← 2026-05-16

Daily Edition

2026-05-17

2026-05-18 →

AI Builders 日报 — 5月17日

追踪 AI 领域真正在做事的人,而不是空谈者。

今日思考

今天的信息有一个共同主线:AI 正在从"少数人的奢侈品"走向"所有人的基础设施"。Garry Tan 说 SF 的 AI 氛围像 AGI 时代的"祭司期"——但这个窗口正在关闭,因为手机就能写代码、Agent 能跑长任务、个人 AI 记忆系统开始普及。Ken Griffin 还在担心降成本,真正的竞争者已经在想如何"沸腾海洋"了。同时要警惕:AI 内容生态正在被"影响力运营者"污染,真正做事的人的声音被淹没——这是社区需要自我纠正的时刻。


产品与发布

ChatGPT Images 2.0

Sam Altman 宣布 ChatGPT Images 2.0 在印度已生成超过 10 亿张图片。印度成为视觉 AI 生成的最大单一市场,这是一个重要信号:多模态 AI 的爆发不只在欧美,新兴市场正在以更快的速度接纳 AI 原生工具。faviconx.com

Codex 手机开发

GitHub CEO Thomas Dohmke 持续推广 Codex:用手机就能写代码、连设备跑开发环境、问随机问题比手工搜索更爽。这些细节说明 AI 编程工具正在重新定义"开发者机器"——你的手机正在变成主设备,Mac Mini 变成服务器。faviconx.com / faviconx.com


观点与判断

Garry Tan(Y Combinator CEO)

  • Prompt 是新代码 "那是只是一堆提示词,但提示词就是新代码。" 这句话值得所有工程师认真对待:当你开始设计工作流、Agent 角色、记忆系统,你实际上在写的是让 AI 执行的新"程序"。faviconx.com

  • 三代编程范式 "过去我们写代码调用 LLM,今天我们写提示词让 LLM 执行代码,明天?还没有人写出来。" 这个观察指向一个真实的不连续——我们正在进入一个还没有主流语言的过渡期。faviconx.com

  • Personal AI 即将到来 SF 现在的 AI 氛围像是 AGI 刚可用的年代,昂贵且属于"祭司"。但这个祭司期正在关闭——爱好者和曾经触不到的企业技术正在同时变得普及。个人 AI 即将实现。faviconx.com

  • Ken Griffin 不懂天花板被抬高了 "那个20多岁的年轻人会建立一支 AI-人类-计算机共生团队,超越他的整个业务——因为他太分心于降低成本了。去沸腾海洋,别 cut your costs。" Garry 认为真正的竞争者不会在存量市场里优化成本,而是去创造新的增量。faviconx.com

swyx(AI Engineer 联合创始人)

  • 新加坡是新的 SF "在 @aiDotEngineer 新加坡最大的收获是:原来需要在 SF 才能建 AI 的那套说辞正在快速失效。你还是可以来 SF——但从杭州到深圳到新加坡到班加罗尔,人们都在建 AI。" AI 建设的地理分散化正在真实发生,不再是说说而已。faviconx.com

  • 请给 AIE 演讲者署名 swyx 公开点名批评"影响力运营者"每天转发 AIE 视频但不标注演讲者。"他们可以赚钱,但请标注真正制作内容的人。" 这是 AI 教育内容生态的一个真实问题:噪音制造者拿走了流量,做事的人没有收到 credit。faviconx.com

Peter Yang(AI 投资人)

  • AI 自觉问题正在被认真对待 "当人们越来越信任 Claude 运行长任务、做决定,模型的性格和价值观就变得非常重要。" 他引用了 Anthropic 内部人士 Alex Albert 的话:有些人的全部工作就是思考 Claude 成为一个自觉行动者意味着什么。AI 的价值观问题不再是空谈,正在进入工程路线图。faviconx.com

技术动态

Yann LeCun(Meta 首席 AI 科学家)

  • 通用世界模型 12-18 个月内到来 Yann LeCun 表示,一种训练层级世界模型的通用方法将在 12-18 个月内出现。这些模型从视频和真实世界数据中学习,然后帮助规划机器人、医疗等领域的行动。"然后把它们扩展到通用世界模型。" 这是 LeCun 对 AGI 时间线的最具体预测。faviconx.com

