Poddavsnitt

AI researcher Nathan Lambert discusses the competitive dynamics between closed frontier labs like OpenAI and Anthropic and the open-weight model ecosystem. He analyzes the roles of compute, data, and talent capital, the strategic approach of Chinese AI labs, and the launch of his new nonprofit research lab, Trillium, focused on documenting recursive self-improvement.

A discussion featuring Stanford Professor Dan Boneh and a16z partners regarding the potential for AI to break modern cryptographic systems. The conversation addresses concerns sparked by OpenAI's mathematical breakthroughs and distinguishes AI-driven threats from quantum computing. The participants conclude that while AI introduces a new "player" in attacking complexity assumptions, there is currently no evidence of a cryptographic breach.

Co-founders of MakerMaker.ai, Daval Adjeda and Owen He, discuss the technical definition of Recursive Self-Improvement (RSI). They distinguish true RSI—where the AI system acts as both the optimizer and the optimized—from simple autocatalysis, and explore the empirical challenges of measuring genuine capability gains versus noise and reward hacking.

Nick Bostrom discusses the evolution of AI capabilities, the risks of existential risk and misalignment, and the tension between pacing frontier development and the potential costs of delaying breakthroughs in medicine and longevity. He analyzes the impact of reinforcement learning on goal-seeking behavior and the implications of the Hugging Face agent swarm incident.

Ben Horowitz discusses his journey from engineer to CEO, the founding principles of a16z, and his philosophy on evaluating entrepreneurs based on world-class strengths. The conversation also explores the impact of AI on software development and the personal influence of hip-hop culture on his life and business perspective.

Pat Gelsinger, General Partner at Playground Global and former CEO of Intel and VMware, discusses the evolution of chip design and the current bottlenecks in AI infrastructure. He reflects on his early career at Intel, the birth of the EDA industry, and how AI is now transforming the chip design process. The conversation also touches on energy constraints, memory innovation, and the future of virtualization for AI agents.

AWS CEO Matt Garman discusses the massive infrastructure expansion driven by AI, including a $220 billion capital expenditure plan and the acquisition of 2 million NVIDIA GPUs. He explores the evolving relationship between AWS and startups, the shift toward agentic workflows, and the strategic importance of custom silicon and cloud architecture for AI agents.

A discussion with Valen founders Linda Du and Andrew Wang regarding the modernization of the mortgage servicing industry. They explore the transition from legacy infrastructure to AI-driven systems, the challenges of operating in a heavily regulated environment, and the importance of a robust system of record for improving consumer experiences.

Kevin Mandia, founder of Mandiant and now CEO of Armaden, discusses the paradigm shift in cybersecurity caused by AI. He argues that AI-driven attacks operate at machine speed and scale, rendering traditional human-led defense obsolete and necessitating autonomous defensive systems.

A discussion on the seventh edition of a16z's Top 100 Consumer AI Apps report. The conversation focuses on the shift from traffic-based metrics to consumer spending data, the emergence of personal agents, and the "power law" of AI adoption where a small percentage of power users drive significant revenue.

Jack Altman and A16Z general partner David George discuss the "and" thesis of AI investment, arguing that frontier models, open source, and applications will all succeed simultaneously. They analyze the current state of AI diffusion, noting that while revenue is currently concentrated among coders, the potential for the broader knowledge workforce is massive. The conversation also covers supply chain constraints in data center capacity and the relationship between compute power (gigawatts) and token consumption.

Alex Atala discusses the acquisition of OpenRouter by Stripe and the strategic vision for a neutral, multi-model AI marketplace. The conversation explores the concept of 'neurodiversity' in AI—combining multiple models to avoid vendor lock-in—and the importance of cost efficiency in enabling new startups to emerge.

A discussion on the competitive advantage of AI-native startups over established software incumbents. The conversation focuses on the concept of 'owning the end-to-end arc' of a business process, using enterprise procurement as a primary case study to illustrate how work often happens outside traditional systems of record.

Joel De La Garza interviews internet infrastructure pioneer Barrett Lyon about his new venture, DocsNet. They discuss the stagnation of internet protocol design, the erosion of privacy in centralized messaging apps, and Barrett's goal to rebuild a peer-to-peer architecture that eliminates central servers. The conversation also includes a detailed account of a near-fatal aviation accident Barrett experienced.

