📡 SIGNAL, NOT NOISE. This page is for people who want to understand where the world is going before it becomes obvious. I cover the forces reshaping markets, technology and society: ₿ Bitcoin & digital assets 🤖 AI & autonomous agents 🧠 Semiconductors & compute ⚡ Energy & infrastructure 🏦 Finance & macroeconomics 🌍 Geopolitics & global power No recycled hype. No blind optimism. No permanent doom. Just the important developments, the numbers behind them, and what they could mean next. I’m especially interested in the intersections: AI × Energy AI × Finance Bitcoin × Monetary Policy Technology × Geopolitics Automation × Work Blockchain × Traditional Finance Because the biggest opportunities—and the biggest risks—usually appear between categories. 📍 Follow for the signal. Question everything. Verify the data. Think independently. The future rarely arrives as one big announcement. It arrives as a series of signals. #AI #Bitcoin #Crypto #Technology #Markets #Finance #FutureTech #Geopolitics

The best part about Chroma Season 2? > $450K+ in cash prizes > $1M in free trials & promotions > Unlimited entries Experiment, try something ridiculous, learn from it, and submit something completely different the next day. how creative competitions should work my friends 👊

CAD DESIGN. BUT UP TO 125X FASTER 🤯 and WAY cheaper, too! @skyfallai's new SKY-001 breaks complex objects into components, sketches each part, reconstructs everything in 3D. 18 mths in the making, built by a team from Maluuba's deep learning lab (acq'd by Microsoft Research)

Last week, these guys jumped out of a plane to promote Pancake.ai. This week? They've stacked 100+ of SF's top GTM engineers into a one-click follow list. I NEED TO SIT IN ON ONE OF THEIR MARKETING MEETINGS 😂

WAIT... YOUR GROUP CHAT CAN BUILD SOFTWARE NOW 🤯 Wabi generates interactive mini-apps inside conversations, alongside AI agents that help get things done. → School pickup changes? Update the family calendar. → Planning a trip? Pick dates, build the itinerary and book it. → Fancy a show? Find a night that works and get tickets. Your people, the apps and the agent, all in one thread. you bet i’m following this 👀

THEY SMASHED A MAC MINI. IT WORKED. 2.2M VIEWS. 3,000 DOWNLOADS. But AgentCloak's launch wasn't just a stunt. Step 1: Big creators sparked the buzz Step 2: Smaller accounts kept it going Step 3: The founder's network reached buyers Reach is nice. Relevance converts 💪

🚨🟢 MULTICOIN CAPITAL JUST MADE HYPE ITS LARGEST HEDGE-FUND POSITION One of crypto’s best-known investment firms is making a major conviction bet on Hyperliquid. Multicoin Capital says $HYPE is now the single largest position in its hedge fund. The firm accumulated its exposure through public markets rather than through a disclosed private allocation. Multicoin has not revealed exactly how much HYPE it owns. But making it the fund’s largest position sends a clear signal about where the firm sees asymmetric opportunity. And there is another important part of the story. While many venture firms are expanding aggressively into artificial intelligence, Multicoin says it sees no reason to chase the AI investment boom. It is choosing to stay specialised in crypto. 📊 WHY THIS NEWS IS USEFUL The important takeaway is not simply that a fund bought HYPE. It is what Multicoin appears to be underwriting. Hyperliquid has built one of crypto’s strongest vertically integrated trading ecosystems, spanning: • perpetual futures • spot markets • on-chain liquidity • its own execution infrastructure • the HYPE token economy That means HYPE increasingly represents exposure to the economics of an entire trading network rather than a single isolated DeFi application. For investors, Multicoin’s positioning is useful because it shows where a specialist institutional investor is concentrating capital while much of the broader venture market is rotating toward AI. 🔎 WHAT TO WATCH NEXT Watch: • Hyperliquid trading volumes • protocol revenue and fee generation • HYPE token supply and unlocks • competition from centralised exchanges • regulatory pressure on perpetual trading • whether institutional participation keeps increasing The strongest signal would be continued growth in Hyperliquid’s underlying business metrics while HYPE remains central to the ecosystem. The weaker signal would be a rising token price without corresponding growth in users, liquidity, volumes and economic activity. The deeper story: Multicoin is not simply betting on another crypto token. It is betting that on-chain financial markets themselves can become one of crypto’s dominant businesses. #HYPE #Hyperliquid #MulticoinCapital #Crypto #DeFi #Altcoins #Blockchain #CryptoNews