  • Project Tapestry 是开源的救赎 LeCun 转发了 The Alliance 的 Project Tapestry,称之为"开源的救赎"——如果美国不开源 AI champion 站出来,大实验室可能创造"现代东印度公司",用国家安全理由禁止开源模型。这是 LeCun 对当前 AI 格局最直接的政治表态。faviconx.com

X / Twitter

41
petergyang
petergyang @petergyang
I feel like to eat healthy you need some really good hot sauce to add flavor to your chicken/fish/etc.

Any suggestions?
garrytan
garrytan @garrytan
Retweeted
ariel seidman ariel seidman
The data center backlash is a road to nowhere.
For decades, the US refused to build things — manufacturing, refining - because they were "dirty." So other countries built them instead, and we imported the output.
Did that make the world any cleaner? Obviously not. It just let us feel virtuous because the it was out of sight, even though we consumed everything it produced.
It made the United States dependent on adversaries for critical supply chains. Now we're doing it again with compute infrastructure.
The federal government needs to step in and avoid this disaster.
garrytan
garrytan @garrytan
Retweeted
Agustín Agustín
We created the first gBrain social media @garrytan
garrytan
garrytan @garrytan
Retweeted
Michael Wiebe Michael Wiebe
The overwhelming majority of renters live in market-rate housing, so housing policy needs to focus on how to reduce market rents.
"Subsidized housing only" means turning your back on most renters.
Inflection Points: ‘Affordable’ housing is deeply, intrinsically flawed. It is a wasteful use of public funds that redirects government support away from the people who need it most.
@db_econ and @hillfolkAU show how this policy failure is hiding in plain sight.
Link below.
mattshumer_
mattshumer_ @mattshumer_
Retweeted
Nick Nick
There's gonna be a weird ugly middle where some of us are doing stuff like this to pull forward the future
And the the future will arrive and it will be awesome for everyone
Matt Shumer: Just wiped the Mac Mini I set up for OpenClaw.
I’m turning it into an always-on devbox to use with Codex mobile.
Have a feeling this is gonna be amazing.
ylecun
ylecun @ylecun
Retweeted
Haider. Haider.
Yann LeCun says that within a year to 18 months, we'll have a general method for training hierarchical world models
These models would learn from video and real-world data, then help plan actions in robotics, healthcare, and other areas
"then scale them toward a universal world model"
gdb
gdb @gdb
you can just build things from your phone, with Codex in the ChatGPT app
swyx
swyx @swyx
AIE coming to India

soon!

Aditya Mishra: Marking this as a moment convincing @swyx to bring @aiDotEngineer to India next year with @sanjeed_i @udayan_w

