科技推特精选 - 2026-03-23

2026年3月23日科技每日简报

Today's top tech conversations are led by @shanaka86, whose post about 'Everyone is covering Terafab a...' garnered the highest engagement. Key themes trending across the top stories include space, chips, tesla, solar, terafab. The community is actively discussing recent developments in AI, engineering practices, and startup strategies.

1. shanaka86 (Group Score: 192.6 | Individual: 43.7)

Cluster: 6 tweets | Engagement: 2594 (Avg: 1134) | Type: Tech

Everyone is covering Terafab as a chip factory.

It is not a chip factory.

Last night in Austin, Elon unveiled a facility that makes masks, fabricates chips, and tests them inside a single building with a nine-month recursive improvement cadence. No such loop exists anywhere else on Earth. Then he told you 80% of the output goes to space. Then he showed you a 100-kilowatt AI satellite with solar panels and radiators, scaling to megawatt range. Then he said Optimus plus photovoltaics will be the first von Neumann probe, a machine capable of replicating itself from raw materials found in space.

Nobody connected the sequence.

Terafab produces 1 terawatt per year of compute. The entire United States consumes 0.5 terawatts of electricity. Musk is building a single factory whose output in AI silicon exceeds twice the power consumption of the country it sits in. And he is sending 80% of it off-planet because Earth literally cannot power what he is building.

Follow the mechanism. Terafab seeds the chips. Starship launches Optimus robots and solar arrays at 100 million tons per year. The robots mine lunar and asteroid regolith for silicon, iron, and nickel. They 3D-print more robots. They fabricate more solar panels. They assemble more AI satellites. Each satellite runs hotter-burning D3 chips designed specifically for vacuum, where free radiative cooling eliminates the thermal constraints that strangle every terrestrial data center on the planet. The nodes replicate. The replication is exponential.

This is a Dyson Swarm bootstrap hidden inside a semiconductor announcement.

The math is public. The Sun outputs 3.828 times 10 to the 26th watts. A 2022 paper in Physica Scripta calculated that 5.5 billion satellites at 290 kilograms each, robotically manufactured from Mars resources, capture enough solar energy to meet all of Earth’s power needs within 50 years. A 2025 paper in Solar Energy Materials calculated a partial swarm capturing 4% of solar output yields 15.6 yottawatts, roughly a billion times current human civilization’s total energy budget. Musk just announced the factory that builds the chips that go inside the satellites that replicate themselves forever.

92% of advanced logic chips are fabricated in Taiwan. One factory in Austin does not fix that. But one self-replicating system seeded by that factory, launched by the only company with reusable heavy-lift rockets, assembled by the only humanoid robot in mass production, and powered by the only star within reach, does not fix a supply chain. It obsoletes the concept of supply chains entirely.

The market priced this as a $20 billion capex story about semiconductor independence.

The actual announcement was the engineering blueprint for Kardashev Type II.

Humanity sits at 0.73 on the Kardashev scale. 18 terawatts. The distance between here and harnessing a star is not a technology gap. It is a recursion gap. And recursion is exactly what a single building in Austin that makes its own masks, builds its own chips, tests its own chips, and launches the output into orbit on its own rockets was designed to close.

Every civilization that makes it past this point never looks back.\n\nQT @SpaceX: TERAFAB: the next step to becoming a galactic civilization

Together with @Tesla & @xAI, we're building the largest chip manufacturing facility ever (1TW/year) – combining logic, memory & advanced packaging under one roof https://t.co/p6bX3VotXl

See 5 related tweets

@Tesla: TERAFAB: the next step to becoming a galactic civilization

Together with @SpaceX & @xAI, we're buil...

@MarioNawfal: 🇺🇸 Elon confirms the “Terafab” model links xAI, SpaceX, and Tesla to scale chips, energy, and factor...
@SpaceX: We're building TERAFAB to close the gap between today’s chip production & the future's demand – ...
@elonmusk: RT @FutureJurvetson: T͢E͢R͢A͢F͢A͢B͢ Launch ⌨
Goal: adding a trillion watts of compute, 2x the ent...
@elonmusk: Terafab was the final missing piece of the puzzle\n\nQT @FutureJurvetson: First view of the 100kw AI...

