科技推文精选 - 2026-06-09

2026年6月9日科技每日简报

Today's top tech conversations are led by @TrungTPhan, whose post about 'Steve Jobs finding out that th...' garnered the highest engagement. Key themes trending across the top stories include https, super, agents, models, infrastructure. The community is actively discussing recent developments in AI, engineering practices, and startup strategies.

1. TrungTPhan (Group Score: 395.4 | Individual: 31.9)

Cluster: 27 tweets | Engagement: 48 (Avg: 229) | Type: Tech

Steve Jobs finding out that the new Siri AI features — 16 years after he acquired the company — are a calendar reminder, photo finder and contact list search https://t.co/2ZqKx4TA5M\n\nQT @TrungTPhan: Apple WWDC just demo’d new Siri AI powered by Apple Intelligence (aka Gemini distillation).

It’s supposed to be more conversational with on-screen visual intelligence and can integrate user context (photos, contacts, calendars).

The WWDC 2024 version was pure vapourware but this looks more legit because Siri responses actually take a while to respond. After getting fined last time for false advertising, Apple can’t mess around with this Siri update.

See 26 related tweets

@testingcatalog: WWDC 🔥: A new Apple Intelligence powered by Gemini, conversational Siri AI, updated image editing, a...
@TrungTPhan: Siri AI with on-screen “visual intelligence” when someone opens an incognito Safari browser, lowers ...
@FirstSquawk: APPLE ANNOUNCES SIRI AI BETA LAUNCH LATER THIS YEAR, HOME APP INTELLIGENCE FOR CAMERA CLIP ANALYSIS,...
@business: Apple unveiled the next generation of its AI platform, including an update to its Siri digital assis...
@FirstSquawk: APPLE INTRODUCED A NEW AI-POWERED VERSION OF SIRI THROUGH APPLE INTELLIGENCE, DESIGNED TO BE MORE CO...

2. FellMentKE (Group Score: 288.2 | Individual: 39.0)

Cluster: 8 tweets | Engagement: 153 (Avg: 142) | Type: Tech

Every AI assistant right now is basically an integration wrapper.

You ask it to book an Uber and it tells you to open the app yourself.

Super actually does it. No MCP, no skill files, no setup required. She runs her own cloud apps and handles the real-world task herself.

https://t.co/2UL6yEHDh9\n\nQT @RohanArun: Introducing Super, the world's first AI with her own apps.

Super can schedule Ubers, order groceries, and have Starbucks delivered, just like a real assistant.

AI was meant to make our lives easier, but instead, we have to learn MCP, skills, and tools that still feel brittle.

Rather than relearning new apps each year, Super automates them for you and meets users at their level of understanding.

She acts as the single concierge for apps and the real world, saving you time and money.

Unlike most AI assistants that only automate your calendar or email integrations, Super has her own email, phone number, and apps in the cloud.

You can integrate her into your life like an assistant without giving up access to your private email or calendar.

We invented a new primitive called Computer-Use Cache, which reduces costs to near-zero for repeatable tasks so cloud automations are effectively free.

No subscriptions! Super can purchase products and services for you from her wallet, which you pay as you go.

Our thesis: apps are a vestige of the mobile era, and we're building what comes next.

You can access Super over iMessage, Meta Display glasses, and iPhone today.

Text Super at https://t.co/jxedA3kxU8, and if you're one of the first 5,000 in the Bay Area, you get a free Starbucks!

Comment and repost ♻️ and I'll DM you free money to buy things with Super! US and Canada only, while supplies last.

See 7 related tweets

@ai_explorer25: The breakthrough here isn't a smarter model.

It's that routine computer-use tasks now run 60% faste...

@cgtwts: Apple CEO right now: https://t.co/N3A8u1bXr9\n\nQT @RohanArun: Introducing Super, the world's first ...
@Scobleizer: Apple just demonstrated some of these things for new Siri, but that won't be available until later t...
@DataChaz: 🚨@Super_Powers_AI just dropped.