🚨📊 FXS JUST COLLAPSED 68% IN ONE HOUR — AND THERE IS STILL NO CONFIRMED CATALYST Frax Finance’s legacy FXS token suffered an extreme intraday move, falling 68.26% within 60 minutes to around $0.25776. The token had traded as high as roughly $0.812 during the broader 24-hour window. So far, there is no confirmed explanation for the sudden collapse. No major exploit has been identified. No protocol failure has been confirmed. No single liquidation event or large-wallet sale has yet been established as the cause. 📊 WHY THIS NEWS IS USEFUL A move this violent is a reminder that not every crypto crash means the underlying protocol itself has failed. FXS is also no longer Frax’s primary active asset. Since 2025, holders have been able to upgrade legacy FXS 1:1 into FRAX, the native asset of the Fraxtal ecosystem. That matters because legacy or migrating tokens can develop: • thinner liquidity • fragmented trading venues • wider spreads • unusual price discrepancies • exaggerated volatility during large orders A 68% one-hour move can therefore be caused by market structure just as easily as by fundamental news. 🔎 WHAT TRADERS SHOULD CHECK BEFORE REACTING Before treating a collapse like this as a protocol disaster, verify: • the exact token contract • whether you are looking at legacy FXS or the newer FRAX asset • liquidity on the specific exchange or DEX • whether other venues show the same price move • large-wallet transfers • liquidation data • exploit announcements • bridge or migration activity The most useful takeaway: Price action tells you something happened. It does not tell you why. Until the catalyst is confirmed, this should be treated as an extreme liquidity and market-structure event — not automatically as proof that Frax itself has broken. #Frax #FXS #FRAX #DeFi #Crypto #Altcoins #CryptoNews #Trading

🚨🇺🇸🇨🇳 FORMER U.S. TREASURY CHIEF CALLS FOR U.S.–CHINA “AI ARMS CONTROL” TALKS Former U.S. Treasury Secretary Robert Rubin says Washington and Beijing should cooperate on the most dangerous AI risks even while remaining strategic competitors. His model? Cold War nuclear arms control. Rubin pointed to the Strategic Arms Limitation Talks between the United States and Soviet Union, arguing that rival powers were able to negotiate safeguards even amid intense distrust. His message is straightforward: The U.S. and China do not need to trust each other completely. They need mechanisms that reduce the chance of catastrophic AI failures affecting both sides. 📊 WHY THIS NEWS IS USEFUL AI competition is increasingly becoming a national-security issue. The risks are no longer limited to who builds the best chatbot. Potential flashpoints include: • AI-controlled military systems • autonomous weapons • AI-assisted cyberattacks • accidental escalation • interference with critical infrastructure • biological and chemical misuse • systems acting without adequate human control The problem is that the world’s two most important AI powers still lack anything resembling the mature crisis-management architecture built around nuclear weapons. That is where Rubin’s Cold War comparison becomes useful. A U.S.–China framework could potentially include: AI hotlines Direct communication during serious AI-related incidents. Human-control guarantees Rules preventing AI systems from independently making nuclear or other catastrophic military decisions. Frontier-system testing Shared expectations around safety evaluation before highly capable models are deployed. Incident notification Mechanisms for informing the other side when a major AI system behaves dangerously. Red lines Agreement on specific AI uses that both governments consider unacceptable. The important point is that cooperation would not require the U.S. or China to share proprietary models, chips or strategic technology. They could continue competing while still establishing rules around catastrophic risk. 🔎 WHAT TO WATCH NEXT Watch whether AI safety becomes part of formal U.S.–China diplomatic talks rather than remaining primarily a technology-policy discussion. Also watch for: • military-to-military AI communication channels • agreements covering autonomous weapons • joint language on nuclear command systems • frontier-model safety standards • crisis-notification mechanisms • involvement from companies such as OpenAI, Google, DeepSeek and Alibaba The deeper question is becoming difficult to avoid: Can the two countries competing hardest to dominate AI also cooperate enough to prevent the technology from creating risks neither side can control? The Cold War suggests rivalry and risk reduction do not have to be mutually exclusive. #AI #China #USA #ArtificialIntelligence #Geopolitics #NationalSecurity #Technology #CyberSecurity

THE NUMBERS ARE GETTING RIDICULOUS Polsia's reported run-rate milestones: → $1M after 12 weeks → $10M another 10 weeks later → $20M just 6 weeks after that → Approaching $25M, with $100M ARR as the next ambition Growth that speeds up at each milestone is worth paying attention to 👀 Now the real question: can Polsia v2 turn that momentum into something sustainable?

