惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

G
GRAHAM CLULEY
V
V2EX
WordPress大学
WordPress大学
博客园 - Franky
Last Week in AI
Last Week in AI
博客园 - 司徒正美
有赞技术团队
有赞技术团队
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
博客园 - 【当耐特】
V
Visual Studio Blog
C
CERT Recently Published Vulnerability Notes
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
cs.CL updates on arXiv.org
cs.CL updates on arXiv.org
Jina AI
Jina AI
Attack and Defense Labs
Attack and Defense Labs
腾讯CDC
The Hacker News
The Hacker News
Hugging Face - Blog
Hugging Face - Blog
K
KPMG report finds enterprise disconnect between AI and its ROI | CIO
J
Java Code Geeks
人人都是产品经理
人人都是产品经理
阮一峰的网络日志
阮一峰的网络日志
T
Tailwind CSS Blog
S
SegmentFault 最新的问题
大猫的无限游戏
大猫的无限游戏
小众软件
小众软件
A
Arctic Wolf
量子位
博客园 - 聂微东
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
N
News and Events Feed by Topic
雷峰网
雷峰网
博客园_首页
Google Online Security Blog
Google Online Security Blog
Spread Privacy
Spread Privacy
罗磊的独立博客
H
Hacker News: Front Page
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
月光博客
月光博客
TaoSecurity Blog
TaoSecurity Blog
Exploit-DB.com RSS Feed
Exploit-DB.com RSS Feed
博客园 - 三生石上(FineUI控件)
宝玉的分享
宝玉的分享
IT之家
IT之家
The Cloudflare Blog
爱范儿
爱范儿
博客园 - 叶小钗
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
Apple Machine Learning Research
Apple Machine Learning Research
酷 壳 – CoolShell
酷 壳 – CoolShell

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor GitHub - GenAI-Gurus/awesome-eu-ai-act: Curated tools, official sources, OSS, templates, and guides for EU AI Act compliance. Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders How to Switch AI Chatbots and Why You Might Want To GitHub - MattMessinger1/agentic_refund_guardrail: Safe refund policy layer for AI agents — Python + TypeScript. Same behavior, shared tests. Adam/papers/emergent_values_whitepaper.md at master · strangeadvancedmarketing/Adam Ask HN: How do you stop playing 20 questions with your AI coding tools How far can automation and AI support psychotherapy? - @theU GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits A Mac Studio for Local AI — 6 Months Later A History of the Early Years of AI at the University of Edinburgh Why AI Coding Tools Still Feel Stuck on Localhost MSN AI Datacenters Are Becoming Strategic Targets twitter.com Penn Researchers Use AI to Surface Unreported GLP-1 Side Effects in Reddit Posts Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 AI models are terrible at betting on soccer—especially xAI Grok GitHub - xialeistudio/echoic GitHub - HimashaHerath/github-dev-wrapped: AI-powered weekly GitHub activity reports deployed to GitHub Pages GitHub - alejandrobalderas/claude-code-from-source: Architecture, patterns & internals of Anthropic's AI coding agent — reverse-engineered from source maps AI and Tech brief: Ireland ascendant GitHub - Titovilal/context0: Context0 - Never Surrender Training for a Marathon with an AI Coach: What Worked and What Didn't Cyber Pulse: Agentic Intel - Apps on Google Play I Built an AI PR Reviewer That Catches Bugs by Not Looking for Bugs Gen Z workers are so fearful AI will take their job they’re intentionally sabotaging their company’s AI rollout | Fortune How AI Is Reimagining the Game of Golf–For Both Players and Courses GitHub - nattergabriel/reseed: A CLI tool for managing and distributing agent skills across projects Is SVG the final frontier? My AI workflow evolved from prompts to a near-autonomous workflow MLSharp Help - 3DGS Viewer & Generator I put my cognitive field based AI's runtime on GitHub Is Numble the first AI-proof game? A3: Kubernetes for autonomous AI agent fleets | Emergent Principles Deepali Vyas ("The Elite Recruiter") GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Unionized ProPublica staff are on strike over AI, layoffs, and wages Unleashing the Advantage of Quantum AI We're heading for an AI-fueled 'dementia crisis,' brain scientist warns The AI-Assisted Breach of Mexico's Government Infrastructure [pdf] GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. MSN GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness We gave an AI a 3 year retail lease in SF and asked it to make a profit | Andon Labs AI Code is Hollowing Out Open Source, and Maintainers are Looking the Other Way What leaked "SteamGPT" files could mean for the PC gaming platform's use of AI AI is the boss at this retail store. What could go wrong? GitHub - Wuzu11517/agentic-proxy: Local proxy meant to help reduce With Drones, Geophysics and ArtificiaI Intelligence, Researchers Prepare to Do Battle Against Land Mines A Single Operator, Two AI Platforms, Nine Government Agencies: The Full Technical Report 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - inevolin/resume-cli: Hit Claude usage limits? Resume any AI coding session elsewhere. Switch tools at zero friction. GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. How to Build a Secure AI PR Reviewer with Claude, GitHub Actions, and JavaScript This Startup Wants You to Pay Up to Talk With AI Versions of Human Experts Intel Arc Pro B70 Brings 32GB VRAM to Local AI for $949 WordPress 7.0: The Good, the AI, and the Still Missing AI on the couch: Anthropic gives Claude 20 hours of psychiatry IatroBench: Pre-Registered Evidence of Iatrogenic Harm from AI Safety Measures AI Agents Know About Supabase. They Don't Always Use It Right. The history and future of AI at Google, with Sundar Pichai Inside an AI‑enabled device code phishing campaign How Meta Used AI to Map Tribal Knowledge in Large-Scale Data Pipelines AI for Systems: Using LLMs to Optimize Database Query Execution Forecasting the Economic Effects of AI Introducing Tinker: Play with AI, bring your ideas to life AI sheds light on an ancient gaming mystery People really hate AI but not as much as Iran—or Democrats | Fortune What is an AI Product Engineer? Phoebe Gates wants her $185 million AI startup to succeed with 'no ties to my privilege or my last name': 'I have a chip on my shoulder' | Fortune
Margin Points - Arnold Engel
by Arnold Engel · 2026-05-25 · via Hacker News - Newest: "AI"

