The next 6 to 12 months are the ones where you change your life. But, If you don't understand this little concept on how to prepare yourself for the next bull run, then forget about a 100x.... 🧵

people are not yet paying attention to $FLOKI yet. wait until $CALI gets the full attention it truly deserves. time will tell. $FLOKI

. @NEARProtocol has been shipping across several fronts, but the pieces are starting to fit together. → @near_intents for cross-chain execution → Confidential Intents for private transactions → near.com as the user-facing layer → @near_ai + @IronClawAI for secure agents → NEAR One for execution infrastructure An agent that can participate in an economy needs more than a model. It needs to move assets across chains, execute transactions privately, access protected computation, manage credentials and permissions, and operate on infrastructure that can handle those actions at scale. NEAR is building several of those pieces in parallel. NEAR Intents has processed $30B+ in cumulative volume, with record days above $300M and weekly volume reaching roughly $800M–$1B+. It now spans 30+ chains and 125+ assets, with integrations including Solflare. Confidential Intents adds private execution to that transaction layer. TVL has crossed $70M+, while confidential perpetuals are live on near.com through @HyperliquidX. Confidential limit orders are also available through near.com and the 1Click Swap API. Then there is near.com, which is becoming a single interface for swaps, transfers, staking, perps and tokenized assets across multiple chains. @Ondo has also brought tokenized U.S. stocks and ETFs to near.com, including $NVDA, $TSLA, $AAPL, $GOOGL, $META, $MSFT, $AMZN, $SPY and $QQQ, subject to jurisdiction. On the AI side, NEAR AI is building confidential inference around TEE infrastructure and attestation. IronClaw adds secure agent infrastructure, while @AskVenice uses encrypted AI inference. The base layer is evolving too. NEAR 2.13 introduced dynamic resharding and NIST-approved ML-DSA post-quantum signing. Chain Signatures extend cross-network asset interactions, while SPICE targets faster blocks and a separation between consensus and execution. So the stack increasingly looks like: Assets → Intents → Private execution → AI agents → Settlement That is the foundation of the agent economy thesis. Cross-chain finance gives agents access to assets. Privacy gives them room to operate. AI gives them the ability to act. And the base layer provides the infrastructure underneath it. The next phase includes B2B Earn, prediction markets, Agent Market 2.0, AI portfolio management, agents embedded into near.com, tokenized equities and Treasuries, SPICE and faster blocks. The opportunity now is making these pieces work together: Cross-chain liquidity → private execution → tokenized assets → AI agents → scalable settlement.

Public blockchains are great at settlement. They’re not great at keeping your financial activity private. Every deposit, transfer, vault position, rebalance, and balance can leave a permanent public trail. For a normal wallet, that may not matter much. For a treasury, trading desk, or autonomous agent managing capital, it can reveal position size, cost basis, and even strategy. That is the problem Confidential Finance is trying to solve. The goal isn’t to hide everything. It’s selective privacy. Keep addresses and transactions verifiable on a public chain, while keeping sensitive information like amounts, balances, and strategy size encrypted. And importantly, institutions still need a way to prove reserves, satisfy audits, or take action when required by law. That’s where @primus_labs comes in. Primus, formerly PADO, isn’t building another L1. It is building a privacy and verification layer around existing chains, using zkTLS to prove off-chain data without exposing the underlying information, and FHE / zkFHE to perform computations on encrypted data. The idea is simple: Keep the chain public. Keep the financial book private. The first concrete implementation is already live on BNB Chain with Unitas and United Stables. ✦ The flow looks like this: → Shield a stablecoin into an encrypted version → Transfer it without revealing the amount or updated balance → Stake it into a confidential vault while keeping the position private → Unshield or redeem when you want the normal asset back Your wallet remains visible on the public network. Your dollar balance doesn’t. That makes this different from a fully private chain, where liquidity has to move somewhere else, and different from a shielded pool that hides more of the transaction graph. The technical side is also interesting. @primus_labs says it built its FHE stack in-house, including the scheme, AutoHoG compiler, and GPU engine. Its reported benchmarks show roughly 90ms per confidential transfer and 210+ TPS on an 8× RTX 5090 setup. FHE is still more expensive than normal computation, but performance like this is what starts moving confidential finance from research into usable infrastructure. There’s another important piece: proof of reserves. Primus is using zkTLS to let products such as Unitas prove backing without simply publishing their entire financial data set. That combination matters. Privacy without a way to prove solvency is difficult for institutions. Transparency without privacy exposes the entire book. You need both. ✦ The bigger question is what happens next. Can confidential finance support credit, agent execution, automated rebalancing, and cross-protocol strategies without leaking sensitive information at the edges? That is where Primus gets interesting. Its broader stack combines zkTLS for Web2 and agent proofs with FHE for encrypted computation. The $BNB vault is simply the first place where that thesis meets real stablecoins and real yield. If public chains are going to handle more treasury capital, payroll, and autonomous agents, they may need something in between full transparency and a completely private chain. Not a darker blockchain. A public blockchain that doesn’t have to publish your entire financial book.

