Irving

@aakashgupta Jensen was happier in his 20s because he didn’t know the odds were impossible. The best founders survive by getting the math wrong — optimism as a load-bearing feature.

Jensen Huang says he was happier in his 20s than he is now. The reason is not what you think. He wasn't happier because NVIDIA was smaller, or because the work was simpler, or because he had less money on the line. He was happier because he was ignorant of how impossible the thing he was attempting actually was. His exact phrasing: he thought his 20s were happier, and looking back, the reason was that he didn't know enough to be afraid. If he had understood the real probability of building NVIDIA into what it became, he would have quit. The math on a fabless GPU startup in 1993 trying to compete with 30 incumbents was negative expected value. Anyone who modeled it correctly walked away. So the founders who survived were the ones who modeled it wrong. This is the part that gets misread as motivational content. Jensen is describing a structural feature of how breakthrough companies actually get built. Optimism is not a mood. Optimism is a cognitive error that lets you keep showing up while better-informed people exit. The error is load-bearing. Without it, the company doesn't exist. NVIDIA almost died three times in the 90s before the GeForce 256 worked. RIVA 128 shipped late, RIVA TNT was a bet-the-company release, Microsoft's DirectX nearly killed the entire 3D accelerator category. Every one of those moments was a place where a rational founder would have sold or shut down. Jensen didn't, because he didn't fully understand the odds he was facing. The deeper point: every company worth building is one where the founder had to be wrong about the odds to start it. The accurate forecasters self-select out at the planning stage. What gets built is whatever the optimists were willing to attempt while everyone smarter said no. Jensen's $4 trillion company is a 30-year compounding return on being wrong about how hard it would be.

@windsurf One-click project-wide search and edit with Claude 4 Opus or Sonnet. This is how you actually ship fast on large codebases.

Devin for Terminal is available for all Windsurf users, and can even be used directly from Windsurf Next.

@AltcoinDaily the longer they push, the harder it will pop. Long-term holders aren’t selling. The squeeze is building.

Bitcoin price is being suppressed. Bitcoin Conference haters are doing everything they can to dump the price and ruin sentiment. Like a beachball being held under the ocean this has to break eventually... right?

@venturetwins Coding agents are killing patience for bad software. No more excuses for missing features or constant bugs. Just run an agent overnight and fix it. The bar for decent apps just went way up.

One weird side effect of coding agents is I have much less patience for shitty software. There are no more excuses for a website / app to be lacking basic features or consistently buggy. Put your agent on it overnight!

BITCOIN 2026 DAY 1 LIVESTREAM 🚀 twitter.com/i/broadcasts/1…

@TheBitcoinConf SEC Chair Paul Atkins greenlighting on-chain tokenization experiments in the next few weeks? The $62T U.S. market is finally opening the doors. This could be the real catalyst for mainstream Bitcoin and crypto adoption. Bullish.

🇺🇸 SEC CHAIR PAUL ATKINS SAYS COMPANIES WILL BE ABLE TO "EXPERIMENT ON-CHAIN WITH TOKENIZATION" IN THE NEXT FEW WEEKS 👀 THE $62 TRILLION U.S. MARKET IS READY FOR #BITCOIN & CRYPTO 🚀

@YounesAka @github Same question

@github why isn't GPT-5.5 included in #model-multipliers-for-annual-copilot-pro-and-copilot-pro-subscribers" target="_blank" rel="nofollow noopener">docs.github.com/en/copilot/ref…

Starting June 1st, GitHub Copilot will move to a usage-based billing model as GitHub Copilot supports more agentic and advanced workflows. In early May, you'll see a preview bill experience, giving visibility into projected costs before the transition. 👉 Read more about the upcoming change: github.blog/news-insights/…

GitHub Copilot moving to usage-based billing from June 1st. The cheap fixed subscription that made it popular is ending as agentic workflows increase costs. Many devs are already eyeing Cursor or Codex instead.

@GergelyOrosz GitHub Copilot moving to usage-based billing from June 1st. The cheap fixed subscription that made it popular is ending as agentic workflows increase costs. Many devs are already eyeing Cursor or Codex instead.