Not just iMessage, it is fully ready for @Meta glasses too 😎

Their...

@svpino: This looks like the AI that everyone has been promising, but nobody has delivered so far.

By the wa...

3. PaulSolt (Group Score: 274.0 | Individual: 41.9)

Cluster: 13 tweets | Engagement: 2328 (Avg: 83) | Type: Tech

RT @steipete: Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore.

You should be designing loops that prompt your agents.

See 12 related tweets

@garrytan: Don’t build Foxconn factories that make agents do the same thing over and over

Agents by and large ...

@rezoundous: Step by step set up guide

Get unlimited tokens
Loop
???
Profit\n\nQT @steipete: Here’s ...

@Yuchenj_UW: On the whole:

“You shouldn’t be prompting coding agents anymore. You should be designing loops that...

@DominikTornow: I don't know how to do this. Does anybody have some pointers?!\n\nQT @steipete: Here’s your monthly ...
@codewithantonio: designing loops is so outdated. if you’re not astral projecting to become one with claude, you’re ng...

4. MTSlive (Group Score: 248.9 | Individual: 38.0)

Cluster: 10 tweets | Engagement: 833 (Avg: 116) | Type: Tech

SITUATION UPDATE: Google has placed an order with Intel to manufacture more than 3 million TPUs in 2028, per The Information.

See 9 related tweets

@wallstengine: Google and $NVDA are considering Intel$ INTC as a backup chip manufacturer as TSMC capacity remains ...
@rohanpaul_ai: The Information reports that Google has picked Intel to manufacture 3M+ Google TPUs in 2028.

Great ...

@jukan05: I’m not sure whether Qianer Liu phrased it that way intentionally, but I don’t understand why Intel ...
@StockSavvyShay: $GOOGL reportedly ordered$ INTC to package more than 3M TPUs.

Google is expected to produce 6M+ TPU...

@mark_k: Google has reportedly ordered more than 3 million AI chips from Intel for 2028 production. The key d...

5. JayminSOfficial (Group Score: 193.1 | Individual: 36.6)

Cluster: 9 tweets | Engagement: 613 (Avg: 517) | Type: Tech

While much of the AI conversation revolves around models, agents, and applications, the real strategic race is increasingly moving down the stack.

The UK's new $1.5 billion AI Hardware Plan is a recognition of a simple reality: nations that control compute, chips, and AI infrastructure will have a disproportionate influence on the next wave of technological and economic growth.

A few things stand out:- • £750 million for a national AI supercomputer • Direct government commitments to purchase next-generation AI chips from startups • Dedicated funding for AI hardware innovation and semiconductor talent • A major effort to attract private capital into British hardware companies

What's particularly interesting is the focus on inference hardware.

As AI adoption scales globally, inference will become one of the largest infrastructure markets in technology. The companies that make AI cheaper, faster, and more energy efficient will create enormous value.

For years, software captured most of the attention. The next decade could see hardware become one of the most important competitive battlegrounds in AI.

The countries investing early in sovereign compute, semiconductor innovation, and AI infrastructure are positioning themselves for long-term leadership.

The UK just made it clear it intends to be one of them.

https://t.co/QZQhKHjwYF

See 8 related tweets

@KanishkaNarayan: RT @SebJohnsonUK: The UK Government has just announced a £1.1bn package for AI Hardware!

LETS GO

H...

@MTSlive: SITUATION DETECTED: Britain has announced a $1.5B plan to build domestic AI computing capacity, incl...
@AMD: “The United Kingdom has the talent, research excellence and ambition to help lead the next era of AI...
@StockMKTNewz: $AMD just announced plans to invest up to £2 Billion in the United Kingdom 🇬🇧 over the next 5 years ...
@Reuters: UK sets out $1.5 billion AI hardware plan with supercomputer, chip funding https://t.co/p2v2cs31q4 h...

6. TeksEdge (Group Score: 184.1 | Individual: 65.4)

Cluster: 4 tweets | Engagement: 1117 (Avg: 27) | Type: Tech

🔥Holy 💩! New King of the Open Source LLMs! Beats Leading OS Models. Time to start testing to verify benchmarks.

