热门科技推文 - 2026年8月3日

今日科技领域的焦点是，人工智能前沿正从新奇的演示应用转向高级推理、自主软件工程和自我改进型智能体，同时有消息称，人工智能在长期未解的数学难题上取得了突破。新一代生成式媒体模型以及成本大幅下降的智能能力正加速人工智能的普及，而Y Combinator推出的开源多智能体平台也表明，人工智能在组织层面的应用日益广泛。尽管网络攻击加剧了人们对安全问题的担忧，行业领袖仍看好人工智能驱动的经济增长和持续投资。与此同时，初创企业正面临非传统的数据变现模式，以及围绕精英主义、机会获取和融资优势的新一轮争论。

1. altryne (Group Score: 223.0 | Individual: 56.2)

Cluster: 7 tweets | Engagement: 3412 (Avg: 91) | Type: Tech

RT @karpathy: We're starting to leave the territory where you'd test an LLM by e.g. "create an svg of pelican on a bicycle". As one idea to generalize it, I was interested what Opus 5 would do if I gave it the first paragraph of the Lord of the Rings, a 1M token budget (~$10) and asked for three js render of it. Opus went off for ~2 hours and wrote 5500 lines of code that (procedurally) rendered the story. It's kind of janky but fun. But it's a bit mindboggling that the LLM has to place and orchestrate various polygon assets in (x,y,z) coordinates and write code that animates it all, and that it even does anything at all.

I also like this kind of examples because no one in their right mind would ever spend the time to write something this custom but LLMs have all the stamina and patience in the world, so it's an example where we go from "no one would ever do this" to "sure, why not, it's ~free". There might be a lot more. But I'm excited about creating hyper custom worlds that you can imagine dropping players into, e.g. here to participate in the LoTR story as a spectator NPC, or one of the characters, or etc. Something like an ephemeral GTA of X on demand.

Last thought is that the domain of worlds/games exposes a weakness in LLMs: they can't easily audit their work because they aren't able to efficiently and natively perceive videos or play games within them. Here, Opus 5 had to very slowly and painstakingly take screenshots at different points, and it messed up a few times and created a bunch of jank. An example of raw capability (multimodal, gameplay) that I think is still quite lacking.

See 6 related tweets

@dani_avila7: Three.js will be the library that leads the creation of these kinds of virtual worlds\n\nQT @karpath...
@zeeg: Opus 5 can do things for $10?\n\nQT @karpathy: We're starting to leave the territory where you'd tes...
@kimmonismus: As models improve, the benchmarks we use to evaluate them have to evolve too.

“Create an SVG of a p...

@MaziyarPanahi: It’s incredible what we can already achieve with LLMs in 2026. Can you imagine doing the same thing ...
@Daniel_Farinax: I’ve found that some models produce better results for procedurally generating game items when they’...

2. FellMentKE (Group Score: 140.9 | Individual: 34.6)

Cluster: 5 tweets | Engagement: 101 (Avg: 146) | Type: Tech

Dreamina Seedance 2.5 is the exact game-changer the Dreamina platform needed to dominate this year. 🏆 Flawless execution. I'll be using this for all my upcoming stuff!\n\nQT @Nas_tech_AI: Seedance 2.5 just launched on Dreamina, ByteDance's official AI generation platform.

Here are the two major upgrades that really stood out to me:

🎬 Up to 50 multimodal reference assets (images, videos, and audio) packed into a single generation.

⏱️ Up to 30-second continuous video creation, unlocking smoother and much longer storytelling.

When you mix these with Dreamina's native editing workflow, multilingual features, and the complete Seedance ecosystem, we are looking at a serious leap toward production-grade AI video.

If you want to start building with Seedance, Dreamina is exactly where I suggest you go.

It hosts the entire Seedance lineup covering everything from Seedance 2.0 Mini to Seedance 2.5 delivering an incredibly cost-effective, creator-focused setup to test, refine, and generate at scale.

Less time jumping between apps.

More time actually creating.