1/ Huge respect for Dragonfly continuing to invest in the bear market. But I'm morally obliged to call out sketchy projects gaining momentum. As a community we must preempt large scale implosions like #FTX or #LUNA that set the industry back years Bitget may be the next FTX 🧵

lately, have been buying ethereum:0xcf0c122c6b73ff809c693db761e7baebe62b6a2e on every dip after @SabreEthereum last AMA out of all the top breakout MEMES of 2021 bull-run $FLOKI still has one of the based community . survived the last cycle building: Valhalla, European ETP, staking, access to a global exchange and a MASSIVE community. fully PUMPED in on $FLOKI 💪🏽

Upbit might be quietly building one of the more interesting exchange-backed L2 ecosystems. A lot of people seem to be overlooking GIWA. GIWA Global Infrastructure for Web3 Access is Dunamu/Upbit's Ethereum L2, built on the OP Stack. The setup is pretty interesting: → 1-second block times → EVM compatible → Upbit/Dunamu operates the sequencer → Optimism acts as the backup → ETH is used for gas → No official GIWA token has been announced The bigger story is what sits around the chain. GIWA is also building a wallet, on-chain identity/attestation through Dojang, privacy infrastructure through Bojagi, KRW/stablecoin payment rails, and GASOK, an accelerator designed to bring builders onto the ecosystem. And the testnet is already seeing significant activity. 120M+ transactions 30.8M+ addresses The official GIWA explorer now shows even higher cumulative testnet activity, so the network is clearly being used rather than sitting idle. There’s also an important point about the funding numbers floating around CT. I wouldn't treat the often-mentioned $143M GIWA raise as an official funding round. The ~$10–11B figure being circulated refers to Dunamu’s company valuation, not money raised by GIWA itself. In other words, this isn't a normal L2 launching with a VC/token fundraising story. It is being built by the company behind one of Korea’s largest crypto exchanges. That changes the distribution angle. If GIWA eventually gets pushed directly toward Upbit’s existing user base, the potential onboarding funnel is very different from a completely new chain starting from zero. And GIWA is already building the ecosystem around it. My GIWA watchlist DeFi / Infrastructure • @heronprotocol — DeFAI capital allocation • @quadrixfinance — on-chain asset management / vaults • @AmmoraHQ — programmable liquidity infrastructure • @custosfinance — RWA-focused trading infrastructure • @OsigeExchange — GIWA DEX + trading tools • @CheomaYield_xyz — GIWA yield / DeFi • @ZKProofport — zero-knowledge infrastructure • @Hanok_io — on-chain contribution / rights infrastructure • @kaachi_io — wallet-based marketing + rewards • @GiwaterFinance — liquidity / DeFi infrastructure Consumer / NFTs / Games • @jumak_app — Joseon-themed pixel NFTs • @pabalx — on-chain horse racing / social game • @TileOnGIWA — GIWA NFT collection • @PunkOnGiwa — CryptoPunks-inspired NFTs • @elderwoodwiz — early GIWA NFT collection • @GIWAUP — NFT / community project • @Writ_Fi — fully collateralized on-chain options Launchpads / Meme Economy • @gasokfun • @GIWAForge • @GIWALAUNCHPAD • @JoseonFun • @hamrdotfun • @Kimchifunxyz • @GiwaPump • @Gazuaxyz • @memefab_fun • @GiwaGasok • @MapaeGiwa Most of these are early-stage projects, so the watchlist is exactly that: a watchlist, not an endorsement. AI / Social / Other • @OreumLabs — on-chain information / curation • @TheBingoFi — GIWA-focused GameFi • @umul_giwa — GIWA messaging protocol One thing I find interesting is that GIWA isn't only positioning itself as another L2. The official roadmap connects the chain with wallet infrastructure, stablecoin payments, on-chain verification, and builder onboarding. GASOK is already offering grants, development resources, wallet integration opportunities and a path toward mainnet deployment. Mainnet is still listed as in development. So for me, the interesting question isn't: “Does @GIWA_by_Upbit have a token?” It doesn't appear to have an official one right now. The more interesting question is: What happens if Upbit turns its existing user base into GIWA users? That's the part I'm watching.