🏆 Nex-N2-Pro beats ✅ Kimi-K2.6: 86% (13/14) ✅ MiniMax: 100% (8/8) ✅ GLM-5.1: 100% (13/13) ✅ DeepSeek-V4-Pro: 92% (11/12) » Opus 4.7: 45% (5/11) » GPT-5.5: 30% (3/10)\n\nQT @NexEcosystem: 📢 Nex-N2 is here! A family of agentic models that doesn't just think, it acts! Coding, search, tool use. All fused into a single agentic reasoning loop.

Adaptive Thinking, auto-scales reasoning depth per step. Saves ~20% tokens, zero performance loss.
Coherent Thinking, one thinking paradigm across search, coding, and tool use. No more fragile mode-switching.

🏆 Result: Tier-1 open-source performance on SWE-bench, Terminal-Bench, GDPval, and more, tracking GPT-5.5 and Opus 4.7.

🎉 Open-weight. Try it now. 🔗 https://t.co/7oLSfyOCxB 📦 https://t.co/c2CGhXWaz6 https://t.co/KJYXZIpk8M https://t.co/vcjdZ9cuB6

See 3 related tweets

@ModelScope2022: Nex-N2 is now open source！An agentic model series from Nex AGI built for coding, tool use, deep rese...
@TeksEdge: 🏆 Nex-N2 Open Source Release a new #1 agentic model built for: ➺ Coding ➺ Tool use ➺ Deep research ...
@ClementDelangue: RT @NexEcosystem: 📢 Nex-N2 is here! A family of agentic models that doesn't just think, it acts! Cod...

7. jerryjliu0 (Group Score: 175.9 | Individual: 38.8)

Cluster: 7 tweets | Engagement: 94 (Avg: 37) | Type: Tech

I really do think we'll see a lot of value accrue in AI startups building "model routing as a service"

Not just OpenRouter - this includes a much broader set of verticalized agents and infrastructure.

For us it's document infrastructure: parsing, extraction, search
Another infra analogy is web search: Exa/Parallel
For verticalized apps it could be anything from Cognition to Harvey

The frontier labs own the underlying models, and their main application at the app layer is the enormous amounts of $$ that they have; but their main disadvantage is that they only own a subset of all the points on the Pareto curve.

Speaking from our team's experience, it is both non-trivial and extremely important to find a point on the pareto curve of accuracy and cost. There's enormous amounts of ML time spent on model evaluations and benchmarking, and infra time making sure that the service can scale reliability, without rate limits, and without blowing up cost. At the same time it's extremely important not just for cost reasons but also latency, and oftentimes long-tail accuracy in use cases that demand it.\n\nQT @brian_armstrong: Good take

My guess is

demand for intelligence is near infinite
but 80% of workloads will be running on 99% cheaper models within 12-18 months
20% of workloads will still run on latest gen models where IQ maxing is important (scientific breakthroughs, higher level ochestrator agents?)
rough analogy might be what % of macbooks or gaming PCs sold have the maxed out specs for CPU/GPU, prices are falling much faster than Moore's law here though
this leads me to think the limiting factor will be energy and compute, not better models

At Coinbase we're working hard on routing prompts to cheaper models where appropriate, and in some cases have been able to keep costs roughly flat, while token usage continues to grow exponentially.

See 6 related tweets

@ClementDelangue: RT @brian_armstrong: Good take

My guess is

demand for intelligence is near infinite
but 80% of ...
@pmarca: Interesting.\n\nQT @brian_armstrong: Good take

My guess is

demand for intelligence is near infini...
@julien_c: Model routing is growing a lot these days\n\nQT @brian_armstrong: Good take

My guess is

demand fo...
@levie: The numbers may be a bit extreme here, but unquestionably use-cases have to stratify in the next yea...
@rohanpaul_ai: Coinbase CEO Brian Armstrong thinks AI demand is almost limitless, but he expects 80% of workloads t...