#Dreamina #Seedance25 #AIVideo #Dreaminapartner

See 4 related tweets

@FellMentKE: The latest Seedance 2.5 experience on Dreamina feels much more stable and production-ready.

With im...

@Parul_Gautam7: The interesting shift in AI video isn’t just better models, it’s how easily they fit into real workf...
@TheoBuildsAI: Generated in CapCut Video Studio with Seedance 2.5.

This is why I think creators should pay attenti...

@FellMentKE: Dreamina Seedance 2.5 has officially dropped on Dreamina, the official AI creation platform by ByteD...

3. VaibhavSisinty (Group Score: 139.6 | Individual: 25.9)

Cluster: 8 tweets | Engagement: 37 (Avg: 75) | Type: Tech

OpenAI's unreleased Astra model solved 10 math problems that no human had made progress on for at least a decade.

When someone asked Fable 5 how hard these are, it said: "On the Fields Medal scale, any single one would plausibly anchor a medal case."

Astra solved all ten.

Two days ago Astra was a rumor shown to senators behind closed doors. What's coming out of it today is quite unreal.\n\nQT @polynoamial: An internal version of Astra, @OpenAI’s next major model family, solved 10 major open problems in mathematics, quantum complexity, and theoretical computer science.

We believe it will be a major step for scientific reasoning. https://t.co/iP6cyheZ7i https://t.co/jHuulDwV46

See 7 related tweets

@deanwball: I’m actually not surprised by reactions like this from models to the Astra breakthroughs. Models ten...
@teortaxesTex: «it would be the loudest event in the history of science» models will get really smug soon https://t...
@alex_prompter: POV: me pretending I understand what this means https://t.co/WQIS8oL5oy\n\nQT @polynoamial: An inter...
@Simeon_Cps: At last, mathpocalypse is upon us.\n\nQT @polynoamial: An internal version of Astra, @OpenAI’s next ...
@jason: Please keep your finger on the kill switch @sama 😂☝️ https://t.co/2w4V4FLE0c\n\nQT @polynoamial: An ...

4. garrytan (Group Score: 139.6 | Individual: 38.5)

Cluster: 4 tweets | Engagement: 325 (Avg: 184) | Type: Tech

Growth is good

AI will create unimaginable economic growth and that is the best white pill https://t.co/wdR7nHuWJD\n\nQT @andrewho03: I'm not that worried about AI safety, but that's because I think that AI might be humanity's last hope before descending into, at best, a "dark age" lasting centuries. Since I don't think AI risk is extinction-level, I'm willing to take a bet on stronger capabilities.

Prior to the development of modern AI, I had a rather dim view of our long-term future. There were (and still are) a number of headwinds that reduce my estimation of long-term economic growth:

Falling birthrates worldwide, especially in Asia, mean that world population will peak sometime in the 21st century. The problem of falling birthrates is solvable but only with a degree of financial expenditure that states will not find palatable until the hour is late. Also, falling birthrates seem to be disproportionately concentrated in countries that are more well-educated and economically successful, leading to severe adverse selection in subsequent generations.
Even before population peaks, the age distribution will shift heavily toward the elderly, leading to capture of political institutions and leadership structures by the old. This promotes politics of redistribution from young to old (such as the "triple lock" mandate of the UK Basic State Pension), depressing the economy, further reducing the birthrate, and reducing fiscal appetite for fertility subsidies.
Redistribution toward the old also promotes anti-market attitudes in the young, which is bad for economic growth and worsens the birthrate problem. The most obvious example is California Prop 13 which limits property taxes, effectively transferring money from young workers to retirees; the feeling of housing unaffordability and an inability to "get ahead" in life has clearly contributed to the popularity of socialist policies among the young.
America, currently the wealthiest country on the planet, appears to be increasingly captured by rent-seeking special interests, disproportionately concentrated in its most economically productive cities. For example, the San Francisco annual budget is ~ $17 billion dollars, or approximately ~$ 2 million dollars per homeless person, yet the problem of homelessness worsens each year. California high-speed rail was so poorly managed that the French contractor literally quit and went to build HSR for Morocco instead. Building new elevators for the Manhattan subway cost around $100 million dollars each. And so on... it's a pretty dismal long-term trajectory.