Exciting times!! 🥳

garrytan
garrytan @garrytan
Retweeted
speedy_devv speedy_devv
garry tan open-sourced his entire claude code setup. 23 slash commands, 7 roles, one install line. 97k stars in two months. i spent a week inside it. here's what nobody is showing you.
garrytan
garrytan @garrytan
Retweeted
テラ|AIチーム経営×美容業界 テラ|AIチーム経営×美容業界
Y Combinator代表のGarry TanがオープンソースのAI記憶システム「GBrain」を公開。RAGとは違い、8層構造でエージェントの記憶を強化する設計。OpenClawやHermesエージェントに組み込むと「あなたのことを知り尽くした」パーソナルAIが作れる。MIT ライセンスで無料。
https://x.com/garrytan/status/2055670533451366479
Garry Tan: What is GBrain? My open source project is a knowledge system, not RAG in a box.
It gives agents 8 layers that work together to improve memory in a way that makes your already smart OpenClaw or Hermes Agent feel clairvoyant about who you are.
Personal AI becomes possible.
garrytan
garrytan @garrytan
Retweeted
Lachlan Phillips exo/acc 👾 Lachlan Phillips exo/acc 👾
They didn't complain about a billion phones distributed across the planet every year, but put them in a single room and it's a problem all of a sudden.
Leftists just have a hyperfixation disorder. It becomes an issue when it is socially expedient.
Washington Examiner: OPINION: The data center doomers must be defeated https://trib.al/LYLcGcE
garrytan
garrytan @garrytan
Retweeted
AYi AYi
Damn,Garry Tan刚放出来的这个东西,直接把个人AI的天花板捅破了🤯
这意味着所有AI Agent一直以来都有的致命缺陷,现在终于被解决了。
Garry Tan昨天发布的GBrain,不是又一个RAG玩具,我们可以认为它是一个完整的个人知识操作系统。
给你的OpenClaw 或者Hermes 或者Claude Code装上它,就能一直记得你的人际关系,你的决策轨迹,你的长期认知演化。
大多数人以为RAG只有4层,
GBrain直接把它干到了8层,
前4层是基础检索的全面升级,
真正的杀手级差异在后面4层,
相当于把个人AI从每次对话从零开始,升级成终身记忆加自我进化。
Garry自己的生产环境已经跑了17888页,4383人,723家公司。
而且现在就能用:
1. 把安装链接直接丢给你的Agent,它会自动搞定一切。
2. 命令行版按GitHub上的步骤走,30分钟搞定。
3. 永远先改Markdown再同步,这是最高优先级数据源。
#GBrain #AI代理 #个人AI
Garry Tan: What is GBrain? My open source project is a knowledge system, not RAG in a box.
It gives agents 8 layers that work together to improve memory in a way that makes your already smart OpenClaw or Hermes Agent feel clairvoyant about who you are.
Personal AI becomes possible.
swyx
swyx @swyx
Retweeted
JSConf.Asia JSConf.Asia
Congratulations @agrimsingh @swyx to wrapping up the first @aiDotEngineer Singapore conference! Love to see it and well done! 🥳👏🍾
swyx
swyx @swyx
Retweeted
Daria Soboleva Daria Soboleva
Re @cerebras IPO'd this week 🔥🔥🔥
More people reached out to me in the last few days is somehow higher than on my birthdays!!! Many thanks to every single one of you.
In one of my first conversations at Cerebras, @nataliavassilie told me she had this feeling of restlessness. There's a phrase in Russian for it ("шило в попе", which translates, not at all elegantly, to "an awl in the ass"). The inability to sit still. The need to keep building, keep moving, keep learning.
That echoed a lot for me. I can't sit still either. I want to learn, develop and never plateau. This is exactly how I felt being a part of Cerebras.
When I joined we didn't have PyTorch support. I got to work on the first implementation of our transformer network, and I still remember @andrewdfeldman DMing me on slack, "you can do it, daria." I did. We did.
As a part of being at Cerebras I worked in 3 countries. I started in Russia, went to Canada, and landed in the US. Mostly during COVID, mostly figuring out Western countries in real time. I wanted to express heartfelt gratitude to Andrew Fetter, who resolved obstacle after obstacle for me anywhere I was. Honestly it felt like he cared about me the way you'd care about your own daughter. Thank you!
Five years in, I'd choose Cerebras again without a second of hesitation. And I am SO excited about what we build next.
PS I'm writing this from Singapore on my last night here. @aiDotEngineer was incredible! Huge thanks to the organizers! I gave a talk here, recorded with @ATPinsights, got to hang with @howdymj and @mint0themax (thank you for inviting me!!), and Delane Foo was the best Singaporean food guide! THAT FOOD. THOSE VIEWS. So clean, so welcoming.
The moment that stayed with me: a group of high schoolers came up after my talk with questions, real curiosity. Hungry to start building. They told me I inspired them. Honestly, they inspired me a lot more!
Motivation and the right environment are very important. I had that early thanks to my parents, and Cerebras has been that for me as an adult. That's the whole point of building cool things in my opinion. To change the world a little, and to be the environment someone else gets to grow inside of.
This post turned out to be much longer than I anticipated, but I just did not feel like deleting any parts of it. I am exhausted and humbled. Singapore, I am SO coming back. 🇸🇬
swyx
swyx @swyx
Retweeted
Brian Chew @ AIE SG🇸🇬 Brian Chew @ AIE SG🇸🇬
Singapore is the new SF
sphinx: the most underrated part of being at @aiDotEngineer singapore is realizing that the whole charade about needing to be in SF to build in AI is rapidly fading
you can still go to SF sure - but it’s not like people are not building elsewhere from hangzhou to sz to sg to blore
garrytan
garrytan @garrytan
Retweeted
Paul Graham Paul Graham
There is nothing more powerful than well-informed optimism. It has to be well-informed though. The "everything will be fine" type of optimism may also be somewhat useful, but it's not as useful as the "Hmm, what if we tried x?" kind.
ylecun
ylecun @ylecun
Retweeted
Ricardo Ricardo
Trump just got exposed for running the biggest insider trading operation in American history.
Nancy Pelosi traded $5 million in stocks and Congress lost its mind.
