Sequoia Capital

Anthropic公司Claude Code的创造者Boris Cherny与红杉资本合伙人Lauren Reeder在2026年AI Ascent大会上共同探讨编程的未来走向。他解释了为何自己在2026年一行代码都没写过，为何如今每天用手机提交数十个PR，以及为何他认为编程实际上已被解决——至少对他所写的代码而言。此外，他还谈到：为何循环是未来趋势，为何他认为一年后Claude Code本身可能只剩100行代码，以及为何印刷术的发明才是即将发生在软件领域的正确类比。 00:00 开场 00:55 Claude Code现场调查 02:39 Claude Code的起源故事 03:35 从自动补全到智能体 05:07 编程是否已被解决 06:50 Boris的个人工作流程 08:51 未来团队与通才 10:26 SaaS末日预言 12:57 观众深度问答 23:35 总结与下一步

Demis Hassabis，Google DeepMind联合创始人兼CEO、因AlphaFold获得2024年诺贝尔化学奖，与红杉资本合伙人Konstantine Buhler在AI Ascent 2026展开了一场涵盖通用人工智能发展路径及未来影响的深度对话。他阐释了为何相信AGI将在2030年前实现，药物研发周期如何从十年缩短至数天，以及为何应将信息而非物质或能量视为宇宙中最基本的元素。此外还探讨了：爱因斯坦会对当前模型局限性提出何种见解，以及未来一两年为何对人类至关重要。 00:00 开场 00:38 Demis的成长轨迹与共性 01:29 游戏作为AI训练场 02:59 Elixir Studios创业启示 04:39 2009年创立DeepMind 07:25 DeepMind使命与AGI赌注 08:52 科学导向的AI文化 10:37 生物学突破与Isomorphic 12:29 通过模拟探索新科学 20:29 意识哲学

Stark showed up to the field with a lightweight plastic case. The soldiers asked for wood — because wood burns, and it’s cold in winter. Co-founder Uwe Horstmann on why Stark builds for the staff sergeant, not just the general. #DefenseTech #Manufacturing

Chase Lochmiller runs one of the most operationally complex companies in AI: Crusoe builds power plants, manufactures electrical equipment, operates gigawatt-scale data centers, and runs a managed AI cloud on top, all under one roof. He just closed a round at a $31 billion valuation, and he got there by selling off the Bitcoin mining business that was paying the bills. Chase explains why Crusoe is modeled on Exxon, not a cloud provider, and why he partners with companies that look like competitors. He makes the case that a company that outgrows the person who got it there is the rule, not the exception, and that CEOs almost always fire too slowly. He also talks candidly about hiring a coach, working hard to become a polished speaker when it didn’t come naturally, and why the data center industry’s biggest problem is a marketing problem. I meet with a group of Sequoia founder CEOs between 20 and 2,000 employees, and one of them told me his business is becoming dizzyingly complex beneath him. Chase has the clearest answer I’ve heard for how to hold that in your head. And yes, we talk about summiting Everest. 0:00 – Introduction 2:12 – The boiling frog: how a founder holds a 2,000-person company in his head 8:16 – DRIs, the weekly staff meeting, and saying it until you're sick of it 11:19 – Urgency at scale: Elon's "change the world or die tomorrow" tension 14:55 – Startups die of suicide, not homicide 19:29 – Energy first: flared gas, Iceland hydro, and nuclear 23:26 – Bitcoin to AI was the plan all along 32:36 – Polish is learned: coaches, public speaking, and Felicis's self-improvement budget 37:48 – Firing too slowly, and when the company outgrows the person 42:39 – Welders and AI researchers, the data center backlash, and mining for bad news 50:54 – Everest: designing around the Khumbu Icefall's stochastic risk 57:31 – Brian's takeaways

Two years after its founding, Stark has roughly 700 employees and strike drones in the field in Ukraine. CEO Uwe Horstmann and CTO Johannes Schaback argue that the real lesson of that war is not any single weapon: the innovation ecosystem behind it is the wonder weapon. They explain why Stark designs every system for mass manufacturing from day one, runs 10 to 15 programs like a venture portfolio, and views the staff sergeant who operates a system as just as important as the general who buys it. Johannes recounts the publicly reported failed trial that became a forging moment, as engineers worked around the clock at a freezing test range until they landed three consecutive hits. They discuss NATO’s principles of responsible AI use and the hard-coded constraints in Stark’s autonomy stack. And Uwe lays out his next big idea: defense contracts structured like cloud contracts with surge capacity built in, so production itself becomes deterrence. Hosted by Pat Grady and Luciana Lixandru, Sequoia Capital Chapters 0:00 – Introduction 1:03 – Billions in contracts in 18 months: Stark's four-part playbook 2:28 – The staff sergeant matters as much as the general, and wood beats plastic 7:42 – Why Europe is ready for a neo prime, and why procurement reform will last 11:45 – Startup DNA, venture-style programs, and "as complex as a washing machine" 16:27 – Born out of fire: the failed trial and the round-the-clock comeback 21:13 – Where Stark finds talent, and why defense is no longer off-limits 23:48 – Anatomy of a strike, from the wooden crate to the kill zone 27:22 – Vertical integration, multiple domains, and shifting left on manufacturing 32:53 – After the Hugging Face hack: guardrails on autonomous weapons 35:56 – Matching the primes on R&D and scaling to 700 people in two years 39:24 – Deterrence by contract, and Europe's next chapter with the US