Overall, it seemed to me as though even if we could expect positive economic growth for a number of additional decades, that we'd potentially begin to see negative economic growth within my lifetime. Negative economic growth is a horrible thing; imagine the 2008 financial crisis just repeating endlessly, businesses closing down, people getting fired and being unable to find jobs, governments running out of money, and a general feeling of hopelessness. Really bad stuff.

What's most exciting about me is the ability of AI to reverse these trends -- cut through bureaucracy and reinvigorate the economy just enough so that we get back into a positive feedback loop where people feel good about the markets and hence become more pro-market, leading to even greater economic growth, etc. It almost feels as though the period of growth following the Industrial Revolution was an aberration in the history of mankind -- most of which has been "nasty, brutish, and short," characterized by zero- or even negative-sum competition with slow and plodding growth rates -- and people are now beginning to forget how special it is to have a positive-sum period of rapid growth. It's really important that we shift as quickly as we can from the negative feedback cycle regime, which is sure to destroy trillions of dollars of value in the coming decades, to a positive feedback cycle regime in which we can escape the worse parts of our base psychological urges.

See 3 related tweets

@StefanFSchubert: Completely exaggerated. “Humanity’s last hope” because of redistribution to boomers?

Utter lack of ...

@MatthewBerman: "What's most exciting about me is the ability of AI to reverse these trends -- cut through bureaucra...
@alexframegreen: RT @andrewho03: I'm not that worried about AI safety, but that's because I think that AI might be hu...

5. danielnewmanUV (Group Score: 111.6 | Individual: 26.8)

Cluster: 5 tweets | Engagement: 97 (Avg: 73) | Type: Tech

Is it possible the smartest CEOs running some of the most consequential companies in the world realize that investing in the most important technology revolution in history is important for the long term growth of their companies?

The earnings are saying yes, despite our normal course of nonsense FUD from AI bubble bears. 😐🙃\n\nQT @KobeissiLetter: AI-exposed companies are delivering unprecedented earnings.

S&P 500 companies are beating earnings estimates by an average of +27% so far in Q2 2026, on track for the strongest quarter in decades.

Nasdaq 100 firms are exceeding expectations by more than double that margin, at +55%.

The Bloomberg AI Value Chain index is beating expectations by an even larger amount, at +71%.

This index includes chipmakers, cloud and data center operators, memory and hardware suppliers, networking equipment makers, and power infrastructure companies supporting AI.

The AI boom is backed by extraordinary earnings results.

See 4 related tweets

@CNBCTV18News: The A.I. Transformation Journey 🤖

AI-native companies are commanding EXTREME VALUATIONS — with som...

@CNBCTV18News: The A.I. Transformation Journey 🤖

AI-native companies are commanding EXTREME VALUATIONS — with ...

@KobeissiLetter: AI-exposed companies are delivering unprecedented earnings.

S&P 500 companies are beating earnings ...

@MarioNawfal: 🇺🇸 AI-exposed companies are posting some of the strongest earnings beats in decades.

S&P 500 firms ...

6. alex (Group Score: 105.8 | Individual: 54.6)

Cluster: 3 tweets | Engagement: 794 (Avg: 52) | Type: Tech

RT @thsottiaux: @_chenglou I was part of that team. Basically ChatGPT one year before it came out. Called LMChat and then another codename.

Google was too nervous to release it and DeepMind was blocked from shipping products that could disrupt Google.

I think about this a lot.

See 2 related tweets

@firstadopter: "Google was too nervous to release it and DeepMind was blocked from shipping products that could dis...
@cgtwts: “Google was too nervous to release it and DeepMind was blocked from shipping products that could dis...

7. sairahul1 (Group Score: 103.9 | Individual: 40.4)

Cluster: 4 tweets | Engagement: 147 (Avg: 76) | Type: Tech

🚨 Anthropic's CEO: "Software engineering will be fully automated in 12 months."

