Latest Interviews
Showing 481–495 of 3,496 transcripts.
Clear all filters- Jane Street48 min
Making GPUs Actually Fast: A Deep Dive into Training Performance
Corwin, Savant Diaz, Sylvain De Wecker
Jane Street engineers optimize deep learning infrastructure by eliminating CPU-GPU synchronization bottlenecks and fusing PyTorch operations via `torch.compile` and Triton to maximize throughput. When automated compilation fails on complex Python logic, they deploy custom CUDA kernels that leverage shared memory and warp-level reductions to achieve nearly 1,000x speedups in specialized tensor operations. This hierarchical approach, ranging from standard PyTorch to hand-optimized C++, ensures efficient utilization of the H100's 132 Streaming Multiprocessors and strict memory bandwidth constraints.
- a16z53 min
Grant Lee: Building Gamma’s AI Presentation Company to 100 Million Users
Founded by Grant in 2020, Gamma has evolved from a rejected startup into a profitable, mass-market platform that generated nearly 100 million users primarily through organic growth and a viral "first 30 seconds" user experience. The company differentiates itself by replacing static slides with malleable, web-like building blocks powered by an "one-button" AI strategy, moving from a prosumer tool to a business application with over $1 million in ARR and enterprise integration capabilities. Despite early investor skepticism, Gamma achieved profitability within three months of launching paid tiers by prioritizing design-centric hiring, maintaining a lean team, and strategically transitioning from founder-led sales to a bottom-up enterprise motion.
Benchmark's GP, Everett Randle on Why Mega Funds Will Not Produce Good Returns
Everett Randle, Ev Randle, Harry Stebbings
Benchmark partner Everett Randall introduces a new framework for evaluating AI companies that prioritizes absolute gross profit and capital velocity over legacy SaaS metrics, while acknowledging the structural limitations of mega-funds in achieving high net multiples. He contrasts this with Benchmark's craft strategy, which focuses on deep founder partnerships and smaller, high-conviction deals to secure superior returns, explicitly rejecting the industry's shift toward massive capital deployment rates. Drawing on his experiences at Founders Fund and recent market corrections, Randall advocates for evolving investment criteria that recognize AI's potential to expand total economic value despite the inevitability of sector-wide consolidation.
- Milken Institute54 min
The Future of Biomedical Research and Innovation | Future of Health Summit 2025
Esther Krofah, Robert Califf, Julie Gerberding, Sharmila Makhija, Elias Zerhouni
A comprehensive review of the U.S. biomedical ecosystem reveals that the nation's seven-pillar research framework is currently threatened by a convergence of reduced federal funding, immigration restrictions, and ideological pressures on universities, all while global competitors like China and Singapore accelerate their own centralized health strategies. Despite possessing a premier research engine, the United States suffers from a critical implementation gap where $5.4 trillion in annual healthcare spending fails to address fundamental issues like low life expectancy, high medical bankruptcy, and a severe erosion of public trust in scientific institutions. Strategic recommendations now urge a shift toward a unified national life sciences strategy that leverages federated data infrastructure, aligns financial incentives with preventive care, and mobilizes political and voter engagement to restore the country's competitive edge in global innovation.
- All-In Podcast1h 27m
Does OpenAI Need a Bailout? Mamdani Wins, Socialism Rising, Filibuster Nuclear Option
Mamdani, Brad Gerstner, Altman, Jensen Huang, chamath, Jason, DavidSacks, friedberg, yung_spielburg, TheZachEffect, sama, JDVance, houmanhemmati
OpenAI announced a revised $20 billion revenue run rate and $1.4 trillion in infrastructure commitments, prompting partner stock declines and clarifications that federal bailouts are not being pursued. While market sentiment remains cautious due to spending uncertainty, panelists predict a shift back to risk-on investment strategies by February alongside a federal regulatory push to preempt state-level restrictions. The discussion further highlighted geopolitical AI competition, a growing youth unemployment debate attributed to structural economic issues rather than AI automation, and the upcoming expansion of Foundry University to new international markets.
