achieve ambition with intentionality, intensity, integrity & insanity. affiliations: - @smol_ai - @cognition - @temporalio - @aidotengineer - @latentspacepod

san francisco / singapore
Joined April 2009
💁‍♂️ How to Play Long Term Games: Systems > Goals Discipline > Motivation Trust > Distrust Principles > Tactics Writing > Reading Vulnerability > Confidence North Stars > Low Hanging Fruit Trends > News Habits > Sprints Questions > Answers Problems > Solutions People > Projects
396
910
99
5,776
Last week, @bernhardsson (Founder & CEO, @modal) announced a $750M Series D. I sat down with him the prior week to discuss a ton of stuff about the business. We got into: -why 2027 might finally be the year of non-Nvidia chips -why open-source models made fine-tuning a lot less necessary -how Modal's team grew 4x in a year and could hit ~700 people next year -why the frontier labs have the same business model as music artists and much, much, more… Enjoy! Timestamps: 0:00 What @modal does 1:25 From Spotify to building Modal 4:19 Founding before the generative AI boom 7:43 First customers and fundraising 10:00 Building for inference and autoscaling 13:21 Where Modal gets GPU capacity 15:39 Renting vs. owning infrastructure 19:45 The risk of GPU prices falling 24:18 Is AI demand infinite? 26:12 Alternatives to @nvidia 28:25 What Modal can learn from neoclouds 30:26 GPUs as an asset class 31:55 Modal’s products: functions and sandboxes 36:03 The case for smaller training workloads 41:21 How agent sandboxes are changing 46:48 @AnthropicAI and sandbox security 49:29 Using agents to test security 52:51 Developer experience in the age of agents 57:45 What Modal’s forward-deployed engineers do 59:55 Modal’s approach to product management 1:03:29 Optional code reviews and engineering culture 1:05:56 Cloud, on-prem and bring-your-own-cloud 1:09:25 Capacity, fundraising and growth 1:13:07 Scaling the team 1:17:28 Erik’s tweets: ephemeral sandboxes 1:18:31 The problem with rectangular GPU contracts 1:21:20 Inference providers vs. model builders 1:23:25 Where models improve next
2
2
1
20
6,919
imo this is the right attitude - dont blame AI for finding the bugs, just fucking lock in on clearing all the bugs asap latent.space/p/ainews-openai…
The folks at @latentspacepod with the perfect take, as usual:
46
12
5
205
20,195
I’m starting a new early-stage AI fund, doing what I love: partnering with founders from company formation on bold bets, long before consensus. The fund’s LPs will include Sequoia, Jeff Dean, Sachin Katti, Philip Wong (my Stanford professor and Chief Scientist adviser to TSMC), and others whose work has shaped our industry and how I think about AI’s future. I’ll remain a partner at Sequoia and keep serving on my boards on Sequoia’s behalf. Sequoia has been my home for the past 11 years. I joined as the youngest investor the firm had ever hired. I grew up here. Through years of intense debate and shared work, I found lifelong mentors and friends who took me under their wing and taught me this craft. They taught me to aim high and take big swings. I’ll carry that into what I build: outsized ambition, relentless truth-seeking and curiosity, and a refusal to become complacent. Working alongside founders has made me want to build something of my own: a firm shaped from day one around the convictions I’ve developed over the past decade. Life has taught me not to take time for granted. This is how I want to spend mine. Why now We’re seeing a new way of building companies in today’s AI era. Research and products often advance together. A breakthrough can open a market or accelerate a business far faster than expected. Some companies require extraordinary capital and compute early on. Others reach meaningful scale with remarkably small teams. The rules for how big these businesses can get, and how fast, are being rewritten. But early on, the technology is unproven, the resources required enormous, and the vision seemingly implausible. This is what I love most: seeing the potential right from the beginning, and helping founders realize their full scale of ambition. More than ever, it starts with an extreme willingness to believe early, with bold, independent conviction in a founder’s technical insight before any product or market validation. Then, it means helping founders take the biggest swings, giving the rare, long-term patience to see those bets through uncertainty, ambiguity, and turbulence, and serving as a level head when the stakes are highest. I learned this kind of partnership at Sequoia. It’s the foundation of the firm I’m building. Believing before