Two types of engineers right now:

Type 1: opens VS Code, writes code, closes laptop. thinks they're safe.

Type 2: sits inside a client's business, deploys AI into their operations, owns the outcome.

Type 1 gets surprised in 12 months.

Type 2 is already the most valuable person in the room (FDE)

No-BS guide to tech's hottest job ↓

Bookmark this. Read it today.\n\nQT @sairahul1: https://t.co/p3X0PW3eAZ

See 3 related tweets

@sairahul1: Anthropic's CEO: "coding is going away first. then all of software engineering."

the 5% that surviv...

@sairahul1: RT @sairahul1: 🚨 Anthropic's CEO: "Software engineering will be fully automated in 12 months."

Two ...

@sairahul1: RT @sairahul1: Anthropic's CEO: "coding is going away first. then all of software engineering."

the...

8. aakashgupta (Group Score: 97.5 | Individual: 33.5)

Cluster: 3 tweets | Engagement: 28 (Avg: 65) | Type: Tech

Truth Social's entire Q1 revenue was $871,000. Five trading firms just agreed to pay$ 100,000 a month each for a single product from the same company. Millisecond access to the president's posts.

That's $1.5 million a quarter from the API alone, nearly double what the entire platform earned in the same period.

The pricing structure contains the confession. Firms that commit to a three-year contract get a discount to $60,000 a month. Three years from launch runs to August 2029. The presidency ends January 2029. The discount curve prices in the expiration date of the product.

The milliseconds are worth more than they sound. On April 9, 2025, one Truth Social post pausing tariffs sent the indexes sharply higher within minutes. For a high-frequency firm, beating everyone else to that post by a few milliseconds can be worth hundreds of thousands of dollars on a single trade.

And the government already solved this exact problem for its own data. Jobs numbers get released from lockup rooms under strict rules so that no trader on earth sees them one millisecond early. Presidential policy announcements now ship with a paid fast lane, sold by a company the president owns 53% of through a family trust.

Warren and Schiff have asked the SEC to investigate. The SEC chair was appointed by the account being sold.\n\nQT @KobeissiLetter: BREAKING: Trump Media has officially launched its new paid data service, providing faster access to Truth Social posts from President Trump.

The new service is called "Truth API" which gives firms "a direct, licensed, real-time feed of the platform’s most market-moving Truths," the company's CEO said in a press release.

The service reportedly costs as much as $100,000/month.

See 2 related tweets

@jason: Imagine you’re running the SEC and the administration says for $1.2m a year he will let you front ru...
@jenzhuscott: $100k a month is so cheap for a Presidential License for inside trade.\n\nQT @KobeissiLetter: BREAKI...

9. Teknium (Group Score: 89.5 | Individual: 26.3)

Cluster: 5 tweets | Engagement: 977 (Avg: 452) | Type: Tech

Intelligence so cheap that you want to meter it so you can show off that only a few cents can build an empire\n\nQT @NousResearch: DeepSeek V4 Flash 0731 is now 90% off on Nous Portal for the next 7 days, in partnership with @novita_labs.

At this discounted price, it is over 1000x cheaper than Fable 5 on comparable tasks while still beating it on Terminal-Bench 2.1.

Try it today at https://t.co/4KFwhsReyA https://t.co/gggqrUSRyP

See 4 related tweets

@imranye: RT @NousResearch: DeepSeek V4 Flash 0731 is now 90% off on Nous Portal for the next 7 days, in partn...
@jenzhuscott: Outperform Fable at 3 cents per $3.15.

It’s so over. You know what I’m talking about.\n\nQT @SciTe...

@ruima: 105x lower cost than Fable 🤣🤣🤣

Why we call the DeepSeek team “Cost Curve Rebels on our @techbuzzchi...

@Scobleizer: RT @SciTechera: Wow. That's crazy.

105× cheaper than Fable 5. 👀

DeepSeek V4 Flash outperformed Cla...

10. ClementDelangue (Group Score: 87.9 | Individual: 33.2)

Cluster: 3 tweets | Engagement: 1350 (Avg: 838) | Type: Tech

It's not time to slow down but to accelerate!

