AI compute demand is rising, and the numbers show the infrastructure side is still highly centralized. + traditional cloud providers control most of the capacity + building new data centers takes years and significant capital + smartphones already exist in the billions and sit idle most of the time ➜ the gap between available consumer hardware and usable verifiable compute is still large. that’s the setup for @Acurast, a decentralized compute network that turns ordinary smartphones into compute nodes. how it works: + workloads run inside the phone’s Trusted Execution Environment + execution is confidential and verifiable without trusting the device owner + developers deploy serverless jobs, APIs, or LLM inference + phone owners earn $ACU by contributing capacity $ACU is used to pay for compute, reward providers, and participate in the network. providers can run the Processor app on a current phone or a spare Android device. ➜ in short: Acurast matches demand for verifiable confidential compute with hardware people already own. take a look: https://t.co/HmTcUOEegJ

The easy $hype trade looks done for average retail. ICE backing. Druckenmiller in. Elon tweeting. The story is already everywhere. Buying now means paying for us growth that still needs CFTC approval, a real launch, and actual us users showing up. Even if hyperliquid pulls it off, how much of that is already priced in. The extra fees and buybacks have to beat what everyone is already expecting. all of this can be good for the platform without making @HyperliquidX an obvious buy at any price. The next leg will be harder to earn than the last. If hype goes above $97, i’m wrong.

📝 @Acurast has been quietly running real workloads for a while now. Not just simple test jobs. Developers have been exploring a broader range of use cases on the network: > confidential LLM inference > scheduled jobs & webhooks > agentic workflows > data processing that needs both privacy and verifiability This is the part I find most interesting about phone-based compute: Anyone can talk about decentralized infrastructure. Actually running code inside a phone’s TEE and getting verifiable results is a different level. But there’s also a pretty accessible entry point even if you’re not deploying workloads yet. You can just run the Processor app on a spare phone, contribute compute, and earn $ACU while the network handles the rest. No need to write code or manage servers. I like this mechanic because it lets regular users participate in the supply side instead of only being consumers of compute. I’ve been watching how the number of active phones and real deployments keeps growing. Worth checking the current stats and recent demos if you haven’t looked in a while. https://t.co/HmTcUOEegJ

$ENA weekly is building a clear accumulation base after the multi-year downtrend. Price is sitting right on the major support zone ($0.10–$0.19). Classic rounded bottom forming with the first higher lows. As long as this zone holds on weekly closes, smart money is accumulating here. Break above opens the path toward $0.50–$0.80+. Lose $0.10 with conviction → one more flush lower. Right now the structure looks constructive - closer to a local bottom than another leg down. Simple: DCA @ethena this zone and let HTF play out.

📝 @tebrabbit built Thumper for @magicmarkets: a free browser extension that shows when Polymarket isn’t the cheapest place to take a sports position. Quick context: @Polymarket sports prices look clean, but a taker fee lands on top, so the 41¢ you see isn’t the all-in number. The same game is often priced differently on MagicMarkets, and almost nobody checks both. The deal: + drops the @MagicMarkets price into the Polymarket trade box, fee counted on both sides + “better odds” tag when MagicMarkets is cheaper, one click to the same market + side panel: best price, price-move insights, open Polymarket positions re-priced vs MagicMarkets + optional alerts when MagicMarkets is 2¢+ cheaper on a game starting soon Why it fits: sports traders already live on Polymarket. Thumper keeps them there while surface-checking the real all-in price elsewhere. ➜ stops the silent fee gap from eating the edge ➜ turns “I thought 41¢ was the price” into a live comparison before the click 2¢ doesn’t matter once. It matters every weekend. watching whether EPL and in-play users actually switch when the tag shows up. that’s the number that matters. Try here: https://t.co/9iCD9Qki9e #MagicMarkets #Prediction #PredictioMarkets

🧠 Started accumulating $ONDO. Here's the thesis: 1. Position. Sitting around $0.47, still ~78% off the $2.14 ATH. Not trying to nail the bottom, just stacking a fixed amount every week. Dips give me more units, rallies slowly lift the cost basis. Simple. 2. Real product traction. Just crossed $3B in tokenized U.S. Treasuries and ranked #1 on https://t.co/wJTMZqAoEs. USDY alone is over $2.3B and already bigger than BlackRock’s BUIDL. Ondo Stocks is leading the tokenized equity market. This isn’t pure narrative anymore - actual assets are moving. 3. Institutional bid is real. IMF just named @Ondo as one of the largest tokenization platforms in their latest GFSR. Live integrations and partnerships with BlackRock, Franklin Templeton, JPMorgan, Mastercard, etc. These names don’t usually touch dead projects. 4. Unlikely to go to zero. One of the more established names in RWA with real AUM, multi-chain presence, and actual products. Worst case is slower growth or losing some market share. Doesn’t vanish overnight. Assets with a visible floor are better for time-based accumulation. 5. Catalysts are stacking. Ondo Private Markets just launched (starting with pre-IPO AI exposure), Intelligent Portfolios rolling out, and the broader RWA narrative is still early. Not trading the news, but building while things are happening feels better than sitting idle. Risks are there though. Still mainly a governance token so it doesn’t automatically capture fees from USDY or Ondo Stocks. Unlocks still ahead, competition is rising, and price remains volatile. Only using spare capital, fixed amounts, no leverage, planning to hold 1-2 years. #ondo #rwa #rwas