Robotics never got the internet's gift of free training data, so someone has to manufacture it. Josh Gao, founder and CEO of Mecka AI, runs one of the largest robot data factories, and he explains what that actually involves: custom baseball caps and tactile gloves because nothing off the shelf captures force, 100-gigabit upload lines to move hundreds of petabytes, and custom quality gates. Josh lays out the training recipe the industry is converging on (roughly a million hours of egocentric human data, tens of thousands of hours of robot data, and a few hours of in-domain teleop ) and why tactile and simulation data still have to enter the "data pot of soup." He argues the field is still orders of magnitude behind Tesla's FSD, which spent a decade on two degrees of freedom while a robot hand has dozens. His take: there won't be a single GPT-3 moment for robotics, but a steady unlock of commercial tasks, starting with constrained industrial work and ending, he hopes, with robots mining asteroids. Hosted by Sonya Huang, Sequoia Capital 0:00 – Introduction 1:46 – What Mecka is: pixels, actions and logic 2:45 – Is robot data economically feasible without the internet? 4:15 – Teaching robots about carbon life 5:27 – Cloud kitchens, stablecoins and the path to robotics 7:03 – Entering robotics with a fresh lens 8:45 – What egocentric data is and why Mecka bet on it 10:09 – The first pilot: aunties and uncles with a mobile app 10:42 – Force data and tactile gloves 11:55 – Inside the operations of a data factory 14:41 – Heat-mapping the taxonomy of human work 15:42 – Quality gates, fraud and the McDonald’s POV videos 17:44 – Earning a margin as a data intermediary 19:33 – Millions of hours vs. Tesla’s decade 20:55 – Why it isn’t binary: unlocking commercial tasks along the way 22:02 – Home data is plentiful; home robots are the hardest 23:24 – Useful robots can be deployed today 24:35 – When robotics gets an FSD-style flywheel 26:42 – Deployment diffusion and the modern systems integrator 28:21 – Favorite vertical robot categories; mining an asteroid 30:08 – The hardware slate: Waymo-fying people 30:55 – EgoVerse 32:04 – The data pot of soup: the converged training mix 33:54 – Move 37 for robotics: AlphaGo and Federer 36:02 – World models and the training mix 36:58 – Will scaling the current mix get us to the promised land? 39:05 – Will the AGI labs steamroll robotics? 40:50 – Sci-fi roots and the Kawasaki Corleo 42:44 – How many robots in a decade?

Single-player Go would be the most boring game imaginable. David Silver on what happens the moment you add a second agent, and why simple environments can produce endless learning. #Go #AlphaGo #AI #ReinforcementLearning

David Silver on simple objectives in complex worlds: how evolution got all of life from a fitness signal, and how humans went from coordinating to communicating to trading to science. #ReinforcementLearning #AI #Emergence

Ineffable Intelligence's mission is to build a super learner: a system that starts knowing nothing and discovers knowledge beyond what humans know. David Silver on why humans are the only existence proof. #AI #Superintelligence #ReinforcementLearning

If you need a gigawatt at 99.99% reliability, you need one-plus-one redundancy, which means building two gigawatts, ideally clean, ideally right next to the data center. Amin Vahdat, head of AI infrastructure at Google, on the statistical multiplexing argument for partnering with utilities instead: spread the risk over a much larger base, feed power back to the grid at peak residential demand, and Google, the grid and the neighbors all win.