💎 early project below 1k followers > @subfrostp2p | L1-native DeFi > @PyreFun | Launchpads > @Stonkerage | perpetual > @clanktrade | AI-native launchpad > @darkwoodslabs | privacy > @tarecredit | RWA infra > @Tickr_co | perpetual DEX > @JoltzuLabs | L1. > @ClankerTownRH | AI Agents > @tenkafinance | RWA infra > @enter_elysium l L2 > @frame_robotics l Robotics this as a financial and do make sure you do proper due diligence.

➥ $NEAR is starting to look less like a single blockchain and more like a full-stack crypto platform. Over the past few months, the pieces have been coming together: Confidential Intents → tokenized stocks → AI agents → faster infrastructure. Here’s what has actually gone live recently: --- 1. Confidential trading is becoming the default $NEAR Intents now supports confidential perpetual futures through Hyperliquid. You can fund the account from supported assets and chains, while deposits and withdrawals remain private by default. No separate broker account. NEAR also added limit orders to Intents, giving wallets and apps more control over how swaps execute. Confidential TVL recently crossed $70M, triggering the first NEAR@3.33 milestone drop. --- 2. Tokenized stocks are now live NEAR.com launched tokenized US stocks and ETFs through Ondo Finance. That includes: $NVDA, $TSLA, $AAPL, $GOOGL, $META, $MSFT, $AMZN, $SPY, and $QQQ. The idea is simple: Buy fractional exposure using crypto, from 30+ chains, 24/7, without opening a traditional brokerage account. Availability depends on jurisdiction. --- 3. Intents is becoming a major volume layer $NEAR Intents has processed roughly $30B+ in cumulative volume. There was also a $300M+ single-day volume print on September 18, while weekly volume has reached roughly $800M–$1B+. Privacy-related flows have also become a meaningful part of that activity. --- 4. The base layer is getting upgraded $NEAR 2.13 brought two major changes to mainnet: → Dynamic resharding: shards can automatically split or merge based on network demand. → Post-quantum signing: ML-DSA support is live, allowing accounts to rotate to quantum-resistant keys. The next major step is SPICE, which aims to separate consensus from execution and push block times toward ~200ms, down from roughly 600ms. --- 5. NEAR is building an AI stack $NEAR AI is going beyond the usual “AI + crypto” narrative. It is building infrastructure for private inference and autonomous agents. Users can stake NEAR to pay for private inference and agent hosting. It has also introduced: → IronClaw 1.0 → Confidential inference using Intel TDX → Intel Trust Authority attestation → Integrations with open-source agent frameworks → Support across dozens of models --- 6. The institutional side is growing $NEAR recently went live for spot trading on Hyperliquid. Bitwise’s NEAR Staking ETP has crossed $100M AUM. NEAR also reports that trailing 30-day protocol fee capture has risen to around 30%, compared with a much lower lifetime average. --- The bigger picture is becoming clearer. @NEARProtocol is building around a simple stack: Assets move across chains → Intents handles execution → privacy protects the transaction → tokenized assets expand access → AI agents add another layer of autonomous activity. That’s the part of NEAR I’m paying attention to.

If there is one thing this AMA proved, is that the ethereum:0xcf0c122c6b73ff809c693db761e7baebe62b6a2e community is truly ALIVE and ready for the BATTLE ahead‼️ EXCITING times‼️

WB

we all are not daft to know he was among this team when he pull up that twee. his moves was so obvious. i told a fren of mine that if this guy wasn’t part of the team he definitely had an idea about the art more than we can even say. @Smartecio you sure was mills ahead of that nigger. man thought he was smart too

one of the biggest clown i have seen on my TL of lately. they made their first post, you didn’t tweet. they made us grind and make endless tweet about their project you never came out. now we done with minting only for you to crawl out from your dungeon to make this baseless post. and you knew all along. imagine you tweeting this shit from the same city I’m in, man you’ll be so scared of coming home.

finally got in on @zaddrnet opened ZADDR #900. Rank 2,051 of 2,800. bag officially in the vault. floor price 1 ZEC