8. StockMKTNewz (Group Score: 154.6 | Individual: 24.3)

Cluster: 11 tweets | Engagement: 127 (Avg: 419) | Type: Tech

Apple $AAPL said it isn’t currently able to launch Siri AI, its redesigned digital assistant, on iPhones or iPads in the European Union 🇪🇺 https://t.co/HZoJAGL9tq

See 10 related tweets

@markgurman: The important details:
Siri AI for developers in English today; all users later this year in beta...
@mark_k: I wouldn't worry too much about Siri AI eating up the entire AI market. You see, every new Android p...
@ivanfioravanti: Time to move to US 🤷🏻‍♂️\n\nQT @markgurman: The important details:
Siri AI for developers in Engl...
@Reuters: When Apple holds its developer conference at its Cupertino, California headquarters on June 8, the b...
@ReutersBiz: When Apple holds its developer conference at its Cupertino, California headquarters on June 8, the b...

9. JayminSOfficial (Group Score: 147.8 | Individual: 32.8)

Cluster: 5 tweets | Engagement: 776 (Avg: 517) | Type: Tech

AI adoption creates leverage. AI learning creates compounding leverage.

As teams deploy more coding agents, they generate a growing body of problem-solving knowledge every day. The question becomes: where does that knowledge go?

Hivemind's approach is compelling because it captures successful patterns, turns them into skills, optimizes them, and propagates them across agents.

When knowledge compounds at the team level, intelligence becomes an organizational asset rather than a feature of any single model.\n\nQT @DBuniatyan: Coding agents that actually get better the more your team uses them.

Introducing Hivemind: continual learning for AI coding agents.

Hivemind turns the traces from every agent your team runs (Claude Code, Codex, Cursor, Hermes, OpenClaw, Pi) into reusable skills, then pushes those skills across all of them. All on your cloud storage.

Now with SkillOpt built in, your skills get trained: +19.1 points of accuracy in Claude Code, +24.8 in Codex, best or tied on all 52 setups tested.

Open source, one line install.

See 4 related tweets

@godofprompt: Coding agents are about to stop acting like isolated tools.

Hivemind gives your whole team one shar...

@kimmonismus: New from Hivemind: continual learning for AI coding agents, available to everyone starting today.

I...

@dr_cintas: Hivemind just launched and your coding agents finally share one brain 🤯

It captures traces from Cla...

@alex_prompter: The future of coding agents is not more prompting.

It’s compounding.

Every trace, fix, workflow, a...

10. wallstengine (Group Score: 141.6 | Individual: 33.1)

Cluster: 5 tweets | Engagement: 468 (Avg: 127) | Type: Tech

$NVDA CEO Jensen on Stock selloff:

“We’re at the beginning of [AI Supercycle], and whatever happened to the stock market, you should be very happy because now you can buy at a discount... Everybody should be very excited

“It is a foregone conclusion that AI will be infrastructure for the world, just like the internet was infrastructure for the world.”

See 4 related tweets

@StockSavvyShay: $NVDA CEO Jensen Huang called Friday’s sell-off a buying opportunity.

“Everyone has the chance to b...

@TheStalwart: Huh\n\nQT @business: Nvidia CEO Jensen Huang calls the global tech stocks selloff that began last we...
@business: Nvidia CEO Jensen Huang calls the global tech stocks selloff that began last week a buying opportuni...
@unusual_whales: Nvidia, $NVDA, CEO Jensen Huang has said that the global tech stocks selloff that began last week a ...

11. kimmonismus (Group Score: 134.5 | Individual: 33.7)

Cluster: 5 tweets | Engagement: 129 (Avg: 310) | Type: Tech

What many misunderstand: Apple doesn't actually need the best model in the world.

It's similar to Meta. Their model only needs to be good enough for 99% of everyday use cases. They don't even want to compete with Frontier Labs, but primarily reach the consumer market.