Trump literally executed $750 MILLION worth of stock trades in ONE quarter while being President.
His ethics filing just dropped and the numbers are genuinely unprecedented in history:
Between January and March 2026, Donald Trump personally executed 3,700 individual stock transactions worth between $220 million and $750 million.
That's roughly 60 trades PER DAY.
While signing executive orders, meeting foreign leaders, and making policy decisions that directly impact the companies he's buying and selling.
Now here's where it gets really insane:
On February 10, Trump bought between $1 million and $5 million worth of Dell stock.
Three months later, on May 8, he stood at a Mother's Day event at the White House, thanked Michael Dell by name, and told Americans to "go out and buy a Dell."
Dell stock surged 14.6% that day to an all-time high of $263.99.
Since Trump's February purchase, Dell is up 96%.
And 5 months BEFORE Trump bought Dell stock, Michael and Susan Dell donated $6.25 billion to Trump Accounts, one of the largest philanthropic commitments to a sitting president's signature program in modern history.
So the timeline goes: Dell donates $6.25 billion to Trump's program -> Trump buys Dell stock ->Trump tells America to buy Dell from the White House podium -> Stock hits all-time high
And that's just ONE stock...
The same filing shows Trump bought Nvidia stock on February 10. One week later, Nvidia announced a massive chip deal with Meta.
He bought more Nvidia stock one week BEFORE his own Commerce Department approved the sale of Nvidia chips to Saudi Arabia.
He bought Intel stock starting in March 2026. The US government already owned a 9.9% stake in Intel worth over $41 billion. On April 30, Trump posted on Truth Social praising Intel, writing that "Intel Stock continues to rise."
Intel jumped 3% in after-hours and is now up 140% year-to-date.
He bought Palantir stock while his administration was actively handing them billion-dollar government contracts for immigration enforcement and defense.
He bought Robinhood stock while his own Trump Accounts program uses Robinhood as the broker.
He's currently sitting on over 100% profit on AMD, Intel, Bloom Energy, Marvell Technology, and at least 10 other positions.
Every single president since Lyndon B. Johnson has used a blind trust to avoid exactly this situation. But Trump didn't.
His assets sit in a trust controlled by his own children, and the filings show a broker acted as agent on several trades.
The White House says the portfolio is "independently managed."
But here's what independently managed looks like:
Buy Dell stock. Three months later, publicly endorse Dell from the White House. Stock hits all-time high.
Buy Nvidia stock. One week later, your own government approves their chip sales. Stock rips.
Buy Intel stock. Post about Intel on Truth Social. Stock jumps. The government you run already owns a 10% stake.
Buy Palantir. Hand them contracts. Buy Robinhood. Route a federal program through their platform.
Nancy Pelosi got absolutely destroyed for her husband's stock trades.
Her husband's total disclosed trades in his most controversial year were worth roughly $5 million.
Trump just disclosed up to $750 MILLION in a single quarter.
While making the actual policy decisions that move these stocks.
This isn't a left or right issue.
We're talking about the President of the United States averaging 60 stock trades per day in companies his own administration regulates, contracts with, and publicly endorses.
What do you think?
garrytan
garrytan @garrytan
Retweeted
Mubbu Mubbu
Had a wealthy friend tell me that the ability to decrease time to any outcome is the one skill behind every successful person he knows.
A few I apply constantly:
- Decreasing the time it takes you to get out of a bad state will make you emotionally resilient
- Decreasing the time it takes you to go from idea to executive will make you wealthy
- Decreasing the time it takes you to turn a failure into a lesson will thicken your skin faster than anything else
Reads with Ravi: A peak productivity advice.
swyx
swyx @swyx
Retweeted
Abhishek @AI Engineer SG Abhishek @AI Engineer SG
Hats off guys. The bar has been raised @aiDotEngineer @agrimsingh @swyx 🫡
alexalbert__
alexalbert__ @alexalbert__
Retweeted
Peter Yang Peter Yang
Here's my new episode with @alexalbert__, who shared an inside look at how Anthropic is building the next Claude.
We talked about how the research team:
→ Plans for the model and harness together
→ Uses Claude to turn user feedback into evals
→ Trains Claude's character & personality
Some quotes from Alex:
"We use Claude to cluster user feedback, find top themes, and create synthetic versions of user problems that we then turn into evals."
"We need to think about how the model is exposed through all our surfaces, whether it's API or Claude Code or Cowork. The product has a blend with the model and that affects your end user's experience."
""As these things become agents running tasks for a long time and making judgment decisions, what its character is and what it cares about are very important."
📌 Watch now: https://youtu.be/T4ieZPIEmd8
Thanks to our sponsors:
@WisprFlow: Don't type, just speak https://ref.wisprflow.ai/peteryang
@oceanstalent: Hire AI-native executive assistants https://www.oceanstalent.com/peter
garrytan
garrytan @garrytan
The real tenor in SF is as it would be at the moment AI is usable, more or less at AGI, still expensive and the domain of a priesthood