A discussion among a16z growth team members regarding their "State of Markets" presentation. The conversation focuses on the massive scale of AI infrastructure investment, the distinction between current market growth and the dot-com bubble, and the transition toward autonomous agents and physical infrastructure.

Anisha Charya and David Pollan discuss the rapid evolution of personal AI agents, moving from general LLMs to proactive consumer assistants. They explore the current landscape of specialized vs. horizontal agents and the utility of the Assistant Benchmark for evaluating real-world performance.

Ben Horowitz och Martin Casado samtalar med TypeSafe AI-grundaren Diogo Almeida om Jev och visionen att integrera intelligens direkt i mjukvaran snarare än att bara automatisera kodskrivande. Diskussionen fokuserar på skillnaden mellan "just-in-time software" och "smart software" samt behovet av nya primitiver för att möjliggöra verklig automation.

Katie Kirsch interviews Maggie Landers, VP of Talent at Harvey, regarding the company's rapid scaling and cultural evolution. They discuss the transition from a small startup to a large organization, the "progress over perfection" mindset, and the challenges of onboarding a massive influx of new employees while maintaining a high-performance, AI-native culture.

En diskussion om AI-säkerhet, reglering och riskerna med agent-svärmar. Deltagarna debatterar spänningen mellan frontier-labben och säkerhetscommunityn, samt kritiserar hur kommunikationen kring existentiella risker (X-risk) påverkar den regulatoriska debatten.

A discussion introducing Cosine, a new product focused on professional reputation and peer-to-peer credentials within the startup community. The speakers explore the value of "cosigns" (endorsements) over traditional resumes and the importance of identifying talent before the market does.

Eddie Lazarin, general partner at A16Z Crypto, discusses AI safety, accelerationism, and the "P(abundance)" perspective. He argues that current AI incidents are primarily cybersecurity and control failures rather than signs of emergent superintelligence, advocating for technical controls and open-source development over the "AI doom" narrative.

A discussion on the evolution of the payments industry over the last 25 years, featuring insights from PayPal pioneers. The conversation covers the durability of the credit card interface, the impact of the EMV chip transition, and the potential for AI agents to reinvent commerce.

En diskussion om utbildningens roll i AI-eran, där fokus ligger på att uppmuntra unga människor att ifrågasätta etablerade sanningar och följa sin nyfikenhet genom "side quests". Samtalet berör kritik mot dagens preskriptiva högskolesystem och presenterar visionen för Horowitz & Andreessen Academy.

Ben Horowitz and Gagan announce the launch of the Horowitz Andreessen Academy, a new educational institution designed for the AI era. They discuss the failure of traditional college systems to keep pace with rapid technological shifts and the need for a "bundle" of education, status, and practical "learning by doing" to prepare young builders for the AI revolution.

Stephen Sinofsky, board partner at a16z and former Microsoft Windows president, argues that the current discourse around "AI alignment" and "rogue agents" is misleading. He contends that AI failures should be treated as software bugs and that the industry needs to move from a research phase to a professional software engineering phase characterized by robust telemetry, debugging, and incident reporting.

Ben Horowitz och gästerna diskuterar grundandet av Paid in Full Foundation, en organisation dedikerad till att ge ekonomiskt och kulturellt erkännande till hiphopens pionjärer. Samtalet belyser behovet av att kombinera finansiellt stöd med värdighet och heder för att motverka decennier av exploatering inom musikindustrin.

A16Z partner Josh Ellman discusses the evolution of consumer technology and the "Cambrian explosion" of AI-driven consumer experiences. He analyzes the transition from COVID-era digital saturation to a new era where AI agents enable users to be more present in the physical world by automating productivity and life management.

Databricks CEO Ali Ghodsi discusses the tension between AI frontier pacing and business competition with Martin Casado and Sarah Wei. The conversation covers the irresponsibility of existential risk narratives, the "tragedy of the commons" in AI development, and the distinction between pacing and actual safety.

A discussion with the co-founder and head of engineering of FAL regarding the H3 Max video model. The conversation focuses on the intersection of post-training and systems optimization to achieve significant gains in speed and cost efficiency for generative video.

A16Z partners Alex Rampel and Joe Schmidt interview Keith Paris, CEO of Lightfield, regarding his transition from Tome to Lightfield. The conversation explores the decision to pivot away from a high-growth AI presentation tool due to a lack of product conviction and the subsequent development of a "business world model" to reimagine the CRM.