OpenAI & YC, IPO oxygen, hacky sack

Lemons, slop mergers, return of the sack

May 20, 2026 · [Essays 49, 50, 51]

OpenAI & YC

Sam Altman announced last night at a YC gathering that OpenAI is offering all current YC companies a $2M investment offer in the form of OpenAI tokens. The $2M will convert in each startup’s next round of financing and is done in the form of a SAFE.

There is some very clever handling of adverse selection in here.

The tokens here can be spent quickly (by Demo Day) or slowly. Startups that are burning tokens quickly will prefer this deal and that is the type of company that OpenAI is most interested in backing and seeing grow quickly. Startups in the batch will explore different applications and approaches and the successful ones will spread what works. There doesn’t appear to be a timeline on accepting or rejecting the offer during the batch, so as early token monsters eat tokens and find success there can be new converts.

While there is certainly pivoting during the batch, for many they’ll already have shovel-ready places to burn these tokens. For others who are clear on their business model and on the fringes of product-market fit, it’s an opportunity to reflect on where this new input could help change the shape of what they can do for customers. Boom—OpenAI also gets free mindshare, even from those that don’t take the deal.

This offer is made easier for OpenAI because they have high gross margins on the API tokens they are using in this deal. If they are at a 70% gross margin, a $2M retail headline figure for value represents closer to $600K of actual cash costs to OpenAI.

There is an opportunity for startups to arbitrage the tokens to other companies and capture the cash and book the revenue today. That pushes out the timeline on which they need to raise their next round and see OpenAI convert onto their cap table.

While that might appear to be an issue for OpenAI—the startup is just brokering the token—that misses some of the value. The startup would be in a great position to build a thicker wrapper around the experience for the customers it was brokering the tokens to and would be developing or deepening customer relationships. It’s not without positive residual exhaust from the effort—this is a far cry from drop shipping things on Amazon.

Demo Day and the incentives for the startup that exist outside of this deal would push to maximize the value, rather than just dumb-piping the tokens to others in a modern revenue financing scheme. YC having screened these companies and working with them during regular office hours cuts a lot of the risk for OpenAI.

The offer will also cleave a line in the sand for the batch. The maxxers and the minners? Everyone will be judged at Demo Day with this X factor: “what did you do with your tokens?” or “oh, interesting you didn’t take them?”

Demo Day also provides a nice opportunity for OpenAI to continue distribution momentum. Everyone can see what was built and through all the launches on X as well, each providing a showcase for the ingenuity of how these resources were used. If these early signs are positive, OpenAI can do something like this with the next batch—it has bought the option with being the first mover on this to keep doing it.

Announcing it on a visit and casual interview on stage with a YC partner makes it seem casual and not deeply considered, which actually makes it much more powerful.

Of course, a different type of offer like this has been popular for a long time. Free credits to get startups started on Stripe, Zendesk,1 AWS, Google Cloud, etc. have been a mainstay. a16z speedrun offers $7M worth of free credits to startups in the program. The trade is conceptually different in subtle but important ways. Free credits to get you started is a marketing concept that is proven. It’s free and you either want to use the product or you don’t. The scope of consideration is much narrower.

“I’m going to give you tokens and take equity on the bet that you’ll unlock disproportionate value from them” is something different to consider. In that consideration, it’s the startups who take the deal that are the most bullish on turning the tokens into company value. It’s aspirational, it’s forward-looking and it’s not something everyone can provide. “We’ll help change your valuation” is a different frame from “we’ll help save you money.”

It may open the door for more creative approaches and trades. Things like a studio-type fund, who takes equity and provides services for free (like an awesome launch video or branding) and pays its own employees with carry and cash raised from LPs. Other vendors might get creative with their offers.

There’s a lending-to-customers concept at play here, although with an equity component. That usually comes with some adverse selection risk, but here it’s washed out by a much bigger potential upside for both sides.