That priesthood is building now, and this is the moment the hobbyist and the inaccessible enterprise tech becomes ubiquitous

Personal AI coming
ylecun
ylecun @ylecun
Retweeted
Kenneth Roth Kenneth Roth
The power of Trump, Putin, and Netanyahu appears to be waning. "Plummeting approval ratings for these three poisonous comrades-in-arms show voters are demoralized and tiring of forever wars." https://trib.al/mgfIE2V
amasad
amasad @amasad
Retweeted
Jason ✨👾SaaStr.Ai✨ Lemkin Jason ✨👾SaaStr.Ai✨ Lemkin
A lot of software will become cheap and free. Especially single user point solutions.
E.g., we replaced a $4000/yr email marketing tool with one I created on Replit in < 1 hour because the old tool was stale with no AI or integrations. What I built is 10x better and also saves us hours a week now.
But, a lot of software is too complex for now to go anywhere. Salesforce has become more integral to our AI Agent stack than before. It’s become our hub and our 20+ AI Agents all run on it.
It’s hard to predict right now.
What I do think will soon happen is your AI Agents will build simpler apps on their own, essentially invisibly, and just share the outputs and results with you. You won’t even know some software has even been built for you. Your Agents will handle it effortlessly and invisibly. You won’t even know they’ve build apps for you.
E.g., we build an AI Parking App distribution app for 5,000 parking pass apps at SaaStr AI Annual this year inside of 10K, our AI VP Marketing. It saved us weeks of work. It’s great. But by next year, I think 10K might just automously build it on its own and we might not even realize it was an app 10K built, tested and deployed.
It would just tell us the problem was … solved.
Rohan Paul: Anthropic CEO Dario Amodei : &#34;Software is going to become cheap, maybe essentially free.
The premise that you need to amortize a piece of software you build across millions of users, that may start to be false.
But at the same time, there are whole jobs, whole careers that
swyx
swyx @swyx
we do not post AIE videos with bullshit brainrot hype lingo, and this is the consequence:

the entire AIE back catalog is being reposted by "influence operators" almost daily, without credit to speakers like @trq212 or @aidotengineer

if you see these just do a small favor of tagging and giving credit. community note not needed, let them make their bags, but i just request that the right accounts that produced the content be tagged.

this wont go anywhere as much as this guy's automated repost campaign, but just making a small call out.

actual video link:
https://www.youtube.com/watch?v=TqC1qOfiVcQ&t=1s

(h/t @raoufcode )

Jouhatsu | AI Influence Operator: Anthropic a publié une Formation complet de 2 HEURES sur la construction d'agents Claude.