I actually think Perturb’s approach makes sense, but not for the usual “decentralized AI” reasons people throw around. AI security is becoming a much bigger problem as models get more capable and agents start operating with less human oversight. And this is where @perturbaix has a structural advantage. Most red teaming is still done in-house or with static open-source tools. Perturb turns adversarial attack discovery into a continuous, incentivized competition, then uses those findings for actual model training. The problem is that finding attacks and turning them into something companies will pay for are two different things. Even if miners keep producing strong adversarial examples, the real question is whether enterprises will treat the resulting scores, datasets, and certificates as credible enough to act on. Moving into a real product with continuous scanning + hardening could give them more control over that distribution layer. We’re also seeing the subnet mature (scanning + training live, product phase starting). But I wouldn’t confuse subnet activity with proof that the commercial model works yet. Leaving pure research mode means Perturb now has to prove it can generate demand and revenue without just relying on emissions. My view is that continuous, external adversarial pressure could become just as important as model performance over the next few years. If Perturb can productize that without the output remaining mostly internal to the subnet, the direction is justified. Otherwise, it risks staying an interesting research network rather than infrastructure. SN26 at these levels is interesting either way. #Bittensor $TAO #Perturb #AI

📝Perturb has been running continuous adversarial scanning and training for months. It might be one of the more practical AI security experiments on Bittensor so far. @perturbaix went live earlier this year and has already moved past pure attack discovery. According to recent updates: - Adversarial scanning and training tracks both live - Miners can now fine-tune models using the attacks they generate - Product phase officially starting after ~5 months of subnet work - Robustness scores, datasets, and certificates as the core outputs – Alpha buybacks and owner lockups already happening – Targeting a commercial launch in the coming weeks What’s driving the approach? Perturb turns finding model failures into an incentivized competition, then uses those verified adversarial examples for actual hardening instead of just publishing a report and stopping. The idea is continuous external pressure rather than one-off audits. I can see why teams working on production AI might pay attention. However, the setup still deserves a closer look. Subnet activity and attack success rates don’t automatically mean enterprises will buy the product, and scaling beyond current vision models will take real work. Also, emissions and early traction don’t guarantee sustainable revenue once the commercial layer is live. I’m keeping an eye on how the product launch actually converts. If Perturb can turn continuous adversarial data into something companies reliably pay for and use, this could become a meaningful experiment in decentralized AI security. #Bittensor $TAO #Perturb

📝 AlphaRidge is starting to turn the usual trading stack into one workflow. Most traders still juggle charts in one tab, a screener in another, and research somewhere else. AlphaRidge puts charting, a whole-market screener (~4,500 US stocks + crypto), RidgeScript (Pine-compatible), and an AI analyst into a single login, with Pro at $99/month versus the ~$289 stack many people currently pay for. The AI analyst writes screens, marks up charts, and shows the platform data it pulled so you can check the work. Paper trading is free today. Live brokerage execution is still rolling out. > Another notable piece is the news layer. SN45 miners process roughly 290,000 financial articles a day from ~13,000 sources, turning them into screenable sentiment, impact scores, and event clusters. That data shows up as news marks on charts and as screener columns, so users can filter the market by what is being said, not just by price and volume. The product is still early, but both sides of the loop are moving: research and screening → rules that run in the background → alerts even with the browser closed → more reasons to stay inside the platform. @alpharidge_ai #SN45 #Bittensor $TAO

Looks like @say_gm_ is starting to get quoted more for SayGTM 👀 The Proof of Customer post is going around right as they opened the affiliate pool. The part I care about isn’t really the referral pitch. But how they’re trying to tie SN28 alpha directly to paying usage. Referred top-up → share of that round’s revenue → same share of the emissions pool, capped at $1 of alpha per $1 spent in the first 30 days. So if the model works, this slice of emissions actually depends on whether SayGM can keep bringing developers who pay, not just sign up. That’s more interesting to me than a referral link paying 3% of nothing. Still early though. The rate drops once referred spend outgrows the pool, and the alpha still moves with the market after close. #Bittensor $TAO #SayGM #SN28