Building open-source AI in Europe? This one's for you! @nebiusai really cooked with its European AI Builder Program expansion 🔥 $10M across compute grants, capital and community. Compute grants support the tools builders need, from open models and agent runtimes to evals and inference tooling. Europe has the talent. Now let's give promising projects room to grow 👊

WAIT.. 1¢ PER PROMPT TO FIND OUT WHY AI RECOMMENDS YOUR COMPETITORS?! 🚨 @treg_ai’s new open-source AEO toolkit shows you the pages behind those answers and the people who wrote them: → see which sources AI cites → find the people behind them → draft outreach to get your product considered .. and more! run it from Claude Code. or Codex. Self-hostable. No subs! Best part? Open source. Full code on GitHub. Repo in 🧵↓

🚨 Dreamina's $1.50 Basic offer is somehow still live. It gives you a full month to turn saved ideas into actual videos with the Seedance lineup, including Seedance 2.5. ICYMI the first time, consider this the bonus round 🧵↓ → dreamina.capcut.com/ai-tool/… #DreaminaPartner

🔥 BIG: Circle CEO Jeremy Allaire says the USDC issuer is eyeing around 40 new developing markets across Southeast Asia, Latin America, Africa and the Middle East as stablecoins hit the "mainstream scaling phase." cointelegraph.com/news/circl…

🚨 OPENUI JUST HIT 1.0 FOR OPEN SOURCE AGENT UI 6 months ago, @openuidev open sourced a framework to get AI agents out of the text box. 10K+ GitHub stars and 6M+ downloads later, OpenUI 1.0 is officially live 🎉 It works with any model and lets your agent answer with real product components instead of dumping paragraphs of text into a chat window. Ask an agent to plan a trip, break down metrics, or configure a workflow, .. and it can render: > a live map > a chart view > a detailed form with editable dropdowns with inputs ready to submit ... you name it! eg here's what that looks like inside their Formula 1 demo. Someone tells the F1 agent "Russell is Mercedes' real number one." Instead of writing a 4 paragraph argument, the agent opens the pit wall: → head to head card, Antonelli vs Russell → the points gap, charted race by race → how many Grands Prix are left to close it The answer is the dashboard. FREE and open-source. Repo in 🧵↓

THE MINDS BEHIND GEMINI AND APPLE'S MM1 JUST TOOK VISUAL AI A STEP FURTHER Most AI models can analyze an image. But what if they could keep looking, zooming in, checking details, and rethinking their answers? 🚨 Meet @ElorianAI Glimpse, a new model built to actually understand what it sees. Give it an image or video, ask a question, and it reasons through what it sees. → reads complex charts → handles tricky visual counting → understands humor in videos → turns robot footage into actionable steps Built for understanding, not generation. Think robotics, factories, construction sites, and everything AI still struggles to see 👀

WAIT.. ROBOTS ARE GETTING PERSONALITIES NOW 😱 Meet Hugo from Dynamic Creatures. A character robot that connects vision and sound to gestures, movement and expression. Not just walking or dancing. The ambition is to react to YOU in real time. I freaking love this.

🚨 @OpenBMB just dropped MiniCPM-V 4.7, and the engineering behind it is seriously impressive. 35B total parameters, ~3B active at inference, and ~3GB memory in its target on-device configuration. Here's what makes it work: → LES predicts which experts the next Transformer layer needs and preloads their weights to reduce memory latency. Misses get rescheduled to maintain parallelism. → DSpark Speculative Decoding drafts tokens ahead of the main model, which verifies them. Same output, faster execution. → Visual Token Compression cuts image and video token overhead by 4x and 16x. That last one makes a real difference. So I built something to test it 👇 🎬 Clip Narrator: drop in ANY video clip and get back a fully narrated version. My test footage? Literally 10 seconds of milk heating in a pot. The 35B-A3B sampled 8 frames, compressed the clip into ~2,500 visual tokens, and generated a timestamped movie-trailer script: "A pot of milk heats up." "Steam rises. The stakes are high." "It boils over! Disaster strikes!" A TTS model brought it to life, generating 8 seconds of audio in just 2.6 seconds. Subtitles synced automatically. Export and done. Here's what impressed me: it wasn't just captioning 8 still frames. It tracked the milk rising, steam building, and boil-over, timing each line to the action. That's temporal reasoning, not just image description! A 1.3B variant also handles OCR, charts, math, images and video. And both support 256K context, with fully open-source weights, code and fine-tuning pipelines. Beyond scaling parameters, it's about smarter execution on consumer hardware. Try it locally and tell me the most boring clip you'd feed it first 👇 FREE and open-source. Repo link in 🧵↓

