a16z
Showing 136–150 of 1048 transcripts.
- 51 min
Will LLMs Get Us To AGI?
Martin and Vishal define Artificial General Intelligence as the capacity to generate entirely new scientific paradigms rather than merely interpolating within existing data manifolds, a capability they argue current Large Language Models lack despite their sophisticated Bayesian reasoning. They detail a formal Matrix Abstraction Model explaining how in-context learning functions as evidence-based posterior updates, while simultaneously critiquing the industry's reliance on prompt engineering and empirical scaling as insufficient for achieving recursive self-improvement or true innovation. The discussion concludes that a fundamental architectural leap beyond probability-based transformers is necessary to transition from generating "confident nonsense" to producing outputs that fall completely outside training distributions.
- 49 min
Sam Altman on Sora, Energy, and Building an AI Empire
Sam Altman, Ben Horowitz, Erik Torenberg
OpenAI CEO Sam Altman outlines a strategic shift toward extreme vertical integration, establishing a massive global infrastructure stack to support research, consumer products, and hardware while securing partnerships with Nvidia, AMD, and Oracle. The organization prioritizes scientific advancement and world model development, predicting that within two years AI systems will autonomously conduct complex research in physics and biology, though this rapid scaling necessitates new per-generation monetization models. Concurrently, Altman advocates for a measured regulatory approach focused only on superhuman capabilities to avoid stifling innovation, while emphasizing that the ultimate economic value of this infrastructure lies in the successful and gradual deployment of artificial general intelligence.
- 56 min
Opendoor CEO: Building the Amazon for Homes
Kaz Nejatian, Alex Rampell, Erik Torenberg
CEO Kaz Nejatian is steering Opendoor away from its failed inventory-heavy iBuyer model toward a software marketplace aiming to control 10% of housing supply to disrupt traditional agent monopolies through 1% commissions. Following a strategic pivot triggered by the "triple whammy" of Zillow's competition, rising interest rates, and capital pullback, the company now operates in every U.S. market with features like Dallas' seven-day return window to build high-frequency demand. The ultimate vision is to internalize the entire real estate transaction chain by bundling financing and insurance, thereby reducing friction and delivering net value to consumers while avoiding the pitfalls of operating as a low-frequency asset-holding business.
- 1h 3m
The Lawyerly Society vs. The Engineering State: Who Owns the Future?
Dan Filgate frames the U.S.-China competition not as a race with a definitive winner, but as a complex synthesis where American strengths in wealth creation and intellectual property must balance against China's engineering-driven capacity for infrastructure and manufacturing. The analysis highlights critical frictions between the U.S. "lawyer culture," which often paralyzes industrial progress through legal hurdles, and China's ability to enforce large-scale goals despite systemic human rights concerns and IP vulnerabilities. Ultimately, the dialogue urges the United States to avoid complacency by adopting more functional industrial policies that prioritize physical production and logistical efficiency rather than rigid legalistic processes.
- 27 min
Software is Eating Labor
The speaker outlines a fundamental shift in the global software economy where autonomous AI agents replace human labor to capture the $13 trillion U.S. wage market, effectively transitioning SaaS from digitizing records to executing end-to-end workflows. This transformation necessitates a move away from seat-based pricing toward outcome-based models, exemplified by pilots at companies like Zendesk and startups that deploy AI to perform tasks ranging from freight negotiation to debt collection. Consequently, venture capital efforts are pivoting to identify enterprises that can fully leverage this automation to monetize vast labor markets previously inaccessible to traditional software solutions.
- 43 min
The Person Who Runs HR For 2 Million Federal Workers
Katherine Boyle, Scott Kupor, Greg Barbaccia
The administration is executing a sweeping overhaul of the federal workforce to secure national leadership in the AI race, projecting a reduction of 300,000 civilian employees while dismantling 43 years of hiring barriers to mandate technical skill testing. This strategy shifts performance evaluations toward a forced distribution model and replaces risk-averse culture with outcome-based efficiency, targeting a critical talent gap by recruiting early-career professionals and private sector executives for short-term secondments. Key initiatives include centralizing citizen data through a "One Government" portal, deploying generative AI tools like ChatGPT on government desktops, and restructuring contractor oversight to prevent cost sprawls driven by non-technical management.
- 52 min
Building an AI Physicist: ChatGPT Co-Creator’s Next Venture
Anjney Midha, Liam Fedus, Ekin Dogus Cubuk
Founded by co-creators of ChatGPT and DeepMind physicists, Periodic Labs operates a frontier AI research facility dedicated to advancing physical science by coupling Large Language Models with automated high-throughput experimentation. The organization deploys a unique "mid-training" methodology and physically grounded reward functions to overcome the epistemic limits of current models, specifically targeting the discovery of high-temperature superconductors exceeding 200 Kelvin. By integrating ML scientists, experimentalists, and simulators into a unified workflow, the company aims to replace theoretical predictions with real-world data loops, eventually expanding its "AI physicist" capabilities to serve aerospace, defense, and semiconductor industries.
- 56 min
Anduril CEO: China Has Scale. Can America Catch Up?