And Apple actually has a good chance there. Because a well-adapted Gemini model, based on (3.1/3.5?) and well integrated into the OS, could achieve exactly the use case that many need: AI that simplifies their daily work.\n\nQT @kimmonismus: WWDC 2026 - Apple rents Google's brain to fix Siri. What we can expect:

Apple's keynote today is a software reset built around one admission: it couldn't build frontier AI alone. What to expect:

Gemini-powered Siri, a rebuilt assistant on a custom 1.2T-parameter Google model, ~$1B/year (Gurman). Runs via Private Cloud Compute, no Google training on your queries.
Siri as an app, standalone, iMessage-style chat with synced history, a "Search or Ask" Dynamic Island pop-up, and an Extensions system. Drafts emails, pulls from mail, calendar, contacts and the web.
Six OS betas - iOS 27, iPadOS 27, macOS 27 ("Big Bear"), watchOS 27, tvOS 27, visionOS 27. iOS 27 is a "Snow Leopard" cleanup release. iPhone 11 and SE2 lose support.
Liquid Glass 2.0 - system-wide opacity slider, fixes for the shadow and transparency complaints.
AI health coach - the watered-down "Health+", now fitness and wellness instead of an AI doctor (pretty cool!)
Model choice (rumored), users may pick the engine behind Apple Intelligence, with Gemini and Claude floated.
Hardware mostly later - M5 Macs, new iMac, foldable iPhone (~$2.5K, Sept), OLED touchscreen MacBook Pro, smart-home hub.

Sources: TechInsider, Bloomberg, Gamebezz

See 4 related tweets

@kimmonismus: WWDC 2026 - Apple rents Google's brain to fix Siri. What we can expect:

Apple's keynote today is a ...

@BrianRoemmele: APPLE JUST BROKE DOWN AND SAID THE AI WORD!

WWDC 2026

I met with them in 2012 about the AI world ...

@Techmeme: Apple overhauls Apple Intelligence, with a new architecture built on Apple Foundation Models develop...
@MTSlive: SITUATION EXPLAINED: Apple's new foundation models were created through a deep collaboration with Go...

12. nvidianewsroom (Group Score: 134.3 | Individual: 37.2)

Cluster: 5 tweets | Engagement: 477 (Avg: 379) | Type: Tech

NAVER is building a full-stack NVIDIA AI factory in Korea with NVIDIA DSX.

Jensen met with @official_NAVER founder and chairman Haejin Lee while in Korea as NAVER plans to expand GAK Sejong to 55 megawatts and beyond to gigawatt scale.

See 4 related tweets

@Reuters: Nvidia announced a series of deals in South Korea with tech giants including SK Hynix and Naver, as ...
@NVIDIAAP: RT @NVIDIAAIInfra: 📣 Useful AI has arrived — and NAVER is building the infrastructure to meet it at ...
@Techmeme: Nvidia says South Korea's Naver will use its technology to build AI factories at "gigawatt scale" to...
@ReutersBiz: Nvidia announced a series of deals in South Korea with tech giants including SK Hynix and Naver, as ...

13. cgtwts (Group Score: 125.1 | Individual: 39.8)

Cluster: 4 tweets | Engagement: 75 (Avg: 39) | Type: Tech

“babe wake up.”

Kimi AI just dropped their own OpenClaw. https://t.co/e1IU0kE9vl\n\nQT @Kimi_Moonshot: Meet Kimi Work - a local AI agent on your desktop that does the work for you.

🔹Native agent swarm: Up to 300 AI agents running in parallel on your local machine. 🔹Browser use: Paired with WebBridge extension, your agent will navigate websites in your browser: search, scroll, click, type and complete tasks. 🔹Built for Finance: Native global market data tool call from Yahoo Finance and World Bank - no complex API setup required. 🔹Memory system: Kimi Desktop keeps a running diary of your preferences, past decisions, and context to know you better.

Available for macOS (Apple Silicon) and Windows.

🔗Try it now: https://t.co/yhiai2VWIy

See 3 related tweets

@VaibhavSisinty: Kimi just dropped something wild. A desktop app that runs 300 AI agents on your machine. In parallel...
@codewithimanshu: RT @codewithimanshu: Kimi AI just dropped Kimi Work.