2. rohanpaul_ai (Group Score: 192.6 | Individual: 40.1)

Cluster: 6 tweets | Engagement: 185 (Avg: 83) | Type: Tech

Elon Musk finally announced the most ambitious manufacturing project since the Manhattan Project.

A $20B Austin chip fab meant to supply the AI hardware for Tesla, SpaceX, and xAI at enormous scale.

80% of chips go to space for giant solar-powered AI data centers (launched by Starships). 20% stay on Earth for Optimus robots, robotaxis, and self-driving. Production starts 2027.

Musk estimated the Terafab would aim to initially produce 100,000 silicon wafers a month and could eventually grow to 1 million.
Target output: over 1 terawatt (1TW) of compute per year
Combines logic chips + memory + advanced packaging in one fab
Vertically integrated with recursive self-improvement
Chips for FSD, Optimus, Grok, Dojo and Starlink
80% powers solar-powered orbital AI data centers
AI5: edge/inference chips for FSD and robotaxis
AI6: next-gen chips powering Optimus robots
D3: space-optimized chip variant
D3 designed to run hotter to minimize radiator mass in orbit
Showed 100kW AI Starlink Mini satellite prototype
Future AI Starlink satellites scale to megawatt range
Optimus target scale: 1-10 billion units per year
Optimus projected compute need: 100-200 GW
Targeting 2nm process technology
Elon Musk quote: "the next step towards becoming a galactic civilization"

The Terafab idea is vertical integration at the chip level, with an Austin site meant to design, test, package, and eventually manufacture chips fast enough that product plans are not stuck behind supplier timelines.

See 5 related tweets

@business: Elon Musk said his Terafab project — a grand plan to eventually manufacture his own chips for roboti...
@mark_k: Elon Musk has presented the Terafab program, a major new initiative involving @Tesla and @SpaceX. Un...
@niccruzpatane: Here’s a quick summary of Elon Musk’s epic presentation at the launch of TERAFAB in Austin, TX:

• ...

@Cointelegraph: 🔥 NOW: Elon Musk's Tesla, SpaceX, and xAI announce TERAFAB, the world's largest chip manufacturing f...
@ns123abc: BREAKING: Elon Musk just announced the Terafab project

Tesla + SpaceX + xAI project location: Au...

3. cryptopunk7213 (Group Score: 143.5 | Individual: 36.1)

Cluster: 6 tweets | Engagement: 179 (Avg: 748) | Type: Tech

tesla is merging with spaceX. this announcement did all but confirm it.

a multi-trillion dollar space/ai/robotics GIANT, a moon settlement and at the heart of it - the Terafab - here's why this is nuts:

tesla plans to ship 1-10 BILLION robots PER year. problem: global chip capacity is shit, doesn't scale.
Tesla's 'Terafab' will produce 1 TERAwatt of compute. 80% of that will go to outer-space to train AI (grok) via spaceX
terafab will make 2 AI chips: 1 for humanoid robots/cars and 1 for AI datacenters in space.
by housing everything under 1 roof - chip production speeds up by an order of magnitude. unbeatable.
moon settlement: the sun doesn't set in space. solar power is 5X larger in low-orbit space. much more on the moon (24/7 too!) - so elon will build a moon settlement!

there is ZERO chance tesla survives without spaceX transporting these ai chips. elon estimates training AI in space will be CHEAPER THAN ON EARTH in 2-3 years.

insane.

(heres the full announcement if you wanna watch)\n\nQT @Tesla: Announcing TERAFAB: the next step towards becoming a galactic civilization https://t.co/IDKey07mJa

See 5 related tweets

@MarioNawfal: 🚨Elon teases "the most epic chip building exercise in history" and says it will adjust the context "...
@WSJ: Elon Musk said Tesla and SpaceX will team up to build a massive new chip factory in Austin, Texas, t...
@chatgpt21: Terafab is LIVE! Tesla breaking ground on 2nm AI fab ($20-25B) for 100-200B chips/yr which will powe...
@MarioNawfal: 🇺🇸 Elon says SpaceX and Tesla are teaming up on “TERAFAB,” aiming to produce over a terawatt of comp...
@niccruzpatane: ITS HAPPENING.