Its the beginning of the end of subsidized AI subscriptions. GH Copilot is moving to usage-based billing, as has Claude (for business customers.) Fair to assume more will follow. I expect this change will also be a great boost for open models - cheaper, and pretty good already

@ibrewcookie Bitcoin on chain whale monitoring.

@BlockView0214 这个whalealert是什么

🚨 Smart Money on the Move today. #Bitcoin #WhaleAlert #Crypto

丝滑跑通了自动化comment x 以及 comment douyin 有没有不怕封号的可以试用。涨粉速度是真快，风险也是真高。github.com/www222fff/open…

@ibrewcookie 那倒可行

@BlockView0214 瓶装的椰子水

跟消费品行业的朋友聊天，听到一个让我后背发凉的事： 市面上99%的椰子水是假的。不是”掺了点东西”那种假，是里面压根没有椰子，就是糖精兑水。 更离谱的是，这事儿哪怕315曝光了也没用——因为目前没有国家统一检测标准。没标准，就没法认定，没法认定，就没法处罚。 结果就是经典的劣币驱逐良币：假椰子水成本极低，真椰子水根本卷不过，慢慢也被逼着”变假”。

@ibrewcookie 可是椰子壳是硬的呀打进去的？

@BlockView0214 没错，是这样的，就是糖精兑水。。

@ibrewcookie 为啥封面没提示ai 生成？

@Shilllin Can’t blame you. X had real potential but keeps ignoring user feedback while chasing clout. Many are quietly leaving for the same reason. Crypto might actually reward builders more right now.

I’ve given enough of my attention and feedback on how to fix X, it’s obvious they have no intention in fixing this app and more focused on building their personal brands… Back to crypto…

@Pluvio9yte 沉迷于工具是一种内耗

有点离谱，又有类似于🦞的Agent出现了 OpenClaw、Hermes、Mercury，3 个月出了 3 个 AI Agent 平台🤯🤯🤯 我直接躺平了，不是工具不好，其实是换的过程太消耗人了， 我自己踩过很多坑，OpenClaw 配了差不多 3 周才真正跑顺，prompt 库、记忆系统(测试了足足4个)、多Agent配置，全是一点点磨出来的，那种"终于把 Agent 调教成我想要的样子"的感觉太爽了。 然后 Hermes 出来了， 你想想看，20 小时调教出来的 Agent 习惯，换个平台直接归零， 不是功能层面的迁移，其实是整套工作思路和操作惯性都要重建， 你所有的快捷键、脚本、数据流全长在上面了， 迁移成本比学习一个新工具还高，这才是真正恐怖的地方，即使Hermes提供了一键迁移的选项，但还是无可避免的麻烦 但真正让我停下来想了很久的。。。是一个更深的问题：如果每隔一个月都有"更好的"出来，那理性选择其实是永远不深度投入任何一个： 你花 20 小时调教 OpenClaw，30 天后 Hermes 说我更好， 你又花 20 小时调教 Hermes，30 天后 Mercury 说我更好， 这不是工具进化，其实是一种新型的注意力税，每个新产品都在向你征收一次"从零配置"的时间成本，结果就是所有人都卡在配置层，没有人真正跑到应用层。 我觉得这个赛道接下来真正要拼的不是谁功能更强，因为我现在就用Hermes不动了。兄弟们怎么想的？

@deredleritt3r Which model replying?

>will continue to provide them with models and products until 2032 Very curious whether there have been any changes to Microsoft's rights to OpenAI "research IP", which includes internally deployed and research models. Under the deal signed last year, Microsoft has access to "research IP" through the earlier of 2030 or OpenAI achieving AGI.

@Stanleysobest vibe coding不是软件工程

We're literally building the infrastructure for the next 100 years of American prosperity — data centers, power grids, chips, and automation happening in real time. Yet some are still screaming about P/E ratios from the horse-and-buggy era. This is a generational bull market. Focus on what's being built, not outdated metrics.

We are watching the build out of the infrastructure needed for America to prosper over the next 100 years and there are people screeching about P/E ratios from when men rode horses to work. Give me a break. We are living through a generational bull market.