A local AI desktop agent that runs 300 paralle...

@testingcatalog: KIMI 🔥: A new "Kimi for Work" AI Agent has been released with support for Native Agent Swarm, Browse...

14. mark_k (Group Score: 120.7 | Individual: 33.4)

Cluster: 4 tweets | Engagement: 293 (Avg: 205) | Type: Tech

GPT-5.6 has hit the Design Arena. This always happens a few days before release... making a Thursday release quite likely.

@OpenAI might aim for a BIG AI WEEK 🔥\n\nQT @synthwavedd: 🚨 A new anonymous model under the name "Kindle" has been added to the Design Arena, very likely to be the same "kindle-alpha" GPT-5.6 Release Candidate checkpoint previously revealed

As foretold! It's coming https://t.co/JN5Lzce421

See 3 related tweets

@synthwavedd: 🚨 A new anonymous model under the name "Kindle" has been added to the Design Arena, very likely to b...
@AiBattle_: Kindle (GPT-5.6) has been removed from the Arena

A new model, Levi, appeared shortly after Kindle w...

@TeksEdge: GPT-5.6 warming up for released (and during Apple's WWDC '26?)\n\nQT @synthwavedd: 🚨 A new anonymous...

15. rohanpaul_ai (Group Score: 120.6 | Individual: 29.8)

Cluster: 5 tweets | Engagement: 117 (Avg: 67) | Type: Tech

American AI startups are routing far more app traffic to Chinese LLMs.

Through OpenRouter, weekly token consumption was mostly driven by U.S. models through much of 2025. But from early 2026, Chinese models suddenly became the main growth engine.

AI model market is becoming less about brand loyalty and more about raw utility.\n\nQT @nxthompson: This is a pretty striking shift toward Chinese models by American AI startups since the start of the year. https://t.co/80uaCWQCRE https://t.co/4vA8xPQRgE

See 4 related tweets

@RnaudBertrand: Extraordinary chart: Chinese AI models have now completely overtaken their US competitors on OpenRou...
@alex: I think this is mostly due to people who use OR are using it specifically to call Chinese models? (I...
@KyeGomezB: RT @rohanpaul_ai: American AI startups are routing far more app traffic to Chinese LLMs.

Through Op...

@Miles_Brundage: One reason I focus on auditing of American AI companies first, not jumping straight to China

(anot...

16. deanwball (Group Score: 119.9 | Individual: 31.9)

Cluster: 4 tweets | Engagement: 143 (Avg: 106) | Type: Tech

Teams of thousands of Mythos-tier agents burning billions of tokens off of a single prompt should provide the intuition that our conceptual universe can grow profoundly larger. Up to now we have merely automated existing tasks. We have yet to hear the machines roar.\n\nQT @deanwball: Ok, so I used my tweeting about this project—plus a rainy afternoon in Taipei that kept me indoors—to experiment with more sophisticated multi-agent teams stuff.

For context: the goal of this topic is to create book-ish length (50-75k word) histories of arbitrary topics from as little as a single word (e.g. “Taiwan”). This was an early project I did with Opus 4.5 when that model first came out. It was structured largely in the way I thought about AI at the time: there are agents, and then there are chatbots. So: (1) CLI interface that allows me to pass <topic> to an LLM api (2) insert <topic> into pre-written “book outline” prompt (3) as part of the outline prompt, instruct LLM to generate prompts for each topic and subtopic within the outline (4) pass those topic prompts, along with broader context about the overall goal, to LLMs (5) format those outputs into a web UI well-suited for long form text.

But after I tweeted about this old project, I realized that I now think of AI in an importantly different way: teams of agents collaborating rather than isolated agents that use LLMs as part of their software toolkit (though the latter is still a useful modality, tbc, it is not as primary as it used to be for me).

So I experimented with this notion: rather than “agents designing scripts to process information by deterministically bouncing text between LLM APIs,” could I design a digital organization or firm whose job is to produce books I would like on arbitrary topics? A publishing house, just for me?