- a16z1h 3m
Amjad Masad & Adam D’Angelo: How Far Are We From AGI?
Adam Cheyer and Amjad Masad debate the trajectory of Artificial General Intelligence, with Cheyer predicting near-term "functional AGI" will automate remote work within two years and Masad arguing that current scaling relies on non-scalable human data while true AGI requires a fundamental architectural shift. Their discussion highlights critical economic bottlenecks, including the potential exhaustion of high-quality training data as expert labor is automated, the risk of eliminating entry-level career pathways, and the emergence of solo AI agents that could decentralize economic power. While both acknowledge that true human consciousness remains scientifically unsolved, they anticipate a decade of agents transforming industries from software development to caregiving, urging students to explore philosophy and neuroscience to address the foundational questions current AI engineering bypasses.
Is a $4.5BN Exit Enough in VC? & Harvey Raises $150M & Why Google is a Buy and Amazon is a Sell
Jason Lemkin, Rory O'Driscoll, Harry Stebbings
This episode dissects the shifting venture capital landscape through the lens of Nirvana's IPO, highlighting how aggressive founder fundraising has compressed VC ownership stakes to single digits and challenged the viability of traditional exit multiples. Experts further analyze the divergent fundamentals of the AI sector, contrasting the high-valuation risks facing giants like OpenAI with the operational realities of mature SaaS companies adapting to the "agentic software" revolution. The discussion concludes by questioning the fiduciary responsibility of board members in trillion-dollar capital expenditure scenarios and identifying the critical need for startups to demonstrate genuine labor replacement rather than mere assistance to avoid market obsolescence.
- Sequoia Capital1h 0m
OpenAI Sora 2 Team: How Generative Video Will Unlock Creativity and World Models
Bill Peebles, Thomas Dimson, Rohan Sahai, Konstantine Buhler, Sonya Huang
OpenAI's Sora leverages Diffusion Transformers and space-time tokens to generate coherent videos that increasingly respect physical laws, driving a daily output of seven million clips through an iterative deployment strategy. The platform prioritizes user creation over consumption via algorithmic incentives and the viral "Cameo" feature, while a dedicated API and emerging IP monetization frameworks support diverse enterprise and creator ecosystems. Looking forward, the technology aims to facilitate scientific discovery in physical sciences and evolve into a persistent digital clone platform for social interaction and knowledge work.
- a16z50 min
Seeing The Future from AI Companions to Personal Software
Eugenia Kuyda, Erik, Anish, Justine, Jenna
Wabi founder Eugenia positions the current AI landscape as a primitive "Microsoft DOS" phase, forecasting a transition to a screen-first "Software 3.0" ecosystem where users build and remix highly personalized "mini-apps" through visual, non-technical interfaces. By replacing isolated chatbots with a unified social operating system, the platform enables influencers and creators to distribute interactive functional tools that merge content with utility while allowing communities to remix and collaboratively refine software for niche needs. This approach aims to solve the AI discovery problem by encapsulating complex prompts into shareable applications, effectively shifting the market from passive consumption to mass-market creation and deep personalization.
Cohere's Chief AI Officer, Joelle Pineau: Why Scaling Laws Will Continue & Future of Synthetic Data
Joelle Pineau, Harry Stebbings
A leading AI expert outlines a trajectory where robust scaling laws and algorithmic innovations converge to deliver near-term enterprise productivity gains, while identifying the inefficiency of reinforcement learning for general AGI and the risks of model collapse from synthetic data. The speaker emphasizes that successful adoption requires on-premise deployments to navigate legacy system friction and security vulnerabilities, predicting a market shift toward efficient inference costs and a global ecosystem of sovereign models rather than US-China dominance. Furthermore, the discussion advocates for pragmatic governance over fear-based existential risk narratives, calling for balanced AI teams that prioritize data curation and open innovation to support future interfaces and scientific discovery.