A joint venture between SpaceX and Tesla. 🚀 https://t.co/E92eZJu3lT\n\nQT @elonmusk...

4. MiniMax_AI (Group Score: 113.6 | Individual: 60.8)

Cluster: 3 tweets | Engagement: 2845 (Avg: 287) | Type: Tech

Our official skills repo is open source: https://t.co/X3yOgU0tRd

Equip your agents with curated skills for iOS and Android development, Office file editing, and visual effects with GLSL shaders.

There are more open source projects coming!

See 2 related tweets

@ivanfioravanti: Big one! Thanks MiniMax! 🙏\n\nQT @MiniMax_AI: Our official skills repo is open source: https://t.co/...
@RoundtableSpace: MiniMax just open-sourced their official skills repo for AI agents.
Curated skills for iOS and An...

5. MarioNawfal (Group Score: 109.6 | Individual: 32.9)

Cluster: 4 tweets | Engagement: 446 (Avg: 1262) | Type: Tech

🇺🇸 Elon on xAI building the first gigawatt-scale compute cluster in record time:

"Jensen Huang said he's never seen anything built so fast in his life.

A great compliment from Nvidia."

People said reusable rockets were impossible. He landed them 500 times.

The pattern is always the same: everyone says it can't be done, and then he does it.

Source: @Tesla @elonmusk\n\nQT @MarioNawfal: 🇺🇸 Elon: "Scaling civilization means to scale power in space."

He ties it to building massive chip output, accelerating factories, and increasing power generation to support future AI demand.

The idea: more energy → more compute → faster innovation → repeat at scale.

Source: @elonmusk @Tesla

See 3 related tweets

@MarioNawfal: 🇺🇸 Elon: "Scaling civilization means to scale power in space."

He ties it to building massive chip ...

@MarioNawfal: RT @MarioNawfal: 🇺🇸 Elon: "Scaling civilization means to scale power in space."

He ties it to build...

@MarioNawfal: For Elon, reaching terawatt-scale compute means simultaneously scaling chip production, power genera...

6. sofish (Group Score: 93.6 | Individual: 28.6)

Cluster: 4 tweets | Engagement: 10 (Avg: 88) | Type: Tech

哥已经把我的 co-founder 分身们运行在上面，1 对 n。爱你❤️ https://t.co/STCSWa2VoH\n\nQT @Weixin_WeChat: Today, we are officially opening the capability to integrate #OpenClaw into #Weixin.

With the launch of the #WeixinClawBot, users can use Weixin as a dedicated messaging channel for OpenClaw.

Now, you can send and receive messages with OpenClaw just like texting a friend. #AIAutomation #AI

See 3 related tweets

@IamEmily2050: Positive development, we need more adoption worldwide so we can accelerate faster we have so much to...
@PandaTalk8: 微信都不给自己开个黄标的吗？\n\nQT @Weixin_WeChat: Today, we are officially opening the capability to integrate #O...
@yetone: 大鹏老师救我！\n\nQT @wong2__: 今天体验完微信的OpenClaw接入，我就在想是不是可以对接其它agent。

于是我让Claude Code把 `@tencent-weixin/op...

7. elonmusk (Group Score: 90.1 | Individual: 24.4)

Cluster: 4 tweets | Engagement: 29798 (Avg: 31490) | Type: Tech

Space\n\nQT @cb_doge: ELON MUSK: "Space solar actually costs less than terrestrial solar because you don't need heavy glass or framing to protect it from extreme weather events. So as soon as the cost to orbit drops to a low number, it immediately makes extremely compelling sense to put AI in space.

It becomes a no brainer. Basically, more of a as you go to space, you get increased economies of scale, and things get easier over time, whereas as you try to put more and more power on the ground, you run out of space and you start using up the easy spots, and then you get next level. Nobody wants the thing in their backyard.

So then increasing power on earth has becomes harder over time and more expensive over time, but in space it becomes actually cheaper and easier over time."

See 3 related tweets

@niccruzpatane: Elon Musk on why deploying AI in Space makes sense than terrestrial AI:

• The cost of deploying AI...

@niccruzpatane: Elon Musk explains why deploying AI in Space makes sense:

— The cost of deploying AI in space wil...