Fortunately I keep quite extensive records with agents of my reading, favorite quotes from hundreds of books, and my entire library. So reasonably high-quality context for “what I would like” was extant.

As I played around with multi-agent orchestration within the coding agent apps, I found that Claude Code’s dynamic workflows feature (activated by including the word “workflow” in your prompt) had the best ergonomics. It causes the model to produce a script that defines a multi-agent hierarchy, though the complexity can be more like a corporate org chart than a simple hierarchy. The model does this, runs it, and the Claude Code CLI app gives you an elementary UI for overseeing the operations of the autonomous digital firm you have instantiated. It “just works.”

Purely as a nerd, it is incredible to watch 100s of agents go to work. In my case, I went with Opus’ off-the-bat organizational structure at first, which involved panels of fact-checkers, editors, researchers, writers, etc. Later I added additional roles and processes, including a group of “submitters” who were placed in competition with one another to write the best proposals.

It is, at the very least, a hell of a way to spend tokens—a bit more than 10 million, in the case of my refined structure.

The output was FAR better than what I used to get with decentralized LLM calls, but probably not worth a ~two OOM increase in token use. The main flaw was cybernetic in nature: the “virtual firm” still behaves too much like a bunch of isolated agents deterministically passing information between one another.

What you want, in any firm virtual or real, is for information to pass in an almost electrochemical fashion between the firm’s constitutive agents. You don’t want each agent to just have context for “the broad project as we conceived of it when we started” but rather the work as it evolves in real time. Eureka moments over here need to make their way over there; a misconception baked into the original proposal needs to be flagged and actioned in a way that is legible to all agents. Very little of this happens, and it is not easy to do as of yet without creating, essentially, layers of bureaucracy which increase costs (ha!).

In short, the way it all works is not very deep-learning-pilled. It is what Hayek would call rationalistic. Perhaps @polynoamial’s research program will push us to deep-learning-pilled, electrochemically linked multi-agent architectures.

See 3 related tweets

@deanwball: Ok, so I used my tweeting about this project—plus a rainy afternoon in Taipei that kept me indoors—t...
@deanwball: There are two main approaches, I think, to resolving the cybernetic problem I identify below. One I ...
@deanwball: tweeting about this caused me to think about the structure of this project, which was one of the fir...

17. wallstengine (Group Score: 117.7 | Individual: 40.6)

Cluster: 4 tweets | Engagement: 603 (Avg: 127) | Type: Tech

$AMZN announced a multibillion-dollar agreement with$ GLW to supply optical fiber, cable, and connectivity solutions for Amazon’s U.S. data center infrastructure.

The deal will create 1,000 advanced manufacturing jobs at Corning’s North Carolina facilities and support hundreds of construction jobs.

Amazon says the agreement strengthens the U.S. fiber optics supply chain as AI data center demand expands.

See 3 related tweets

@StockMKTNewz: AMAZON $AMZN JUST SIGNED A MULTIBILLION-DOLLAR DEAL WITH CORNING$ GLW

Corning will supply the optic...

@ShanuMathew93: And told you there will be more multi-year framework agreements across the stack... chips, semicap, ...
@StockSavvyShay: $AMZN announced a multibillion-dollar deal with$ GLW to supply optical fiber, cable and connectivity...

18. NVIDIARobotics (Group Score: 115.4 | Individual: 30.6)

Cluster: 5 tweets | Engagement: 71 (Avg: 238) | Type: Tech

As part of LG Group's new AI factory collaboration, @LGUS is building home robots like CLoiD, trained and validated in NVIDIA Isaac Sim and Isaac Lab, and exploring Isaac GR00T for humanlike reasoning and complex task execution. 🤖

Across its businesses, LG is leveraging Isaac and Cosmos to advance physical AI across home, manufacturing and logistics. ⬇️\n\nQT @NVIDIAAIInfra: 📣 LG Group and NVIDIA are building an AI factory based on the NVIDIA DSX platform.