David Silver on what happened when his team gave up human data, and why he says LLM labs are "human-in-the-loop learning algorithms with thousands of people in the loop." #AlphaZero #AlphaGo #AI #ReinforcementLearning

David Silver spent two years building AlphaGo on human data. When he removed human data altogether building AlphaGo Zero, the AI surpassed AlphaGo in under two days. He left DeepMind to run that same bet in pursuit of superintelligence. David and Lasse Espeholt, co-founders of Ineffable Intelligence, explain their plan to build a “superlearner”—an agent that starts knowing nothing and discovers science, mathematics, and language for itself by direct experience instead of through a pre-training run. Silver argues that human-language reasoning is a ceiling, not a foundation, that alignment via chain-of-thought monitoring is already reaching its limits, and that the science of how superintelligences coexist should be worked out in simulation before anything touches the real world. Espeholt describes the GPU cluster and the lineage from IMPALA and SEED RL that makes the experiment possible. Silver’s forecast: strong evidence within a year, superintelligence in three. Hosted by Sonya Huang, Sequoia Capital 0:00 – Introduction 1:09 – What held true from Go to today, and what the LLM era changed 3:35 – Why leave DeepMind to start a new lab 5:58 – Why "Ineffable" (and the secret second reason) 7:03 – Lasse's path: scaling RL at DeepMind and Google Brain 9:31 – Defining a super learner and first contact 12:35 – Super learners vs. LLMs that make discoveries 13:59 – Day to day: the RL stack, crucibles and human-journey milestones 16:13 – Don't breadcrumb the objective: simple rewards, emergent complexity 21:16 – What a crucible looks like 24:14 – Go: one agent is boring, two is infinite 25:41 – Will agents rediscover language? 27:22 – Removing human priors: AlphaGo to AlphaZero 32:05 – Keeping human bias out of the crucibles 33:35 – Confidence, timing and super forecasting 36:24 – Is compute the driver? 38:13 – The ceiling on simulation and Rich Sutton's objection 43:44 – Alignment and the science of superintelligence 47:45 – Monitoring agents that don't think in English 51:01 – Pacing the frontier 52:54 – The compute setup: Vera Rubin on GCP 55:18 – Training, inference, simulation and RL scaling laws 57:40 – IMPALA and SEED RL 1:00:58 – 20 people and the compute-per-person ratio 1:04:49 – A research-led culture 1:05:47 – Predictions

Luca Ferrari is co-founder and CEO of Bending Spoons, a serial acquirer of AOL, Vimeo and Evernote, Airtable, etc, that buys to hold forever, and whose only real edge is that it operates software companies better than almost anyone else. Bending Spoons got 800,000 job applications last year and hired fewer than 300 people. Luca runs hiring the way a quant runs a portfolio — hundreds of signals tested against future performance — and he's blunt about what the data says. Harvard and Stanford barely predict anything. Interviews are the worst signal in the process unless they're scripted. The best candidates were politically active as teenagers and stopped in their 20s. He hires for slope over experience, which is why a 25-year-old routinely shows up to run a company Bending Spoons just bought, and why he thinks experienced hires from big tech lose motivation and struggle to adapt. He also explains why Bending Spoons has no executive team, why the burden of proof always sits with whoever wants to add complexity, and why he stayed in Milan on purpose. I called three founders whose companies Luca acquired before this interview. All three, unprompted, said the same thing: exceptionally high integrity. His answer about where that comes from is the most honest closing I've had on this show. Chapters: 0:00 – Introduction 1:13 – The Bending Spoons model: buy to hold forever, then rebuild from the ground up 8:07 – 800,000 applicants, fewer than 300 hires: industrializing talent density 13:45 – Hiring like a quant trader: hundreds of signals, scripted interviews 21:45 – Why strategy and talent trump org structure 25:34 – Leader or manager? Head in the clouds, feet on the ground, but never both at once 30:38 – Slope over experience: why 25-year-olds run the companies Bending Spoons buys 41:08 – Radical candor, the 360 that found Brian's bugs, and picking one flaw a year 47:00 – No exec team, no sacred org chart: whoever is CEO should rewrite the system 51:27 – Question everything, add nothing: the burden of proof is on whoever adds complexity 57:08 – Why Milan? The uncool is cool; the world needs more than one Silicon Valley 1:01:16 – Never lie, never omit: the high-integrity code 1:04:02 – Brian's takeaways