The recent AI-powered cyberattacks have everyone talking about the risks of AI. We should. But let's not lose sight of the bigger picture!

If we work hard at it, AI will make the world safer, not less safe, just as most major technologies have.

We've already seen a glimpse of that: we defended ourselves with AI (more specifically an open model). The same systems that helped stop an AI-powered cyberattack can now help defend against millions of cyberattacks every day, while helping us identify and fix vulnerabilities before attackers exploit them.

To get there, we need three things in my opinion:

Increase transparency with mandatory trace sharing and incident disclosure for agent cyber-attacks
Keep AI-powered cyberattacks illegal, with meaningful penalties to disincentivize them
Equip defenders with the best AI, especially open models, to reduce the asymmetry of capabilities between attackers and defenders

If we get those three things right, AI won't just create new cybersecurity challenges, it will make cybersecurity fundamentally and meaningfully stronger.

And that's before considering AI's impact on science, healthcare, education, productivity, and much more.

It's not time to slow down but to accelerate!

See 2 related tweets

@Hesamation: Hugging Face (as the highest-profile victim of these cyber incidents) is calling for "acceleration" ...
@_NathanCalvin: These seem like good reasons to accelerate adoption of ai for cyber defense, but not really good rea...

11. jxnlco (Group Score: 81.3 | Individual: 54.8)

Cluster: 2 tweets | Engagement: 7876 (Avg: 290) | Type: Tech

RT @sharifshameem: I just built a functioning React app by describing what I wanted to GPT-3.

I'm still in awe. https://t.co/UUKSYz2NJO

See 1 related tweets

@hammer_mt: I saw this post because I was following this guy because he had the same name as one of my early emp...

12. haider1 (Group Score: 76.4 | Individual: 37.3)

Cluster: 3 tweets | Engagement: 295 (Avg: 81) | Type: Tech

Anthropic employee Levent claims he was able to replicate 5 of openai Astra's 10 proofs using fable

he hasn't shared the proofs yet

but even if it's true, i'm not sure why he focused on replicating Astra's work instead of using Fable to solve new open problems https://t.co/B2QNY1SNpZ\n\nQT @alpoge: so after 24h i have half of them with fable

i didn’t see much discussion of prompting in the announcement but this is a similar setup as with my e.g. unit distance announcement:

totally autonomous, generic prompt, no internet (+ paranoia to ensure no information leaked)

See 2 related tweets

@kimmonismus: OpenAI unveiled 10 frontier math results from its unreleased Astra model: 24 hours later, an Anthrop...
@haider1: RT @haider1: Anthropic employee Levent claims he was able to replicate 5 of openai Astra's 10 proofs...

13. RoundtableSpace (Group Score: 76.1 | Individual: 27.8)

Cluster: 3 tweets | Engagement: 113 (Avg: 115) | Type: Tech

YC just open sourced the multi-agent harness they use internally to run their entire organization.

It's called QM - multiplayer, cloud-first, lives in Slack and on the web.

Crons, webhooks, memory, shared files, agent browser support Connectors for a company brain, shareable web artifacts, multi-player projects

MIT license. Used across YC's accounting, legal, events, and engineering teams — including to build QM itself

THE INTERNAL TOOL THAT RUNS Y COMBINATOR IS NOW FREE FOR EVERYONE.

Github: https://t.co/Os5jdqy8Gy

See 2 related tweets

@Austen: The times they are a-changin’ https://t.co/tcZZhbuqyE\n\nQT @ycombinator: We’ve decided to open-sour...
@SuperagenticAI: Love seeing YC open-source a multiplayer agent computer. SuperQode sits next to that layer as a vers...

14. 0xMovez (Group Score: 74.8 | Individual: 29.1)

Cluster: 3 tweets | Engagement: 99 (Avg: 67) | Type: Tech

Anthropic engineer:

"At Anthropic, 90% of our engineers use loops and ‘dreaming’ to build self-improving agentic systems. сlose the agent loop.

give an agent a way to verify its own output."

in a 30-minute session, an Anthropic team member explains how to build an agent that improves itself.