🔥 HUGE: Anthropic is reportedly paying Blackstone over 10% yield on $9B of debt backed by its chip leases, beating most of DeFi with an AI lab as the borrower.

🚨⏱️ THE WORLD IS SCRAPPING THE LEAP SECOND — AND UTC WILL RUN CONTINUOUSLY FROM MAY 2027 One of the strangest pieces of global infrastructure is being retired. Since 1972, timekeepers have occasionally added a leap second to Coordinated Universal Time (UTC) to keep ultra-precise atomic clocks aligned with Earth’s slightly irregular rotation. That ends on 20 May 2027. The General Conference on Weights and Measures has voted to let UTC run continuously instead. Under the old system, UTC was kept within 0.9 seconds of astronomical time. Under the new system, the difference can eventually grow as large as one hour before a larger correction is required. 📊 WHY THIS NEWS IS USEFUL A single second sounds trivial. For computers, it isn’t. Modern infrastructure depends on machines agreeing precisely on what time it is. Leap seconds have historically created problems for: • cloud servers • telecom networks • financial trading systems • databases • GPS-related infrastructure • power grids • distributed software systems Different companies have also handled leap seconds in different ways, creating inconsistencies between systems. The risk became more urgent because Earth has recently been rotating fast enough that scientists may eventually need a negative leap second — removing a second rather than adding one. That has never been done before and many systems were never designed for it. 🔎 WHAT CHANGES FOR YOU? Probably almost nothing. Phones, computers and clocks will continue working normally. The real impact is behind the scenes. UTC becomes a cleaner, continuous time standard for the digital world, while astronomers and geodesists will separately account for the growing difference between atomic time and Earth’s actual rotation. The deeper takeaway: Modern civilisation increasingly needs machine time to be predictable more than it needs clock time to perfectly follow the spinning Earth. That is why a 55-year-old system is finally being retired. #Technology #UTC #Science #Time #CloudComputing #GPS #Finance #Telecoms

🚨💻 META’S NEW BUSINESS TERMS COULD LET IT SCRAPE NEWS SITES FOR AI TRAINING A major shift in the AI-versus-publishers battle is coming on October 30. Under Meta’s updated commercial terms, businesses using products such as Facebook, Instagram, WhatsApp and Messenger authorize Meta to use automated tools including crawlers, bots and web scrapers to access websites associated with their products or services. And the language goes further. Content collected from those sites may be used to: • develop Meta products • improve them • research new systems • train AI models That potentially puts news publishers in a difficult position. Many publishers rely heavily on Meta platforms for audience distribution and business tools. But continuing to use those commercial accounts may now also mean granting Meta broad permission to crawl the publisher’s external website. 📊 WHY THIS NEWS IS USEFUL This could change the economics of AI content licensing. Other AI companies have spent significant sums signing licensing agreements with major publishers. If Meta can rely instead on permissions embedded inside its business terms, publishers could lose some of their negotiating leverage. That creates a much bigger question: Can access to a dominant distribution platform effectively become permission to use a company’s content for AI training? There are still legal uncertainties. Meta says website owners can manage crawler access through robots.txt. But legal experts have questioned whether accepting Meta’s commercial terms could give Meta a separate contractual basis to crawl the same content. 🔎 WHAT TO WATCH NEXT Watch for: • publishers closing or restructuring Meta business accounts • legal challenges to the new terms • changes to robots.txt policies • new publisher licensing agreements • regulators examining platform power and contractual consent • whether Meta actually expands its AI training datasets using this authority The deeper issue is not simply web scraping. It is who owns the economic value of online content once that content becomes training material for AI. If this model survives legal scrutiny, platform terms of service could become one of the most powerful tools in the AI data race. #Meta #AI #ArtificialIntelligence #Publishing #NewsMedia #Copyright #Facebook #Instagram #TechNews