recently divested its Bitcoin mining operations to NYDIG to pivot exclusively toward energy-powered data centers and large-scale "AI factories." Leveraging a diverse portfolio of power sources ranging from stranded natural gas to small modular reactors in locations like Iceland and Norway, the company aims to serve as the "supermajor" utility provider for the global intelligence economy. To manage this exponential growth, Lockmiller emphasizes a culture of intellectual flexibility and continuous improvement, utilizing deep delegation of domain experts to navigate the complexities of scaling operations and maintaining organizational resilience.
- 44 min
European Defense Sovereignty Starts in the Factory: Stark’s Uwe Horstmann and Johannes Schaback
Uwe Horstmann, Johannes Schaback, Pat Grady, Luciana Lixandru
Founded by Uwe and Johannes Stark, the European defense startup Stark secured billions in contracts within 18 months by applying a venture-first model to develop scalable, AI-enabled weapon systems for Ukraine and NATO forces. The company differentiates itself through a "make or break" engineering culture that integrates rapid soldier feedback loops with mass-production readiness, treating autonomous programs as internal startups rather than traditional bureaucratic projects. With plans to invest €600 million over the next three years, Stark aims to fundamentally restructure European procurement toward agile innovation while adhering to responsible AI principles to strengthen transatlantic security architecture.
- 45 min
Inside the Robot Data Factory: Mecka's Josh Gao on Sensorizing Humans to Train Robots
Mecca AI addresses the critical data constraints in robotics by deploying proprietary sensorized wearables to capture high-fidelity egocentric video and tactile force data for training physically online foundation models. Having already amassed millions of hours of curated human demonstrations through rigorous quality gates, the company is transitioning from a pure data provider to a platform enabling commercial deployment in controlled industrial environments like hotels and data centers. By converging multimodal inputs including simulation-based self-play and force feedback, Mecca aims to bridge the vision gap between human and robot interaction, ultimately scaling toward a future where robots outnumber humans in highly specialized, task-specific applications.
- 1h 8m
David Silver and Lasse Espeholt of Ineffable Intelligence: A Superlearner for the Age of Experience
David Silver, Lasse Espeholt, Sonya Huang, Rich Sutton, Vera Rubin
Co-founders Dave Silver and László Espohold established Ineffable Intelligence to develop autonomous superlearners that discover new scientific and mathematical knowledge from scratch using complex simulated environments called "crucibles." Leveraging a dedicated Rubin GPU cluster on Google Cloud and a team of elite DeepMind alumni, the company aims to bypass human data pre-training to build systems capable of organic emergence. The organization prioritizes solving alignment in simulation before physical deployment, forecasting that the first evidence of this approach will emerge within a year and true superintelligence contact within three years.
- 1h 6m
Luca Ferrari: The Operating System Behind Bending Spoons
Bending Spoons operates as a publicly traded tech platform with an $18 billion valuation, executing a "buy-to-hold" strategy that integrates acquired digital businesses into a shared operating system to drive permanent efficiency. The Milan-based firm differentiates itself through a data-driven recruitment model that processes 800,000 annual applications using quantitative metrics rather than interviews, alongside a culture of radical simplification and extreme ownership led by CEO Luca Ferrari. By prioritizing raw talent and fluid organizational structures over industry experience, the company plans to expand its international footprint with new US offices while continuing to compound growth through disciplined, technology-enabled acquisitions.
- 1h 4m
Google's AI Infrastructure Chief, Amin Vahdat, on the Physics & Economics of Frontier AI
Google is executing the largest capital expenditure build-out in human history, deploying over $200 billion to construct specialized AI data centers that co-design hardware and software to maximize "Good Put" performance metrics. This strategy shifts from general-purpose infrastructure to purpose-built clusters powered by deep collaboration between DeepMind and infrastructure teams, enabling near real-time failure recovery and a six-month doubling of effective serving capacity. Facing power as the primary constraint, Google is pioneering advanced optical networking and considering orbital data centers to sustain long-horizon agent workloads while adhering to open standards and a multi-vendor ecosystem.
- 1h 15m
Databricks’ Ali Ghodsi Never Wanted to Be CEO. Now He’s Among the Best
Ali Ghodsi, Ben Horowitz, Brian Armstrong
Since 2015, CEO Ali Ghodsi has guided Databricks from a $1.5 million revenue startup to a market leader by pivoting from product-led growth to an aggressive enterprise sales motion under CRO Ron Gabsco. This strategic overhaul involved replacing technical hires with emotionally intelligent sales leaders and relentlessly championing the controversial "Lakehouse" architecture to undermine competitor Snowflake on cost and AI integration. Ghodsi now applies similar principles of multi-year focus and internal AI integration to manage organizational scaling, prioritizing long-term market transformation over short-term public market pressures.
- 1h 5m
Box's Aaron Levie: On Reinventing Yourself in the AI Age and Enterprise Diffusion
Strategic analysis of the current AI landscape highlights a market pivot from raw model development to application-layer "Neo Labs" that bridge legacy systems with enterprise workflows, driven by the recognition that value will accrue across the entire stack rather than solely at the infrastructure level. Box exemplifies this shift by transforming into an agentic harness that deploys long-running, asynchronous agents to extract structured data and automate complex workflows, utilizing a model-agnostic garden to balance cost and accuracy while adhering to strict domain-specific evaluation protocols. Ultimately, successful market penetration depends on overcoming adoption barriers through robust data hygiene and systems of record, as execution capabilities and cultural integration will determine which organizations capture the trillion-dollar opportunity in applied AI.