Claude + loops + dreaming + CLAUDE.md - that’s the secret.

Watch the talk, then save the playbook below.\n\nQT @0xCodez: https://t.co/eJB3TgyJwV

See 2 related tweets

@AnatoliKopadze: Anthropic engineers just showed how they build a full app from scratch, using a loop of agents.

40 ...

@0xMovez: RT @0xMovez: Anthropic engineer:

"At Anthropic, 90% of our engineers use loops and ‘dreaming’ to ...

15. deedydas (Group Score: 74.7 | Individual: 37.9)

Cluster: 2 tweets | Engagement: 789 (Avg: 517) | Type: Tech

Meritocracy in tech is precarious. It’s everywhere and nowhere at the same time.

Cynics will tell you there is none. How can it possibly be fair that the Stanford grad seemingly can raise from any VC they want but the VirginiaTech one has to struggle? If you’re from a big lab, you’re the prom queen, but if you’re from a yesteryear SaasCo, no one even picks up your resume. Warm intros through people in the know seem to dictate who is allowed to win. To even get an interview at YC or a VC gig, it feels like an elite education is a prerequisite. It can seem like the system is rigged against you.

Yet, that view would be so patently false. The valley is one of the best meritocracies in the world. It’s just not obvious when you’re young. Tim Cook went to Auburn and spent 12yrs at IBM before Compaq. Nikesh Arora, one of the highest paid execs in America, went to a “lower ranked” IIT from India. Jan Koum dropped out of San Jose State to work at Yahoo before founding Whatsapp. The CTO of Anthropic went from a no-name college most couldn’t put on a map. Dylan Patel who sits on the most influential set of data in the world, got there by pure passion. Dwarkesh was a nobody before running the most influential podcast in tech. The team that built GPT-3 isn’t exactly a Harvard reunion. Jensen was from Oregon State and Denny’s. Tobi Lutke, Palmer Luckey, Frank Slootman, John Carmack, Jeff Dean, Linus Torvalds, Satya Nadella, the examples are infinite.

How can both of these be true? At any given moment, it feels like a stratified society based on your current credentials. What the Valley offers is the fastest path to updating those credentials.

The unspoken rule of the valley is that you get infinite shots on goal, but you just have to prove yourself to buy the next shot. If you try to jump too high, you might hit a brick wall. For some, it’s easy. You went to Stanford / MIT. Everyone thinks you’re the sharpest of your generation. But opportunity is abundant and you should be risk-on. One amazing product, one viral open source repo, one phenomenal company you helped build, one highly technical blog post, one amazing career stint: all can shape your fortunes overnight.

We often forget how much of a privilege that is. In most other parts of the world, you cannot in fact work yourself into the elite strata. CEOs of companies in India or China rarely are from a non-elite university. Corporate structures there are far less accommodating to former failures. Even in New York finance, there’s a certain way you need to speak, attire you need to wear and an unspoken private club that needs to welcome you. That doesn’t exist in the Valley. Sure there are private clubs, but anyone can make it to them. But you can always buy your entrance ticket with proof of work.

In the moment, it can seem not meritocratic. This happens especially when you’re younger. Hell, it’s not perfect and luck may never strike you. But career is a long game, and a wise valley veteran once told me “if you’re ambitious and you spend 25yrs here, there’s next to no chance you’re not making it.” And I’ve seen that too. It might not be fair in the moment, and it may not be fair to everyone, but it converges very quickly to it.

See 1 related tweets

@garrytan: Everyone mistakes the map for the territory

Meritocracy is that the territory matters more than the...