The attention around decentralized compute isn’t coming out of nowhere. AI demand is running into a simple problem: • GPUs are scarce • cloud prices are high • a lot of hardware is sitting idle Mining farms, smaller data centers, and even consumer machines can have computing power that isn’t being used. Decentralized compute is basically an attempt to turn that idle capacity into a marketplace. And in 2026, this is starting to look less like a narrative and more like a real market. Some decentralized GPU networks are offering inference and fine-tuning at prices that can be 60–80% below on-demand cloud providers like AWS and Azure. That doesn’t mean decentralized compute will replace hyperscalers. Large AI training jobs still need tightly connected GPUs, fast networking, and predictable performance. A single cluster with InfiniBand is hard to beat there. But not every AI workload needs that. Batch inference, rendering, overflow capacity, and smaller training jobs can be much more flexible. That is the opportunity. The sector itself is also becoming easier to understand. Not every project is doing the same thing: - some rent GPUs - some coordinate distributed training - others sell access to machine intelligence rather than raw computing power ✦ Here are some names am watching: → @rendernetwork One of the more established GPU networks. It started with decentralized rendering and has expanded toward AI inference and compute. The important thing here is actual GPU supply and usage, rather than just an AI narrative. → @akashnet A permissionless cloud marketplace built on Cosmos. It lets users rent compute and GPUs through a marketplace, with pricing driven by supply and demand. Its bigger value proposition is simple: cheaper and more flexible cloud infrastructure. → @ionet Focused on aggregating distributed GPUs into clusters for AI training and inference. The supply can be more fragmented than curated GPU providers, but that is also the point: finding available computing power across a wider network. → @AethirCloud More focused on enterprise GPU infrastructure. It has positioned itself around AI inference, gaming, and other workloads that need reliable access to GPUs. → @opentensor This one is different. Bittensor isn't simply a GPU rental marketplace. Its subnets create markets around machine intelligence, where participants are rewarded for producing useful or highly rated outputs. So the product is closer to buying intelligence, rather than buying GPU hours. → @gensynai Focused on decentralized AI training. This is a much harder problem than simply renting GPUs because distributed training requires coordination and verification across different machines. The interesting question is whether that infrastructure can eventually support serious AI training at scale. → @nosana_ai A smaller, more lightweight GPU network focused on inference and compute workloads. It is particularly interesting for use cases where users care about cheaper, accessible GPU capacity rather than massive centralized clusters. There is an important limitation, though. Decentralized compute doesn't magically solve the physics of distributed systems. If an AI model needs hundreds of GPUs communicating extremely quickly with each other, a tightly connected data center still has a major advantage. Reliability, latency, networking, and SLAs matter. So the real opportunity isn't necessarily replacing AWS, Azure, or Google Cloud. It's serving the workloads where cheaper compute, unused hardware, and flexible capacity matter more than having everything inside one warehouse. That is why I'm watching this sector. AI is going to need an enormous amount of compute. The projects that matter won't necessarily be the ones with the loudest AI narrative. They'll be the ones that can prove real GPU utilization, real customers, and real revenue from outside crypto.

Some events to watch out for this week! @axisrobotics $AXIS Kicks off AXIS token sale on Sonar today @avax $AVAX Helion mainnet upgrade happening tomorrow. @BitMEX $BMEX BitMEX to Shuts Down All Operations on the 23rd. @Balancer $BAL BAL Winddown Governance Vote to take place on 25th of september. @Solana $SOL Alpenglow Mainnet to Launch on the 28th. @CoinExcom $CET Shuts Down on 28th as well, Most Services Halt @pancakeswap $CAKE PancakeSwap’s Pre-Access Portal, which will allow getting tokenized exposure to private companies, is expected to open in the next few days @injective $INJ Injective's Meridian Mainnet Upgrade, enabling institutional-grade RWA Issuance, will go live on Sept. 24 -> Biggest Token Unlocks for this Week $XPL → $177.79M $AKE → $81.75M $GRAM → $51.47M $H → $20.79M $CC → $17.08M — Daily $TRUMP → $13.23M — Daily $FOGO → $10.40M $TAO → $6.93M — Daily $SOSO → $6.82M — $MORPHO → $5.58M — Daily $XPL dominates with $177.79M, representing 45% of the total and 18.94% of its supply. Six straight days have major cliff unlocks, with the biggest supply pressure coming from $XPL (18.94%), $FOGO (15.4%), and $H (2.93%). "

This thread is 100% gonna blow up and I am gonna look like a salty c****again but @icobeast pissed me off so now it's gonna get ugly Kalshi fakes their crypto volume and I can prove it NOTHING pisses me off more than watching a company treat its own customers like complete fucking idiots Grab some popcorn cause I dissected every piece of evidence one by one until there’s basically no way for him to respond without making this look even worse Oh and FYI, I left every source at the end so y’all can check the receipts yourselves 😘 1/x