- 52 min
Making Cities Awesome: Peregrine’s Nick Noone & Ben Rudolph
Nick Noone, Ben Rudolph, Sonya Huang
Founded by ex-Palantir and UNHCR veterans Nick Benes and Ben Hallowell, Peregrine deploys a "forward-deployed" engineering model to help municipalities build privacy-preserving data infrastructure without creating a surveillance state. The company leverages agentic AI to automate 95% of complex data integration and execute rapid analytics, such as identifying hidden crime patterns in Florida or simulating hurricane impacts for local leaders. By maintaining strict local data sovereignty and reducing delivery costs below one million dollars annually, Peregrine enables thousands of unique cities to preserve institutional memory and solve specific community safety challenges.
- 55 min
Parallel’s Parag Agrawal: Building a New Web for AI Agents
Parag Agrawal, Sonya Huang, Andrew Reed
Parallel, founded by ex-Twitter CEO Parag Agrawal, is building a "web systems" infrastructure that replaces human click data with direct agent feedback to optimize search indexing and ranking for software agents. The company has launched a specialized search agent product and secured a strategic partnership with Google Cloud to serve as the primary grounding provider for enterprise AI, achieving retrieval speeds of 200 milliseconds for top-relevant tokens from a trillion-page web. By shifting from traditional advertising economics to a "Shapley value" model for content attribution, Parallel aims to monetize high-quality data extraction for workflows ranging from financial modeling to autonomous agent triggering.
- 54 min
Rich Sutton and Khurram Javed: Why AI Models Stop Learning, and How to Start It Again
Rich Sutton, Khurram Javed, Sonya Huang, Alfred Lin
Rich Sutton and Oak Lab advocate for "continual learning" as the essential default for true intelligence, critiquing current Large Language Models for freezing weights and relying on finite human-curated data rather than adapting through real-world experience. To overcome the barrier of catastrophic forgetting, the lab proposes a 12-step research agenda utilizing individual step-size optimization and meta-learning to enable neural networks to continuously update their internal world models. This approach aims to create energy-efficient, self-maintaining agents capable of forming general abstractions across diverse domains, moving the field beyond static training paradigms toward systems that evolve alongside their environments.
- 22 min
Continual Learning: How AI Agents Get Better With Every Use | Arjun Karanam, Trajectory
Arjun Karanam, Ronak, Gabe, Harrison, Nico, Harvey
Trajectory, co-founded by Arjun and Ronak, addresses the lack of accumulated experience in AI by building a platform that enables models to continuously learn from the 100 trillion daily tokens generated by real-world agent interactions. The company utilizes a dual-learning architecture combining differential privacy with reinforcement learning on user-corrected traces, allowing organizations to transition from static models to systems that compound capability through automated post-training and flexible harness optimization. By abstracting complex training parameters into a 15-minute workflow, Trajectory empowers enterprises to retain ownership of their specialized models while refining agent performance directly against production traffic.
- 24 min
When to Build Your Own Agent Harness | Harrison Chase, LangChain
The framework defines autonomous agents as systems built from three owned components: the model, context, and a prioritized harness that orchestrates data flow through an iterative LLM loop. Organizations can customize this harness via middleware for domain-specific optimizations or maintain off-the-shelf versions for in-distribution tasks, ensuring compatibility through dynamic model profiles. Continuous improvement is driven by a flywheel where trace data from evaluations using the Harbor benchmark feeds into an automated engine that identifies failures and suggests prompt, code, or context fixes.
- 26 min
RL Environments Explained: How AI Agents Learn Real-World Work | Brendan Foody, Mercor
Mercore has expanded its revenue run rate to $2 billion by transitioning the AI data market from basic crowdsourcing to high-skilled "agentic data" services that enable frontier labs to build complex reinforcement learning environments. The company leverages expert networks of lawyers, engineers, and doctors to create realistic simulated worlds with precise human-verified rubrics, demonstrating a fivefold increase in model performance on specific legal tasks during recent training. As the primary data vendor for major application layer companies, Mercore addresses the industry's need for ultra-long horizon tasks and social dynamics evaluations that synthetic models cannot yet self-generate.
- 28 min
Post-Training Is How You Keep Your Taste | Fireworks CEO Lin Qiao
Fireworks CEO Linh Nguyen advocates for a strategic industry shift from relying on rented APIs to owning intelligence through deep model customization, enabling companies to preserve unique business judgment while reducing inference costs by five to ten times. This approach utilizes a structured lifecycle of data curation, fine-tuning, and serving loops to transition from generic prompting to specialized models, as demonstrated by success stories like Cursor and niche vertical leaders in healthcare and security. Ultimately, post-training is positioned as the critical mechanism for startups to scale after product-market fit by converting proprietary user data into unclonable domain expertise before high API expenses threaten unit economics.