IPO oxygen

SpaceX’s IPO anticipation is taking all the oxygen out of the room.

From CNBC:

“Nobody wants to be caught in the SpaceX blast radius,” said Renos Savvides, head of equity capital markets at Neuberger Berman. “If you’re a smaller IPO and you’re on the road the same time as SpaceX, no one is going to pay any attention to your deal.”

Wait your turn, OpenAI & Anthropic. And if you aren’t part of the big three for the next year, well, your luck might be in the care of the great IPO sorting hat.

Obviously, there are ways around this if you are an aspiring public company. You can grow huge on your own and IPO or merge with a public company or merge with other private companies and IPO together.

Merging seems very underrated at this moment.

SpaceX is IPOing into being one of the top 10 largest public companies in the world, which is astounding. Saudi Aramco had a mega $25B IPO fundraise in 2019 and is also in the top 10, but that was a well-established business in a well-established market.2 Nothing else comes close to SpaceX.

SpaceX itself is the product of the Musk bundle. Starlink, which accounts for the majority of its revenue today, and xAI and X are all part of it. The Musk bundle kicked the door open for others to also combine.

The first mega merger company will get attention, and disproportionately. A first-merger-advantage, if you will.

Modern merger theory was built on efficiencies and cost savings first. Then ‘synergies’ that would unlock greater revenue growth. Then ‘moat’ or defensibilities to make it harder to compete with and allow for higher prices or lower costs, translating to more profit.

But now, another thing? Attention. Specifically attention from investors. Big, moving targets draw attention. The new bundle is not for efficiency but for hyper-relevance and attention. Attention breathes oxygen into the room for companies.

AI-for-management-information cuts the downsides of traditional bloated conglomerates. Sure, there are some efficiency gains too, but that’s not the driver here. If your revenue is being valued at a high multiple, anything you can add to it will likely be valued more highly in your orbit than on its own.

Big public companies also don’t have much of the downside of smaller ones. Information and financials can be categorized into bigger buckets that obscure margins for specific product lines. AI-native businesses are also easier to integrate with and run through an M&A process.

It’s not that hard to find tie-ups that are beneficial in terms of market, inputs, or customers that can be packaged. We’re quite a ways away from seeing our first slop-merger, purely executed on the say so of a model. Executives have more time to think deeply and pursue courtships.

Focus? Well, each offering of sufficient size today already has the ‘mega-menu’ on their website listing every possible solution to anything related to their area and any possible potential customer that could possibly walk through the door. The era of focus is now one of vague sense: Stripe is money stuff, Salesforce is sales stuff, Adobe is design stuff, Uber is convenience stuff. Details get filled in after the mental shortlist for a buyer has been set.

Combinations can happen now to leapfrog and find oxygen in the public markets. At a different point in the cycle, they’ll be about finding a port in the storm.


Hacky sacks are back

Intrepid reporting from the WSJ tells us that hacky sacks are officially back in:

The little bean bags that once appeared relegated to the same cultural graveyard as mixtapes and Blockbuster have taken over high-school sports fields, hallways and sometimes classrooms seemingly overnight.

Cue nostalgia. Cue Backstreet’s back?

This story goes beyond nostalgia.

There is no brand associated with this, and that makes it feel more genuine. There is a Hacky Sack brand, but the brand name has become generic. Technically, it’s a foot bag—but that seems too odd a concept, like some sort of plastic wrapper you’d put on a walking boot to cover it in the shower.

Products from a slew of different makers are easily available on Amazon and Etsy and everywhere else hacky sacks are sold. No one is really pushing a brand in a major way, which is almost quaint. Retailers are happy to sell here and drive local awareness, but there probably isn’t gold sitting in the middle of hacky sacks.3

Contrast this no-brand scenario with others. Hoverboards were hot for a moment as well but tapered off quickly when some started catching fire. Without known brands behind them, the safety risks and financial risks of buying a dud started to outweigh the cool factor. Had a known electronics brand stepped in earlier or an upstart leaned heavily into establishing a trusted brand, the story may have turned out differently.

We are somewhere on that journey now with peptides. Brands for distribution and manufacturing are starting to coalesce in the peptide space. They will likely drive for differentiating if at all possible to protect their own investments in the space and prevent unbranded generics from cutting corners that could lead to customers being turned off from the category.

Both peptides and hoverboards have higher costs and real safety issues. Hacky sacks are gleefully cheap, and any safety issues are entirely self-inflicted. They are able to stay in a brand-less state indefinitely without damaging the category or impeding the growth prospects.

This isn’t even the product of nostalgia marketing, like Coinbase’s Backstreet Boys Super Bowl ad or Tin Can’s appeal to millennial and Gen-X parents. It’s driven by the kids, not the parents.

In all of it, a generation is getting exposure to some business thinking: it’s nice to have both a single-player mode and a potential group option when you play with others on the playground. Come for the tool, stay for the network.


Like these essays? I write a trio of essays on wide ranging topics delivered four times a week. Subscribe and forward it to your friends. Have feedback, thoughts to share or comments? Comment or reply.