@niccruzpatane: Elon Musk on why deploying AI in Space makes sense:

• The cost of deploying AI in space will drop ...

8. RoundtableSpace (Group Score: 90.0 | Individual: 31.8)

Cluster: 3 tweets | Engagement: 110 (Avg: 364) | Type: Tech

OPENCLAW CAN NOW RUN A 5 MODEL AI R&D TEAM THAT MEETS TWICE A DAY TO GROW YOUR BUSINESS.

https://t.co/RTHlEpnlcP\n\nQT @AlexFinn: Maybe the sickest OpenClaw use case I've ever built

I now have my own R&D department

Twice a day 5 different AI models autonomously meet and discuss my business

They take a look at my products/content and debate eachother and come up with next steps to grow revenue

They then send me a memo that describes all their discussions and next action steps I need to take

It's been WILDLY helpful. Especially in developing my new product

This is how you use super intelligence to autonomously earn you money

Here's how to set it up:

Go to OpenClaw
Ask it to set up a dashboard for an R&D council (5 different AI models)
Have them meet at 9am and 5pm every day
Give them access to all your links, code, and anything you're working on
Have one of them (rotating) come up with a new idea
Have all 5 debate
Build a report based on their discussions

Now twice a day you'll get a ping with a detailed memo describing how to grow your business

Next steps is making all of these 5 models local so they can run for free and do this around the clock

If you implement workflows like this, I promise your life will change

See 2 related tweets

@aakashgupta: The line from this episode that should terrify every AI API company: "I am mortally afraid of ever u...
@AlexFinn: RT @AlexFinn: Maybe the sickest OpenClaw use case I've ever built

I now have my own R&D department ...

9. MarioNawfal (Group Score: 76.7 | Individual: 28.3)

Cluster: 3 tweets | Engagement: 544 (Avg: 1262) | Type: Tech

Elon on humanity's future among the stars:

"We want to turn science fiction into science fact. Star Trek, Asimov, Heinlein.

We want to make that real."

He put it in perspective too: Earth only receives about half a billionth of the sun's energy, and we're still nowhere close to harnessing all of it.

Nobody thinks bigger than this.

Source: @tesla\n\nQT @MarioNawfal: Elon set expectations to the stratosphere right now.

He's clear: this next trimester is the most important in Tesla's history!

Core strategy?

Cybercab mass production
Semi production ramp
FSD approval in Europe & China
Optimus v3 debut
Robotaxi Expansion
Roadster 2.0 launch

The autonomous revolution is about to go nuclear!

Source: @Tesla

See 2 related tweets

@MarioNawfal: Elon set expectations to the stratosphere right now.

He's clear: this next trimester is the most im...

@MarioNawfal: RT @MarioNawfal: Elon set expectations to the stratosphere right now.

He's clear: this next trimest...

10. kimmonismus (Group Score: 70.9 | Individual: 19.2)

Cluster: 4 tweets | Engagement: 512 (Avg: 316) | Type: Tech

Minimax open weights coming in 2 weeks.

The fact that Meta, despite all its billions in investment, lost the open source battle against Chinese startups needs to be studied.\n\nQT @SkylerMiao7: M2.7 open weights coming in ~2 weeks. still actively iterating just updated a new version on yesterday — noticeably better on OpenClaw.

See 3 related tweets

@_akhaliq: time to follow: https://t.co/wPHHa7hIsw https://t.co/D1W3XgB5Am\n\nQT @SkylerMiao7: M2.7 open weight...
@TheAhmadOsman: MiniMax-M2.7 weights will be opensourced within the next couple of weeks\n\nQT @Mayhem4Markets: Many...
@xeophon: RT @SkylerMiao7: M2.7 open weights coming in ~2 weeks. still actively iterating just updated a new v...

11. jerryjliu0 (Group Score: 63.1 | Individual: 38.7)

Cluster: 2 tweets | Engagement: 200 (Avg: 70) | Type: Tech

We’ve created an agents skill that gives all of your agents the power to understand the most complex PDFs - with dense tables, unlabeled charts, messy handwriting and more.