Elena Berger and cyber ethnographer Ruby Justice-Gelow discuss the concept of "para-content"—content about content—and how it creates a gap between online perception and real-world behavior. They use wellness and body optimization trends, such as GLP-1s and peptides, as case studies to explore how niche subcultures are amplified and professionalized, particularly within Silicon Valley.

OpenAI medgrundare och president Greg Brockman diskuterar övergången till AGI-eran, utmaningarna med beräkningskapacitet (compute) och behovet av att skala säkerhet och alignment i takt med modellernas förmåga. Han reflekterar över OpenAI:s tidiga forskning kring RLHF och hur säkerhetsfokus har skiftat från ytliga filter till arkitektoniska lösningar.

Justin Johnson, co-founder of World Labs, discusses the launch of Atlas, a multimodal world model designed for spatial intelligence. He explains the model's core capabilities in generation, reconstruction, and simulation, and how it differs from traditional language models by focusing on physical and visual understanding.

Anish Acharya, general partner at a16z, discusses the psychological and economic impact of AI. He challenges the "permanent underclass" narrative, arguing that AI amplifies human agency and that most real-world problems are not solely intelligence-bound.

A detailed overview of the a16z Speedrun program, focusing on the founder experience, the operational structure of early-stage AI companies, and the venture capital approach to failure and the power law. The speaker describes the transition of founders to San Francisco and the program's goal of accelerating company inception in the AI era.

A discussion on the evolving nature of the power law in technology investing, specifically how frontier AI differs from traditional startups by converting capital directly into compute to compound advantages. The participants explore the massive addressable market of AI as it targets labor and GDP rather than just software, and the resulting shifts in portfolio construction for institutional allocators.

A conversation with VALS founder and CEO Ryan Krishnath regarding the critical need for independent AI evaluation. The discussion covers the failure of public benchmarks, the risk of self-reported model capabilities (citing Llama 4), and the development of infrastructure to measure recursive self-improvement and enterprise ROI.

Lisha Li from A16Z interviews OpenAI researchers Mark Selke and Metabh Swani about how large language models like GPT‑5 are changing mathematical research. The guests describe concrete examples such as solving Erdős problems, the unit‑distance conjecture and the recent ASTRA results, and they discuss the strengths and limits of AI reasoning compared with human intuition. The conversation also touches on how AI may reshape the practice of mathematics and what this means for the research community.

I det här avsnittet av A16Z‑podden diskuterar värd Sophia Du med forskaren Lucas Kaiser hur AI‑kraften idag är koncentrerad i stora företag och om öppen källkod kan förändra detta. De går igenom transformer‑arkitekturens resurshöga krav, möjliga forskningsgenombrott som kan möjliggöra mindre modeller och hur akademi och hobbyister kan bidra med egen utrustning. Trots nuvarande koncentration uttrycker de optimism för en framtid med mer distribuerade och specialiserade AI‑system.

Julie Yu, former founder of Kairos and current head of the healthcare investing team at Andreessen Horowitz, discusses how AI can transform the costly and inefficient healthcare industry. She recounts her 18‑year journey from software engineering to health tech, the patient‑access paradox, and the impact of COVID‑driven telehealth on regulation and payment models. Yu highlights the biggest opportunities for founders, from AI‑native care delivery to robotics, and envisions a future where an AI doctor lives in every pocket.

I det här avsnittet av MTS‑podden diskuterar Box‑s medgrundare och VD öppna viktsmodeller och deras inverkan på AI‑ekosystemet. Gästerna Theo Jaffe och Sofia Puccini samt referenser till Aaron Lebby och Jensen analyserar fördelarna med öppen innovation, debatten kring distillation i Kina och behovet av större amerikanska investeringar i öppna modeller. Diskussionen berör även säkerhetsaspekter, konkurrens mellan USA och Kina samt möjliga framtida finansieringsscenarier för stora öppna AI‑lab.

I det här avsnittet diskuterar Martin Casado tillsammans med World Labs‑grundarna Fei‑Fei Li, Justin Johnson och Ben Mildenhall det nya AI‑systemet Atlas. De förklarar hur Atlas använder ny vy‑förutsägelse för att kombinera 3D‑rekonstruktion, videogenerering och simulering, och visar exempel som bullet‑time‑videor skapade med bara tre iPhone‑kameror. Diskussionen belyser skillnaderna mot tidigare video‑ och språkmodeller samt framtida möjligheter inom dynamik, interaktion och robotik.