At 100,000-accelerator scale, something fails multiple times an hour, which is why Google's Chief Technologist for AI Infrastructure Amin Vahdat thinks FLOPS is a vanity metric. The metric that matters is what he calls goodput: the useful work a workload actually delivers through real failures. Amin walks through the calculus that split the TPU line into 8i and 8t for the first time, why the TPU's core primitives haven't changed since v1, and how Google and DeepMind co-design in the same rooms, intercepting chip architectures mid-flight before tape-out. He explains why long-horizon agents are sending demand for CPUs and storage through the roof alongside accelerators, how optical circuit switches reroute light to a spare rack in milliseconds, and why Google would rather wait on a utility than build its own gigawatt. We also cover orbital data centers and the multi-megawatt rack of 2036. Hosted by Sonya Huang, Sequoia Capital 0:00 – Introduction 1:47 – What makes a data center an AI data center 5:30 – Goodput, not FLOPS: holding yourself accountable when something fails every hour 11:52 – Doubling token capacity every six months, and where the gains actually come from 15:32 – The TPU bet: from a contrarian call in 2013 to splitting 8i and 8t 23:30 – The case for and against co-design 26:11 – Shoulder to shoulder with DeepMind: intercepting chips mid-flight 34:16 – Long-horizon agents change the shape of the data center 37:50 – Optical circuit switching and the state of networking 42:35 – Power is the binding constraint: utilities, gigawatts, and sizing a data center 49:23 – Training vs. serving clusters, seven-year-old TPUs, and open standards 58:29 – Orbital data centers and the supercomputer of 2036

模型已就绪，组织尚未跟上。Ali Ghodsi 探讨为什么推广人工智能意味着重新改造地球上每家公司的流程，以及为什么这至少需要十年时间。

Ali Ghodsi 用两天时间构建了一个连接器，而他的团队之前需要三个季度。他了解到时间究竟花在了哪里，以及他们如何实现每个季度交付七个连接器。

一次好的高管搜寻需要 6 到 12 个月。Ali Ghodsi 探讨为什么人们会妥协，为什么信号一直都在，以及为什么非正式推荐胜过了正式推荐。

Databricks 拥有 150 万美元营收，董事会却在面试外部 CEO 人选。Ali Ghodsi 谈论为什么 a16z 仍然选择了他，以及那段无人告知他的试用期。

Ali Ghodsi 从未想当 CEO。2015 年，Databricks 有一个极其成功的开源项目，150 万美元营收，董事会却在背后悄悄面试其他候选人。他得到了这个职位作为试用期。他们甚至没有给他 CEO 的薪水。当我在本播客中邀请 Ben Horowitz 并问他谁是他投资组合之外硅谷最好的 CEO 时，他毫不犹豫地说是 Ali。 我们深入探讨他实际上如何经营公司：找到公司里最大的单一瓶颈，然后在两三年内投入巨大且不成比例的注意力。这就是 Lakehouse 类别的构建过程，当时战略顾问、他自己的销售团队以及互联网上的大多数人都还在嘲笑这个名字。Ali 分解了他针对 Snowflake 的策略，为什么他拒绝了那些至今仍心存怨恨的杰出高管，以及他自己写代码时学到的经验。 他还直言，回避冲突会让你失去这份工作的资格。在 HubSpot 时，我过度看重受人喜欢。Ali 不会。 0:00 – 引言 1:22 – 秘密的 CEO 搜寻及董事会为何押注创始人 4:11 – 教授还是 CEO？总是选择更难的那条路 8:18 – 建立你自己的策略，将所有精力投入到一个瓶颈 12:16 – 扼杀产品引导增长，了解优秀企业销售人员的真正实力 19:09 – 提前招聘：销售领导者、高管和非正式推荐 27:02 – Snowflake 的竞争：研究你的敌人，绝不模仿他们 33:22 – Lakehouse：信念、嘲讽以及第二幕的案例 41:02 – 杀手本能及为何回避冲突的 CEO 会失败 44:10 – 邓巴数及围绕人工智能重新思考组织架构 48:00 – 通用人工智能已经到来——企业只是将其用作聊天机器人 52:55 – 他还在写代码吗？两天完成一个连接器对比三个季度 57:18 – 一天的生活，以及为什么周一会议不是演戏 1:04:53 – 为什么 Databricks 会上市，只是还没到时候 1:08:09 – 克服冲突回避，否则不要当 CEO 1:11:07 – Brian 的总结