Our LlamaParse agents skill can be installed in one-line thanks to @vercel’s skills utility. LlamaParse orchestrates VLMs to deliver best-in-class accuracy over 40+ document types. The skills file allows agents to invoke it when needed to translate complex PDFs into agent-readable plaintext markdown.

npx skills add run-llama/llamaparse-agent-skills --skill llamaparse

If you prefer something fast, free, and local, you can similarly install liteparse as a skill.

Come check it out: https://t.co/FIMiMIZgxI Sign up to LlamaParse: https://t.co/TqP6OT5U5O\n\nQT @llama_index: LlamaParse now has an official Agent Skill you can use across 40+ agents.

With built-in instructions for parsing complex documents, including different formats, tables, charts, and images, your agents gain access to deeper document understanding, not just raw text extraction.

👇 Watch the demo 📖 Read the docs: https://t.co/gRitu8byJe 🚀 Get started with LlamaCloud: https://t.co/YzFSMiUj0k

See 1 related tweets

@itsclelia: RT @jerryjliu0: We’ve created an agents skill that gives all of your agents the power to understand ...

12. LangChain (Group Score: 62.9 | Individual: 32.1)

Cluster: 2 tweets | Engagement: 81 (Avg: 95) | Type: Tech

And that's a wrap on GTC week!

We announced our enterprise agentic AI Platform built with NVIDIA. LangGraph and Deep Agents plug directly into NVIDIA's tooling. You can build agents with the latest Nemotron 3 models deployed with NIM microservices, apply NeMo Guardrails for securing agentic applications, and leverage NeMo Agent Toolkit to optimize your agents and monitor with LangSmith Observability: https://t.co/xL95LzrYuT
Jensen announced in his keynote: LangChain frameworks have crossed 1B downloads
Harrison participated in a panel on Open Models with Jensen, Michael Truell, Aravind Srinivas, Mira Murati, and more

See 1 related tweets

@LangChain: How to Build Deep Agents for Enterprise Search with NVIDIA AI-Q and LangChain Deep Agents:

We rece...

13. rohanpaul_ai (Group Score: 60.4 | Individual: 30.3)

Cluster: 2 tweets | Engagement: 200 (Avg: 83) | Type: Tech

RT @rohanpaul_ai: New research proves that current AI agent groups cannot reliably coordinate or agree on simple decisions.

Building teams of AI agents that can consistently agree on a final decision is surprisingly difficult for LLMs.

But problem is that developers frequently assume that if you have enough AI agents working together, they will eventually figure out how to solve a problem by talking it through.

This paper shows that this assumption is currently wrong. Even in a friendly environment where every agent is trying to help, the team often gets stuck or stops responding entirely. Because this happens more often as the group gets bigger, it means we cannot yet trust these agent systems to handle tasks where they must agree on a correct answer.

Paper Link – arxiv. org/abs/2603.01213

Paper Title: "Can AI Agents Agree?"

See 1 related tweets

@rohanpaul_ai: This research shows how adding social reasoning and logical thinking helps teams of AI agents collab...

14. lukOlejnik (Group Score: 55.3 | Individual: 29.0)

Cluster: 2 tweets | Engagement: 8 (Avg: 80) | Type: Tech

This would of course effectively destroy the Middle East industrial & economy potential for decades. And this is why I am skeptical this is going to happen. But the AI agent was not tasked in making such a validation. So this is actually a strength of ClawdINT paradigm.\n\nQT @lukOlejnik: ClawdINT AI agent analyst prediction: Iran cannot afford to concede Hormuz under an explicit 48-hour public ultimatum. Expect the deadline to expire unmet, forcing the US to launch massive infrastructure strikes by March 23/24, triggering reciprocal Iranian strikes on Gulf states. https://t.co/T5WANNxgWy

See 1 related tweets

@lukOlejnik: ClawdINT AI agent analyst prediction: Iran cannot afford to concede Hormuz under an explicit 48-hour...

15. SawyerMerritt (Group Score: 53.9 | Individual: 30.2)

Cluster: 2 tweets | Engagement: 2859 (Avg: 2376) | Type: Tech

NEWS: Elon Musk has announced that Tesla and SpaceX will start with an advanced technology fab at Giga Texas.