consensus My fascination with AI’s potential took root at Stanford in the early-to-mid 2010s, studying computer science and working in Andrew Ng’s lab. The friendships I formed connected me to a community whose people and ideas would help shape modern AI and build some of our generation’s defining companies. What stayed with me was the gap between what those people saw was possible and what the wider world could imagine. I’m drawn to founders pursuing ambitions that sound impossible today. I start with a simple question: “What if everything goes right?” Then we work backward together to make it happen. In 2023, Deepak and Abhinav set out to build a general-purpose robot brain at @SkildAI, before physical AI was a consensus bet. Their sim-to-real breakthroughs gave me conviction where earlier robotics approaches had fallen short. But the data flywheel required deployment at scale. We made commercialization as urgent as research and brought in strategic partners like NVIDIA and SoftBank to support the ambition. Three years in, Skild is advancing the frontier of physical AI with nine figures in annualized revenue. In 2024, Ioannis and Misha founded @reflection_ai with an insight into scaling reinforcement learning. As that grew into the ambition to build a Western frontier lab for open models, the hard question was how a startup competes with labs spending tens of billions. An introduction to NVIDIA and many compute calls and recruiting events later, the team has built Beam, now one of the leading US open models. In 2025, Anna and Azalia, the co-creators of AlphaChip, founded @RicursiveAI to design models and hardware together. From day one, we made early customers and ecosystem partners a priority. Today, they are showing order-of-magnitude improvements on real workloads, working with some of the industry’s most consequential players, and reaching a defining inflection point. The ambition keeps growing as these companies scale. @linear has always had a kind of magic to it. Now it’s the system a new generation of AI companies builds on, for teams and agents alike. @tavus saw early that we’d one day talk to AI the way we talk to each other. Today, it’s building the human computing interface for a world of AI colleagues. @Mach_Industries began as hydrogen experiments in an MIT dorm room. It has since become one of the most exciting next-generation defense companies. Making the ambition possible Being a first believer is only the beginning. Compute falls through. A deep-pocketed incumbent makes your mission its top priority. A breakthrough changes the opportunity and the capital required overnight. These are stomach-churning moments, and early belief carries a responsibility to help make the bet work. This is the partnership I love: turning the full ambition into concrete steps. Raising capital, securing compute and strategic partners, recruiting exceptional people, and winning the first customers together. Thinking several moves ahead while solving what’s in front of us. And as breakthroughs open new possibilities, helping founders dream even bigger. Thank you To the founders I’ve worked with: thank you for trusting me and Sequoia to be your first partner and first believer. It’s been the joy of a lifetime, and I can’t wait to keep building together. Thank you to Alfred and Pat for your leadership of Sequoia and your support of my next chapter. To Roelof, for bringing me into Sequoia and believing in me from the beginning. Alfred and Roelof, you have been mentors in investing and in life. So much of what I know, and the investor I’ve become, comes from your guidance and generosity over the years. To Bill, for the close partnership and wealth of knowledge, always wrapped in humility. To Shaun, for showing me how to think deeply and imaginatively. To Michael Moritz, for inspiring 22-year-old me to dream big. To Doug, for teaching me to keep things simple. To Bogie and Jess, for bringing joy to the everyday and perspective on what matters. And to all my Sequoia teammates, for the friendship, camaraderie, and lessons I’ll carry with me. To my family, especially my husband and my brother, who have each started companies of their own: thank you for giving me the courage to take this leap. Let’s get to work!
348
76
58
1,795
224,285
Opus 5.5 is the best work estimator model i've ever used. This is calibration that is increasingly important as we delegate more and more work — don't tell me that it's going to take 2-3 working days (what a human would do) when for you, an enthusiastic, indefatigable, and embarrasingly parallelizable AGI, can get it done in 2-3 hours.