Starting a company is hard. Reinventing your company for AI as a public company with quarterly earnings results is even harder. Aaron Levie has pulled off the transition with Box and offers hard-won advice for founders. The cofounder and CEO of Box argues the value isn't only in the model; it's in the bridge from a model's raw capability to the actual workflow inside a bank, a law firm, or a pharma company. That's the case for the application layer, and Box is building it: an agent harness tuned so tightly to its own file system, permissions, and search that it beats handing the raw API to Claude or ChatGPT on both accuracy and latency. Aaron explains why token subsidies from the labs can't last, why you want a model-agnostic company routing your tokens rather than the one selling them, and why coding diffused fast while the rest of knowledge work won't. (There's no "give us your GitHub" for a sales rep.) His prediction: within five years, 90% of enterprise tokens go to work no human user ever initiated. Hosted by Sonya Huang, Sequoia Capital 0:00 – Introduction 1:55 – Are application companies the hottest neolabs? 6:56 – Will the labs move up the stack? 12:34 – Box and betting the company on AI 16:50 – Hero use cases: reading a million contracts and long-running agents 18:42 – Work slop: why AI code is embraced but AI content isn't 24:08 – Building Box's agentic harness and the evals that matter 27:23 – The state of the model race 29:25 – Open-weight model adoption in the enterprise 32:34 – Memory, continual learning, and what belongs in the weights 37:29 – Box Labs and systems of record in a world of agents 44:55 – Will chat be the dominant UI for enterprise AI? 48:00 – Why coding diffused fast and the rest of knowledge work hasn't 54:31 – Staying wired in, making a company AI-first, and what it takes to win

A Florida county couldn't explain a spike in flood rescues. The agent traced it: three consecutive days of a weather pattern, carving sand channels, creating perfect rip current conditions. #ai #publicsafety

Nick Noone was flying in and out of Baghdad. Ben Rudolph was working refugee crises in Africa. In 2016 they decided to come together to work backwards to a single answer: safety sits at the bottom of the pyramid. #ai #publicsafety

Flock, Axon, and most of public safety tech grew by owning the input. Nick Noone on why Peregrine inverted the entire model, and why the problem was never how much data an agency had. #ai #flock #publicsafety

Peregrine's first agent was a cold case agent — and they tested it by asking a department to grade it against a case they'd already cracked.

Most public safety technology companies grow by collecting more data. Peregrine inverted the model: no sensors, no new data, a business built on connecting the data and information cities already own. Co-founders Nick Noone and Ben Rudolph received more than two dozen no's before San Pablo PD let them in the door in February 2018. Today, Peregrine powers law enforcement, emergency medical services, fire and rescue, and other services in more than 400 cities and communities globally. Nick and Ben explain their north star for data sovereignty, and discuss how Peregrine's philosophy and privacy-first approach to data access and ownership preserves individual privacy and cities' sovereignty. They walk through how AI and long-horizon agents are being deployed: a cold case agent that reproduced an exoneration detectives had reached by hand, a Wisconsin county that placed a suspect using cell records buried in 300GB of evidence, identifying threats to a synagogue, root-causing an escalation in weather-related incidents, and more. Hosted by Sonya Huang, Sequoia Capital 00:00 Introduction 02:07 What Forward Deployed Engineering Means 03:58 What Silicon Valley Gets Wrong 05:23 UNHCR, Dimagi And Downstream Data Problems 08:25 Why Cities, Why Safety 10:45 Two Dozen Nos And San Pablo PD 14:19 Building Through Defund The Police 18:16 The Inversion Of The Collection Model 21:20 Data Ownership And Governance 22:57 From Nice Search To Deep Analysis 29:59 Agents Writing The Integrations 31:45 The Cold Case Agent 35:02 The Anti-Network-Effect Proposition 38:40 Facial Recognition And Hard Decisions 40:48 Technology For The Underdogs 42:54 Trusting The Individual Contributor 48:50 Ten Thousand Cities

如果一项任务值得做，并且有可用的GPU，代理今天就会去做。那么明天还剩下什么工作？Parag Agrawal认为，剩下的唯一工作是由变化触发的工作——而其中一个触发因素就是网络上发生的一切变化。搜索不再是一个请求，而变成了一项持续指令。

如果持续学习有效，最终结果会是一个既能荡树枝又能造飞船的单一心智吗？Rich Sutton：“这是一个单一的设计。会有很多心智。”Khurram Javed的理由读起来像是一个证明——两个同样复杂的系统无法相互建模，因此没有哪个单一系统能学会一切。 #ai #llm #持续学习

Parag Agrawal讲述Parallel名字的由来及其背后的理念：一个为AI构建的平行网络，具有与我们为人类构建的网络不同的特性。这意味着你现在发布的每个页面都有两个受众，无论你是否有意如此。

《苦涩教训》可能是AI中被引用最多的文章，这也使它成为被误解最多的文章之一。Rich Sutton给出了他自己26个字的压缩版本——以及人们一直忽略的部分：它并非反对复杂算法，而是支持那些随计算量扩展的算法。 #ai #llm #机器学习

如果一个代理运行十次搜索来写一段文字，谁得到报酬？Parag Agrawal解释了沙普利值——一种用于分配协作创造收益的博弈论方法——以及Parallel如何利用它来估算任何单一来源对代理输出的实际贡献。 #ai #互联网 #网络搜索