Ben, Marc, Erik Torenberg, Brian Schimpf, Chris Power
A strategic assessment reveals a critical gap in U.S. defense industrial capacity, where Russia currently outproduces NATO on 155mm munitions and existing stockpiles would deplete within six to seven days of conflict. This deficit stems from decades of offshoring, an aging skilled workforce, and a reliance on technical superiority rather than mass production, leaving the U.S. unable to sustain high-intensity wars or effectively deter China in scenarios such as an invasion of Taiwan. To address these vulnerabilities, experts recommend a "factory-first" approach featuring government-backed long-term offtake agreements, concentrated capital investment in scaled manufacturing entities, and regulatory reforms to rebuild domestic supply chains and reinvigorate the talent pipeline.
- 59 min
The Common Thread of All Technology: Monitoring the Situation, Ep.1
Erik Torenberg, Katherine Boyle, Eddie Lazzarin
Eric Jackson, Mark Andreessen, and other industry leaders explore how a shared techno-optimist ethos unifies diverse sectors ranging from crypto and defense to consumer hardware, arguing that American innovation culture remains the primary competitive advantage against global rivals. The discussion also addresses critical societal shifts, including the redefinition of medical authority through AI diagnostics, the misaligned incentives driving rising ADHD diagnoses, and the changing role of traditional schooling in an era of hyper-specialized learning. Finally, panelists analyze the fragmentation of internet culture and the evolution of social media platforms like X, emphasizing their function as translation layers that bridge isolated digital subcultures while reshaping public discourse.
- 53 min
From Vibe Coding to Vibe Researching: OpenAI’s Mark Chen and Jakub Pachocki
Mark Chen, Jakub Pachocki, Anjney Midha, Sarah Wang
OpenAI researchers Mark and Jakob Sutskever outline a strategic roadmap centered on GPT-5, which aims to mainstream advanced reasoning and automate scientific discovery by merging the capabilities of instant-response and deep-thought models. This approach shifts evaluation metrics from solving static competition problems to generating economically relevant insights and extending autonomous time horizons to several hours through reinforcement learning. The organization distinguishes itself by balancing protected fundamental research teams with product accountability, prioritizing talent that persists through failure to overcome current limitations in coding autonomy and physical robotics.
- 1h 0m
Software Finally Eats Services - Aaron Levie
Aaron Levie, Erik Torenberg, Steven Sinofsky, Martin Casado
The event analyzes how AI is accelerating a universal adoption curve that empowers young founders and small teams to achieve unprecedented productivity gains by acting as a turbocharger for domain expertise rather than a replacement for it. Participants debate the implications of Reed Hastings' visa salary proposal and discuss how incumbent corporations face disruption from agile startups leveraging non-deterministic workflows to redefine vertical industries. The discussion concludes that while historical leaders will likely expand, the most significant future value creation will emerge from entirely new categories built by the next generation of AI-native entrepreneurs.
- 59 min
Taking Bold Bets: NIH and the Future of Biomedical Science
Dr. Jay Bhattacharya, Erik Torenberg, Vineeta Agarwala, Jorge Conde
Secretary Xavier Becerra announced a $50 million federal investment to launch the Autism Data Science Initiative, selecting 13 teams from 250 applicants to address the rising prevalence of autism and the lack of effective therapies. Concurrently, the NIH is implementing comprehensive reforms to resolve the scientific replication crisis, centralize peer review, and prioritize early-career investigators to foster high-risk innovation. These strategic shifts aim to rebuild public trust through transparent communication, integrate AI for capacity augmentation, and tackle chronic diseases while ensuring rigorous, reproducible medical standards.
- 1h 39m
Dylan Patel on the AI Chip Race - NVIDIA, Intel & the US Government vs. China
Dylan Patel, Erik Torenberg, Sarah Wang, Guido Appenzeller
NVIDIA and Intel have established a transformative $5 billion strategic partnership to jointly develop custom data center and PC products, a move that significantly boosted NVIDIA's stock and acknowledged the industry's shift from CPU to GPU dominance. This alliance contrasts sharply with China's aggressive pursuit of semiconductor self-sufficiency, where Huawei navigates US export bans and high-bandwidth memory bottlenecks through domestic innovation and alleged smuggling channels. Simultaneously, hyperscalers like Oracle and AWS are capitalizing on surging demand with massive infrastructure deals and repurposed capacity, while the market faces technical complexities in deploying next-generation Blackwell architectures and optimizing hardware for specific inference workloads.
- 45 min
The Death of Search: How Shopping Will Work In The Age of AI
The panel analyzes how AI agents are disrupting traditional search and affiliate models by redirecting high-consideration commerce away from SEO-dominated platforms while leaving impulse buys and immediate gratification intact. Key speakers highlight the divergence between walled garden ecosystems and the emerging need for specialized agents that automate price optimization or require new infrastructure for machine-readable product data. The discussion identifies Costco's value-driven model and specialized startups focused on friction reduction as durable counterpoints to Amazon's potential decline, emphasizing that future growth hinges on resolving latency and attribution complexities.
- 56 min
Faster Science, Better Drugs
Erik Torenberg, Patrick Hsu, Jorge Conde
ARC Institute is establishing a physical hub to integrate machine learning with diverse biological disciplines, aiming to solve complex diseases like Alzheimer's by simulating human cells as digital foundations. Patrick O'Brian leads this effort to transition biological discovery from slow physical experiments to high-speed computation, targeting a "virtual cell" capable of predicting cellular perturbations with 90% accuracy. To validate this progress, the organization has launched the $100,000 Virtual Cell Challenge, seeking to replace traditional trial-and-error methods with AI-driven models that can replicate historic scientific breakthroughs.