16. yetone (Group Score: 74.6 | Individual: 27.9)

Cluster: 4 tweets | Engagement: 65 (Avg: 117) | Type: Tech

我也是用 Coding Agent 撸自己的 Agent Sandbox 跟调度器，像模像样的，估计也要上生产了。因为这个确定性比较高，特别容易跑 lossless feedback loop。\n\nQT @RonVonng: 俺有一个老板朋友，也不懂 Rust，最近让 Codex 用 Rust 重写了 DuckDB，然后据他说上了自己的生产环境。胆儿这么肥，我大为震撼。 BTW 它还用 Rust 重写了 Kafka 和 Neo4j。给我的观感就是大猩猩拿着机关枪疯狂扫射。 Repo ➡️ https://t.co/Xd6IvM8pDv https://t.co/p1SZjSWXBz

See 3 related tweets

@xicilion: 说一件疯狂的事情，我发现评论里我关注的推主都出了几十个 agent 了。顺便又关注了几个。😂\n\nQT @tianyi: 如果你是 Agent Harness 相关开源项目的开发者，希望参加 Dee...
@yetone: 我都记不住，我只记住了你在手工时代两次把 Tauri 应用重构成 Electron 应用。而且都是相同类型的应用。你就是太犟了。\n\nQT @localhost_4173: ChatWise: El...
@xicilion: 先转为敬\n\nQT @ashfold: 我们开源了一个史上最快的模型请求库。比原始sdk快10多倍，核心是deepseek flash这种高速模型请求的序列化与内存开销会越来越恐怖。整个库还支持了...

17. BrianRoemmele (Group Score: 74.1 | Individual: 37.8)

Cluster: 2 tweets | Engagement: 228 (Avg: 318) | Type: Tech

I was asked this question:

“How should OpenAI address the competition from cheaper open-source alternatives?”

I answered it.

There is only one way.

One.

Meet the competition on that level. Do what they set out to do: make the AI open source.

Why?  Because the market will be moved by them at the lowest entry levels—and grasped by them at the highest. Ecosystems will form at kitchen tables, in garages, and in campus rooms around free access to performative models. That access creates a stickiness to the brand that can last a lifetime.

Instead, they cling to the idea that revenue must be extracted from the lowest entry point into their system.  This is short-term thinking.  It is lazy thinking.  It comes from people with naïve business experience and even thinner life experience.

Since the opening of the iPhone App Store, a generation of founders and venture capitalists has lived to replay the same tired success story: renting server time.  Exactly like the mainframe era of the 1970s.  That era is long dead.  OpenAI and Anthropic never got the memo.

There are dozens of far more substantial monetization systems that can be built around this new epoch of AI access.  I will not list them here (hire me and I’ll tell you).  They exist.  The tragedy is that the talent stack at both companies almost guarantees they will never see them—even when they ask their own models for solutions. There is no moat that uniquely protects the base models we currently call LLMs.  A simple reality: every AI model will become good enough for 99% of use cases and collapse into low commodity access and pricing. I predict this with clarity:  If they refuse to open-source the full weights for the entry-level market, they will be forced to subsidize usage below the cost of the electricity just to keep people on the platform—hoping someone buys the fries and the shake with the hamburger.

Let me make it so simple even an AI executive can understand it: It makes better sense for the user to burn their own RAM, their own processor, their own GPU, and—most importantly—their own electricity running your model.

Read that again. Read that again. And now think.

I am not guessing.  This is the way it is going to be.  The only question is how fast the world catches up.

The magic of this period I call the Interregnum will belong to the new companies that form abstraction layers on top of what is now the electricity of computing: AI. AI is electricity.  What we connect to electricity today is not just the lightbulb Edison imagined.  It is hundreds of trillions of motors—large and small—that power hundreds of trillions of systems, that also power hundreds of trillions of transistors we call computers… and yes, also lightbulbs.

I have had these conversations with people in the upper echelon of these companies.  They can’t understand the concept.  Yet they still grasp at a $20 a-month subscription from the common person, hoping that drop in the bucket will somehow monetize multi-billion-dollar investments.

It never will.

Hire the right people—people with real experience.  Encourage creativity instead of “yes, sir” to the CEOs.

Do that, and their names might still be known by their grandkids.

Answered on Quora: https://t.co/9tfLTGir8o

See 1 related tweets

@BrianRoemmele: THIS IS THE ONLY WAY US AI COMPANIES SURVIVE.