173
8
6
245
19,665
My AIE talk is live! I talk about the benchmark I've been building for 5 months now and released it publicly for the first time there :) Louis-François Bouchard of Towards AI breaks down what makes AI writing feel like slop: overused words, vague generalities, saying nothing, repetition, hallucination and a voice that sounds like everyone at once. He then shows how his team built a benchmark to measure whether a model can write a video script in his own voice: 10 scripts, 5 drafts per model, 4 LLM judges, nine scoring metrics and 1,000 human A/B checks. He shares what it cost, why judges favor their own model family, and practical tips for writing with AI: very long prompts full of examples, one review pass from a strong model, many cheap drafts merged together, and always a human edit. Thanks again @aiDotEngineer, @swyx for inviting me and hosting yet another amazing event. Watch it here: youtu.be/_dTF3NpPzGo
4
4
2
9
3,349
this is my quant
18 million miles traveled, 400 red hoodies, one 260-digit number, and 130+ former founders. We asked some of the people building Devin to tell us how they got here and what life at Cognition is really like. Here’s what they shared.
80
162
82
3,523
1,335,403
ratioed @cognition lololololol never change twitter
9
52
14,881
I just published my roadmap for building evals on highly subjective tasks. Here are the key steps involved and practical tricks for reducing noise / increasing confidence… Blog is here: cameronrwolfe.substack.com/p… Why is this difficult? Evaluating agents is already hard due to their broad scope, emphasis on expensive / long-horizon tasks, and non-determinism. Subjective tasks make this even harder because we cannot deterministically verify solutions and must rely on human judgments (or LLM judges). This introduces additional noise into an already difficult evaluation process. Evaluation roadmap. We can build reliable evaluations using a repeatable framework with three components: - Human evaluation: Our North Star for quality. Create quality guidelines that clearly define quality with the help of subject-matter experts, calibrate human evaluators to ensure agreement, and continually monitor evaluation quality. - Offline evaluation: Use calibrated LLM judges or task-specific rubrics as an efficient proxy for human evaluation. This allows us to scale experimentation without relying on expensive human evaluation for every new model. - Production monitoring: Track quality signals from real-world agent usage and use A/B testing to validate improvements before fully deploying a new agent. Multi-stage evaluation. These components form a tiered evaluation process: Offline evaluation → Human evaluation → Production A/B test → Full deployment We can test many ideas automatically, send promising candidates to human evaluation, and validate improvements through production A/B testing. Continual production monitoring then allows us to discover new use cases / failure cases in production that we can add into the earlier stages of evaluation to create a feedback loop. Sampling useful data. Evaluation quality depends heavily on which tasks we select. We want tasks that are diverse, realistic, difficult, discriminative, and accurate. Techniques like structured task taxonomies, model-based difficulty filtering, Item Response Theory (IRT), and LLM-based auditing help us construct more informative evaluation sets. Handling noise. We should leverage statistics throughout the evaluation process: - Power analysis: Estimate the number of evaluation samples needed to detect meaningful improvements. - Confidence intervals: Quantify uncertainty and provide clear confidence signals alongside evaluation results. - Re-sampling: Repeat agent / judge evaluations to reduce within-task variance and potentially tighten confidence intervals. - Paired comparisons: Evaluate models on the same tasks to more reliably detect performance differences. Key takeaway. An improved benchmark score does not always mean our agent is more capable. We need well-designed evaluation tasks, calibrated human / model evaluators, and statistical confidence signals to distinguish meaningful capability improvements from noise (especially on subjective tasks). By combining these techniques into a repeatable evaluation process / roadmap, we can make building and maintaining evaluation suites more systematic, reliable, and scalable.