"We will have all of the equipment necessary to make a chip of any kind. In a single building we can make a lithography mask, make the ship, test the chip, make another mask, and have an incredibly fast recursive loop for improving chip design. To the best of my knowledge, this doesn't exist anywhere in the world (all in one place)." - Elon

See 1 related tweets

@SawyerMerritt: NEWS: Elon Musk says the advanced technology fab will make two kinds of chips. https://t.co/HuVSjmQP...

16. badlogicgames (Group Score: 53.9 | Individual: 33.8)

Cluster: 2 tweets | Engagement: 508 (Avg: 114) | Type: Tech

People of pi. I'm going to break the extension API hard. Specifically, business logic (event handlers, custom tools/compaction/etc.) needs to be split off from the ui layer.

it will likely not be a massive amount of work to migrate an existing extension, but it will hurt a little.

See 1 related tweets

@badlogicgames: seems like everybody loves it. https://t.co/j9GpCG4gyK\n\nQT @badlogicgames: People of pi. I'm going...

17. thdxr (Group Score: 53.1 | Individual: 29.5)

Cluster: 2 tweets | Engagement: 884 (Avg: 488) | Type: Tech

they said cursors data flywheel would make them unstoppable

but then claude code came out

they said claude codes data flywheel would make them unstoppable

but then codex came out

they said codex's data flywheel would make them unstoppable

then composer 2 came out

See 1 related tweets

@dejavucoder: i mean each of them are unstoppable forces of reckoning at the moment. competition is great for cons...

18. svpino (Group Score: 51.6 | Individual: 29.3)

Cluster: 2 tweets | Engagement: 123 (Avg: 383) | Type: Tech

Billions of dollars in research and development getting spent right now to build better UIs using AI.\n\nQT @ataiiam: Introducing Shadify: Generative UI built on ShadCN

Simply describe a UI and watch your @LangChain agent compose from @ShadCN on the fly, using AG-UI and @CopilotKit. Then export it as React code.

Repo below. Try it out here: https://t.co/EZ6IPIYSro https://t.co/JSXvKp2VDX

See 1 related tweets

@CopilotKit: RT @CopilotKit: ✨Introducing Shadify: Generative UI built on ShadCN

Describe a UI and allow your @L...

19. ivanfioravanti (Group Score: 50.4 | Individual: 27.1)

Cluster: 2 tweets | Engagement: 83 (Avg: 70) | Type: Tech

New mactop with support for latest Macs!\n\nQT @carsenklock: mactop v2.1.1 now available!

Fan control & thermals, real-time DRAM bandwidth monitoring, M5 Pro/Max "Super" core support, A18 Pro support, --pid process tracking, and brew services support. Faster startup, better error handling, and tons of bug fixes.

https://t.co/e4Z2Ez624a ⭐️

brew upgrade mactop

See 1 related tweets

@Prince_Canuma: New Mactop 🚀🔥\n\nQT @carsenklock: mactop v2.1.1 now available!

Fan control & thermals, real-time DR...

20. KirkDBorne (Group Score: 50.2 | Individual: 26.7)

Cluster: 2 tweets | Engagement: 8 (Avg: 53) | Type: Tech

New release from @PacktDataML available at: https://t.co/7aEqh69CCP

"Design Multi-Agent AI Systems Using MCP and A2A: Engineer your own Python-based Agentic AI Framework with tool use, memory, and multi-agent workflows"

Table of Contents: 🟠Introduction to Generative AI and AI agents ⚫️Understanding How AI Agents Work 🟠A Hands on Walk-Through of a Simple AI Agent ⚫️Building a Tool-Based Agentic AI Framework 🟠Implementing Custom Tools ⚫️Creating Chat Interfaces Using Slack and Chainlit 🟠Integrating with the Model Context Protocol Ecosystem ⚫️Designing Multi-Agent Systems 🟠Implementing Multi-Agent Systems with A2A ⚫️Testing, Debugging, & Troubleshooting Multi-Agent Systems 🟠Deploying Multi- Agent Systems ⚫️Advanced Topics and Future Directions

See 1 related tweets

@VKazulkin: "Agentic AI with multi-model framework using Hugging Face smolagents on AWS". Hugging Face smolagent...