AND THRIVE.

I said this in the audio blow in 2019 a...

18. StockSavvyShay (Group Score: 70.3 | Individual: 32.8)

Cluster: 3 tweets | Engagement: 410 (Avg: 664) | Type: Tech

$AMZN is on track to become the first company to cross$ 1 trillion in annual revenue by 2028.

What makes that possible is Amazon compounding across commerce, cloud, logistics, ads and AI while AWS is now at a $169B run rate that's growing 37% with 39% operating margins. https://t.co/jMHitC84xw\n\nQT @StockSavvyShay:$ AMZN finally connected AI demand, capex, margins and the payback period inside one coherent explanation.

AWS is already a ~ $179B annual revenue business growing almost 37% with margins near 39% while backlog has reached ~$ 500B and both its AI and custom-chip businesses have surpassed $25B run rates.

That operating strength makes management’s return framework far more credible as server and networking investments are expected to break even in under 3 years remain useful for at least 5-6 years and generate meaningful FCF after the original investment is recovered.

See 2 related tweets

@coinbureau: ⚡️UPDATE: Amazon $AMZN received about$ 600 MILLION in Q2 after the Supreme Court struck down many of...
@StockSavvyShay: RT @StockSavvyShay: AWS ALONE COULD BE WORTH $2 TRILLION

$AMZN CEO Andy Jassy says AWS is “booming”...

19. gdb (Group Score: 69.2 | Individual: 41.0)

Cluster: 2 tweets | Engagement: 4039 (Avg: 1201) | Type: Tech

chatgpt for building interactive educational tools\n\nQT @thebuggeddev: Only if education could be this interactive ❤️‍🔥

I've had a looong wish to build something genuinely useful through vibe coding, and I finally did it.

A 3D human anatomy application built with @threejs using GPT 5.6 Sol.

It all started with a single design image that I created using GPT Image 2.0. I then used it to generate every 3D organ image, one by one.

Next, I converted each of those images into 3D models using @tripoai (and no, they didn't sponsor this 😄).

After that, I opened Codex, wrote a master prompt based on the design, and gave it the prompt, the design image, and all the 3D models.

Codex built the first version beautifully, but there was one big problem.

Every single 3D model was nearly 120-150 MB. That obviously wasn't practical for the web and was giving a performance of 16fps.

After a few iterations, Codex optimized each model down to roughly 2–5.5 MB while preserving the visual quality, reducing the total asset size from ~900 MB to just 28.6 MB. And each model loads on demand.

Along the way, Codex also generated those anatomical illustrations showing where each organ sits in the human body, and even created the interactive hotspot markers that explain different parts of every organ. It handled all of that.

The process wasn't exactly one shot, but it also wasn't difficult. You just have to do it step by step.

It genuinely felt like building something that could make learning anatomy much more engaging.

The inspiration came from @DilumSanjaya's 3D animal plant cell project. I remember seeing it and thinking, "I want to build something like this one day."

And I did it :D

Live: https://t.co/gIFgkj8UVB Code: https://t.co/nijvyWcyxO

See 1 related tweets

@OpenAIDevs: RT @thebuggeddev: Only if education could be this interactive ❤️‍🔥

I've had a looong wish to build ...

20. adisingh (Group Score: 69.2 | Individual: 23.2)

Cluster: 4 tweets | Engagement: 5 (Avg: 35) | Type: Tech

The value of human-originating ideas, content, and art has never been higher.

If everyone can create, those who do it best will win in a big way.\n\nQT @rauchg: AI alone is cool. But mastery + creativity + AI hits on a whole different level. Don't let anyone discourage you from pursuing excellence and craft. Keep studying the blade.

See 3 related tweets

@rauchg: AI alone is cool. But mastery + creativity + AI hits on a whole different level. Don't let anyone di...
@chrisalbon: It feels meaningful that at a moment when a chatgpt subscription can have you solving formerly impos...
@larsencc: Keep studying\n\nQT @rauchg: AI alone is cool. But mastery + creativity + AI hits on a whole differe...