The AI factory will provide LG Group with accelerated computing infrastructure to train, simulate, validate and deploy AI-based applications across robotics, autonomous driving, data center technologies and GPU cloud services.

Learn more about the collaboration: https://t.co/eziUcVcKcR

See 4 related tweets

@nvidianewsroom: NVIDIA and LG Group announced plans to build an AI factory to support LG’s robotics, autonomous driv...
@Techmeme: Nvidia and LG expand their partnership in AI, robotics, and mobility tech, including building an AI ...
@nvidia: RT @NVIDIAAIInfra: 📣 LG Group and NVIDIA are building an AI factory based on the NVIDIA DSX platform...
@FirstSquawk: Nvidia, LG Collaborating on Robotics, Motors and Data Center Development, CEO Says...

19. VaibhavSisinty (Group Score: 112.9 | Individual: 32.5)

Cluster: 6 tweets | Engagement: 15 (Avg: 87) | Type: Tech

$1,000,000,000 a year. That's what Apple now pays Google to power the new Siri. 🤯

At WWDC today, Apple unveiled the biggest Siri rebuild in its history.

The surprise wasn't the features. It was what's running underneath: a custom Google Gemini model, reportedly around a billion a year for 1.2 trillion parameters.

That's a genuinely big deal, because Apple's whole identity has always been control. They design the chip, the OS, the hardware, the services, the entire stack. Owning every layer is the Apple way.

So choosing to build Siri's brain on a partner's model is a real shift in philosophy. Apple promised this version of Siri back in 2024 and spent two years trying to ship it.

Rather than keep waiting on a fully in-house model, they made the call to partner and get a great product into people's hands now.

And the result looks strong. On-screen awareness, personal context across your messages and photos, multi-step actions across apps, all wrapped in Apple's privacy-first architecture so your data stays yours.

If the new Siri delivers the way the demo suggests, nobody will care what's running underneath.

They'll just notice their iPhone finally got smart. And that's the bet Apple just made.

See 5 related tweets

@markgurman: While there was nothing revolutionary, Apple just rebooted the foundation of its platforms with fun...
@TechCrunch: Meet the revamped Siri…Apple’s new effort to catch up with the AI competition and bring its longstan...
@WIRED: From a standalone app to a Google Gemini partnership, here’s everything you need to know from WWDC 2...
@SeijinJung: apple is keynoting WWDC today.

siri is reportedly rebuilt on Google Gemini, processed through Apple...

@SeijinJung: today is WWDC. apple is finally unveiling the new siri.

it's powered by google gemini.

two years o...

20. alexcooldev (Group Score: 111.2 | Individual: 32.8)

Cluster: 4 tweets | Engagement: 171 (Avg: 142) | Type: Tech

$20k MRR is a long road.

When you’re scrolling X, it feels like everyone is doing $100k,$ 500k, or even $1M MRR.

What you don’t see:

The years of failure The products that never worked The sleepless nights The money they lost before making money

If you’ve reached $20k MRR, be proud of it.

That’s already life-changing for most people on the planet.

Don’t compare your Chapter 3 to someone else’s Chapter 20.

Everyone has a different starting point, different advantages, and different goals.

Success isn’t a race.

Enjoy the journey.

Build at your own pace.

And remember: the version of you from 3 years ago would be amazed by where you are today.\n\nQT @marclou: I documented my SaaS journey to $20K MRR.

It took 512 days.

I got excited, I cried, I laughed, I lost hope. I got literally every moment on camera.

This video is the raw story of what building a startup actually feels like. https://t.co/vBIJELqrLl

See 3 related tweets

@seraleev: Had to delay the start of my workday by 30 minutes just to watch this video.

Honestly, I’m a little...

@jackfriks: this guy deserves $100k MRR for this video alone\n\nQT @marclou: I documented my SaaS journey to$ 20...
@jackfriks: RT @marclou: I documented my SaaS journey to $20K MRR.

It took 512 days.

I got excited, I cried, I...