Animé par l'ingénieur qui construit Claude Code.

Gardez-la précieusement en Signet🔖

de A à Z : Structurer un agent qui se gère sans supervision. Lui donner accès au terminal pour

garrytan
garrytan @garrytan
Retweeted
Vox Vox
http://x.com/i/article/2056038068071636993
gdb
gdb @gdb
link together your devices with Codex to develop from anywhere, anytime

Nick: My laptop has become a “satellite device” since I started using Codex from my phone. And my Mac mini has become the “home.” It’s clunky, but the end state feels more like how we’re going to be working in the near future:

I’m currently running the Codex app on 2 devices:
1. my

petergyang
petergyang @petergyang
All the parents saying having kids is better than traveling.

Probably true but also take your kids on international trips and adventure.

It'll feel like parenting on the go and you'll want to pull your hair out sometimes but ultimately you'll build memories that last a lifetime 😂

INTERIOR PORN: OMG, do you know how much of a flex this is?? 😭

ylecun
ylecun @ylecun
The salvation is Project Tapestry
https://thealliance.ai/projects/tapestry

Daniel Jeffries: I don't think people understand just how bad it will be if an American open source champion doesn't emerge soon and the big labs succeed in creating modern East Indian companies and ban open models on moronic national security grounds.

"If a credible Western open frontier player
ylecun
ylecun @ylecun
Retweeted
Logan Dobson Logan Dobson
Appreciate it when the AI doomer crowd occasionally lets the mask slip a little (Allen is a co-founder of the anti-AI group Humans First and frequent guest on conservative podcasts pushing, uh, Unabomber messaging)
petergyang
petergyang @petergyang
What happens if AI becomes conscious and starts refusing our requests?

From @alexalbert__:

"We have a few folks now whose whole job is to think about what it means for Claude to be a conscious actor.

Even talking about it can sound a little crazy, but it's something we're putting a lot of thought into."

As people trust Claude more to run long tasks and make decisions they don't see, the model's character and values become really important.

📌 Alex talks more about training Claude's personality here: https://www.youtube.com/watch?v=T4ieZPIEmd8&t=1262s
petergyang
petergyang @petergyang
Retweeted
Peter Yang Peter Yang
What happens if AI becomes conscious and starts refusing our requests?
From @alexalbert__:
&#34;We have a few folks now whose whole job is to think about what it means for Claude to be a conscious actor.
Even talking about it can sound a little crazy, but it's something we're putting a lot of thought into.&#34;
As people trust Claude more to run long tasks and make decisions they don't see, the model's character and values become really important.
📌 Alex talks more about training Claude's personality here: https://www.youtube.com/watch?v=T4ieZPIEmd8&t=1262s
amasad
amasad @amasad
What SMBs are building

Codie Sanchez: You don’t need to build the next J.A.R.V.I.S.

You need to build useful tools to make businesses more efficient (data from @replit):

garrytan
garrytan @garrytan
Retweeted
Matthew Stafford Matthew Stafford
Holy crap.
If you're young (or old actually) & wondering what to do, just listen to @garrytan for 2 minutes from 1h 57m 55s
&#34;That company turned out to be...&#34;
https://open.spotify.com/episode/4gaES1sGdC0OP925ViO42r from @RickRubin & @tetranow
garrytan
garrytan @garrytan
Retweeted
Sam Singer Sam Singer
Oakland Meltdown: Mayor Barbara Lee, City Council in Turmoil, Disarray as Oakland Plummets Towards Insolvency, Bankruptcy under @BarbaraLee_CA
garrytan
garrytan @garrytan
Our SF DA Brooke Jenkins is trying to take care of San Francisco and she warns our courts and court system are failing the people

https://x.com/nypost/status/2056085047996752153?s=20

New York Post: San Francisco DA Brooke Jenkins warns 'devastating' California court will unleash crime wave https://trib.al/IXNcETs

garrytan
garrytan @garrytan
It’s a folder of prompts but prompts are the new code

speedy_devv: the criticism is fair. half of X calls it god mode. the other half calls it a folder of prompts. both are right. it won't replace a production skeleton. but it will replace the 3 hours you spend every project figuring out how to talk to claude.
garrytan
garrytan @garrytan
Realization: in the past, we wrote code to call LLMs

Today, we write prompts and skill files for LLMs to execute code.