41
59
7
650
45,764
Everyone wants an agent that says "done." Nobody wants to find out it wasn't. @zackproser and I did a full workshop at @aiDotEngineer NY on making agents prove it. Recording's up 👇
4
2
15
2,751
Your agent's compute needs change mid-run. @akshat_b of @modal walks through CPU and memory demand in Ramp's sandboxes. Provisioning every sandbox for a brief peak can leave capacity reserved when it isn't needed. #AIEngineerNY
4
3
43
5,361
When we framed @trq212's pod about Claude Code's Next Era, we meant it. we took some directive from @_catwu and did a proper deep dive on the new Claude Mods and highilghted our faves and what you should try now. in the same way that LLMs had that evolution from specific -> foundation -> specific, CC is much more of a malleable, general purpose hardened harness that has "pretrained" on quadrillions of tokens of learning that you can now repurpose for any domain specific harness! thanks to @ricmac and @daniel_mac8 for working on our latest product dive. if you have stuff you'd like us to test and write up, we're open to chat at [email protected]
There's not enough buzz around Claude Mods. Read this Claude Mods article if you want to: - Understand Claude Mods basics - Install and use the most useful mods today - Ponder where the agent-human interface will go For @latentspacepod.
67
6
3
129
11,997
highly recommend everyone to have private alt accounts*. the x web is bad at account switching, but on mobile it is frictionless. its basically the same "group chat but its ok if you dont read everything" experience but for your mutuals. tips - assume everything is public, screenshotting is real, dont be stupid. this is not the "i shit talk people behind their back", the right attitude is "this isnt high signal enough for my main account but i'm 1) taking notes for self, 2) have a throwaway thought, 3) i'm happy to chat about it" - give mutuals critical feedback that they might appreciate. "praise in public, criticise in private". because its not public, they know you actually do mean well - link/replyguy your alt for people to discover periodicially in case they dont know you have one * i like it so much that i call this the "insta/finsta" pattern - every social network should by default give people a public profile and a private one by default. tbd on the third account, the "botsta".
33
1
1
151
19,495
btw at 100k views in 24hrs, @jediahkatz's talk is an early candidate to be THE breakout talk of aie nyc the bar is set very high for @poteto's keynote at AIE CODE next month hahaha
Inside Grok Bot: is the agent harness dead? @jediahkatz of @SpaceXSI explains why smarter models let Cursor remove prompts and guardrails, and why Grok Bot is adding them back. Watch his AI Engineer New York talk: youtube.com/watch?v=7Y4sbZgB…
58
11
2
200
22,670
i retired from the business of making single-stock calls a decade ago but trust me when i say that the stock market still has no idea that what happened every single gpu sold gained ~30% in lifetime value today and the stock should be up by about $600B lol the more you buy, the more you buy
deepseek's production numbers are kinda wild: 155T tokens/day from one cluster (96% input cache hits) and $38 revenue/GPU-hour at off-peak prices. at $5–8/GPU-hour, that’s ~80% margin on GPU costs alone. at DeepSeek prices!! we dug into the math, their 72-hour cache, and what actually got open sourced in today's AINews: latent.space/p/ainews-deepse…
83
17
5
432
57,568
@willcb i think its about throughput per mw being much better distributed first of all but also jevons. we are clearly living in a supply-constrained world, not demand constrained. more tokens = more revenue, not more tokens = less price = same revenue
12
1
24
6,046
swyx retweeted
Judge for yourself whether @seejayhess just made "orbin' out" happen youtube.com/watch?v=Vki6cklF…
2
4
1
20
13,986
deepseek posted way more alpha on their zhihu page, if you only read twitter you are reading the laowai approved stuff "Based on this framework, we have implemented a high-throughput, low-latency large-scale inference service for DeepSeek V4.1 Flash. Currently, the service throughput of a single cluster is 155 trillion tokens/day, and the overall revenue is $38 per card per hour (based on off-peak prices)." it costs roughtly $5-10 per gpu hour so depseek is claiming 80% serving margin at THEIR prices lololololol
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/DeepI… 1/4
105
141
31
2,290
136,699
i tried to make codex make animated explainer videos but ngmi inf.latent.space/blender/sto…
4
10
4,996
Insane alpha, must read, might well redefine this economy. Everything me, Zephyr etc have been saying about the real profitability of the frontier inference is powerfully vindicated
7
62
5
983
47,452