- a16z1h 17m
Sacks, Andreessen & Horowitz: How America Wins the AI Race Against China
Andreessen, David Sacks, Marc, Ben, Erik
The Trump administration is executing a strategic pivot to establish the United States as the global hub for crypto and artificial intelligence by replacing enforcement-heavy regulations with codified legislation and pro-innovation frameworks. This approach includes passing the GENIUS Act to stabilize the stablecoin market, negotiating the Clarity Act for the broader token ecosystem, and dismantling previous AI restrictions to counter Chinese competition while rejecting open-source restrictions. By prioritizing deregulation, energy infrastructure access, and federal preemption over state-level patchworks, the policy aims to secure long-term industry growth and prevent technological stagnation caused by bureaucratic approval delays.
How Ryan Serhant Sold $20B in Real Estate with Media & AI
Ryan Serhant restructured his portfolio into Serhant Technologies, a media and AI-driven holding company that completed over $20 billion in sales and recently secured a $45 million seed investment from Jeff Berman and Harley Miller. The firm distinguishes itself through its production unit, Serhant Studios, which launched the high-stakes Netflix series *Owning Manhattan*, and its proprietary AI platform, Simple, which automates administrative tasks to serve a demographic of younger, digitally native agents. This strategy has enabled the company to pivot from a traditional brokerage model to a scalable technology enterprise that integrates venture capital activities and operates across four primary verticals in 13 states.
- All-In Podcast1h 4m
Pete Buttigieg: The Left's Identity Crisis, Wealth Tax, 2024 Mistakes, Plans for 2028
Pete Buttigieg, Chamath, Jason
Former Transportation Secretary Pete Buttigieg critiques the Silicon Valley shift toward Republican deregulation and warns that extreme wealth concentration and unchecked AI adoption threaten the stability of the republic. He advocates for a progressive wealth tax, contrasts his pragmatic vision for government efficiency with current executive excesses, and argues that the Democratic Party's reliance on identity politics has fractured its coalition. Despite acknowledging bipartisan fiscal irresponsibility, Buttigieg emphasizes the need for a unified economic message to address inequality while supporting strict safety regulations for emerging technologies.
OpenAI Restructuring: Who Wins and Who Loses & Mercor Raises $350M at a $10BN Valuation
Jason Lemkin, Rory O'Driscoll, Harry Stebbings
OpenAI has finalized a legal restructuring with Microsoft and state attorneys general to establish a Public Benefit Corporation topped by a $135 billion charitable foundation, unlocking the path for a potential $2 trillion IPO while granting Microsoft a 27% stake despite CEO Sam Altman retaining zero equity. Concurrently, Andreessen Horowitz deployed a $10 billion "Red Army" fund strategy to dominate the venture landscape through massive check sizes and operational media dominance, forcing late-stage investors to pivot from high-multiple speculation to rigorous selection as market valuations for companies like Merkur and Ramp compress toward public stock growth rates. This shifting landscape redefines venture returns, where mega-funds utilize optioning strategies to secure follow-on rights and founders face a stark choice between early M&A exits or the operational risks of navigating a future public market under intensified antitrust scrutiny.
- Jane Street55 min
Neil Mitchell: Pyrefly: Type Checking 1.8 Million Lines of Python Per Second
Meta engineer Neil Mitchell introduced PyreFly, an open-source Python type checker reimplemented in Rust to address performance and scalability limitations for massive codebases like Instagram. The tool utilizes an aggressive memory eviction strategy and file-level concurrency to deliver rapid IDE feedback while supporting complex type features such as structural subtyping and flow narrowing. Released under the MIT license with over 100 contributors, PyreFly aims to replace legacy systems by prioritizing broad ecosystem adoption and seamless integration with build tools like Buck.