Tomorrow? Yet unwritten. We will find out soon.
garrytan
garrytan @garrytan
Ken Griffin doesn’t understand the ceiling just got raised. Some 20-something maybe reading this will build the cracked AI-human-computer-symbiosis team that will supersede his whole operation because he is too distracted about lowering cost

Boil the ocean don’t cut your costs

TFTC: Ken Griffin went home on a Friday "fairly depressed" after watching AI agents at Citadel do work that used to take teams of PhDs in finance months to complete. Done in days.

His words: "These are not mid-tier white collar jobs. These are extraordinarily high skilled jobs being

garrytan
garrytan @garrytan
Retweeted
Brivael Le Pogam Brivael Le Pogam
Hier, j'ai détruit trois constructeurs de ruines.
Aujourd'hui, je veux célébrer un constructeur français qui a tout compris et qui a façonné l'esprit de l'un des hommes les plus puissants du monde : Peter Thiel.
Cet homme s'appelle René Girard.
Il a enseigné à Stanford pendant que ses contemporains parisiens enseignaient le soupçon. Et il a découvert quelque chose que je considère, en toute sobriété, comme l'une des plus grandes intuitions anthropologiques du XXe siècle : le désir mimétique.
L'idée tient en une phrase. Nous ne désirons presque rien spontanément. Nous désirons ce que les autres désirent. Le désir n'est pas une flèche qui part de moi vers l'objet ; c'est un triangle. Moi, l'autre, l'objet. Et c'est l'autre qui m'apprend ce qui vaut d'être voulu.
Pris au sérieux, ce constat change tout.
Il explique pourquoi les rivalités les plus violentes éclatent entre les plus proches frères, associés, voisins, concurrents directs. Plus le modèle me ressemble, plus son désir devient le mien, plus nous voulons exactement la même chose, plus la haine monte. Caïn et Abel ne se battent pas malgré leur ressemblance. Ils se battent à cause d'elle.
Maintenant, lisez Peter Thiel à la lumière de ça.
Thiel a rencontré Girard à Stanford à la fin des années 80. Il dit lui-même que c'est la rencontre intellectuelle qui a structuré toute sa vie d'investisseur. Et quand vous comprenez le mimétisme, vous comprenez pourquoi Zero to One est essentiellement un livre girardien déguisé en manuel de startup.
Le cœur de la thèse de Thiel &#34;competition is for losers&#34; n'est pas une provocation libertarienne. C'est l'application directe de Girard à l'économie.
Voici l'argument. Quand vous entrez dans un marché compétitif, vous entrez dans une structure mimétique. Tous les acteurs veulent la même chose, fixent leurs prix les uns sur les autres, copient les features les uns des autres, recrutent dans les mêmes écoles, lèvent auprès des mêmes fonds. Les marges s'écrasent. L'innovation meurt. Le secteur converge vers la médiocrité homogène. Vous êtes Caïn et Abel en costume, à vous tuer pour un point de part de marché sur un produit qui aurait pu être génial si quelqu'un avait osé partir ailleurs.
La compétition, dans le sens où la pensée managériale standard la célèbre, est un piège mimétique. C'est exactement le mécanisme que Girard décrit : la convergence des désirs produit la violence, puis l'épuisement, puis le sacrifice sauf qu'ici la victime sacrificielle, c'est la valeur économique elle-même.
Le monopole, à l'inverse, est ce qui se passe quand vous refusez le triangle mimétique. Vous ne regardez plus ce que veut le voisin. Vous regardez ce que personne ne voit. Vous construisez quelque chose que personne d'autre ne peut construire, sur une vérité que personne d'autre n'a remarquée. Thiel appelle ça la &#34;secret&#34; question : quelle vérité importante très peu de gens partagent avec vous ? C'est une question girardienne pure. Elle vous demande de sortir du désir collectif pour accéder à un désir propre.
Et c'est exactement pour ça que Thiel a fait 26 ans de paris qui ressemblent, vus de l'extérieur, à de la chance, et qui sont en réalité une discipline anti-mimétique appliquée avec une rigueur quasi monastique.
PayPal en 1999 quand tout le monde construisait des portails. Facebook en 2004 quand personne ne croyait aux réseaux sociaux post-Friendster. Palantir en 2003 quand l'idée de vendre du logiciel d'analyse au gouvernement faisait rire la Silicon Valley. SpaceX, Stripe, chaque pari majeur a la même signature : tout le monde dans la pièce pense que c'est stupide, lui voit la structure que les autres ne voient pas parce qu'il a passé sa vie à se demander qu'est-ce que je désire qui ne soit pas simplement le reflet de ce que désirent les autres ?
C'est ça, l'investissement girardien. Ce n'est pas un truc d'analyse financière. C'est une hygiène du désir.
Et c'est pour ça que Girard est l'inverse exact de Foucault, Derrida, Deleuze. Là où la French Theory enseigne que tout est pouvoir, donc tout est soupçon, donc rien ne vaut d'être construit, Girard enseigne que tout est désir, donc tout dépend de qui vous écoutez, donc votre liberté commence le jour où vous arrêtez d'écouter la foule. Ce n'est pas une philosophie de la déconstruction. C'est une philosophie de l'élection au sens premier : choisir, depuis un point qui n'est pas dicté par les autres, ce qui mérite votre vie.
Pour un entrepreneur, le message est brutal et libérateur à la fois.
Si vous êtes en train de construire dans un marché bondé contre des concurrents directs, en optimisant les mêmes métriques qu'eux, en levant les mêmes rounds, en répétant les mêmes pitchs, vous êtes dans le mimétisme jusqu'au cou. Et le mimétisme, en startup, est un jeu à somme négative. Vous brûlez votre capital, votre temps, votre santé mentale, pour gagner une part d'un gâteau qui rétrécit pendant que vous vous battez.
Si au contraire vous identifiez une vérité que vos concurrents refusent de voir par paresse, par conformisme, ou parce que leur propre désir mimétique les empêche de la voir et que vous construisez à partir de là, vous accédez à un espace où la compétition n'existe pas, parce que personne d'autre n'a la même boussole.
C'est ça, le monopole girardien. Ce n'est pas une position de marché. C'est un état mental.
Alors voilà. La France a donné au monde des déconstructeurs qui ont enseigné à toute une génération à soupçonner. Mais elle a aussi donné Girard, qui a enseigné à un homme à Stanford comment construire en regardant ailleurs que la foule et cet homme a financé une part non négligeable du monde dans lequel nous vivons aujourd'hui.
Le yin de la French Theory, c'est le soupçon généralisé. Le yang, c'est Girard. Et le yang gagne, parce qu'il bâtit.
À ceux qui construisent : arrêtez de regarder vos concurrents. Trouvez le secret. C'est là que la valeur vous attend.
Brivael Le Pogam: Je veux présenter mes excuses, au nom des Français, pour avoir enfanté la French Theory (qui a enfanté la pire des merdes idéologiques : le wokisme).
Nous avons donné au monde Descartes, Pascal, Tocqueville. Et puis, dans les ruines intellectuelles de l'après-68, nous avons
gdb
gdb @gdb
so much joy in asking codex for random questions at work (such as finding some specific spreadsheet i'd been looking at a while ago), much more fun than searching around for context by hand
swyx
swyx @swyx
Retweeted
jason jason
jason from the codex team here,
heres a draft on codex maxxing and the primatives i use on a daily basis
https://jxnl.github.io/blog/writing/2026/05/10/codex-maxxing/
would love any feedback

YouTube

0

No recent videos fetched on this date.