ChainFeeds published its Aug. 5 research roundup with five featured topics: on-chain privacy, Uniswap’s apparent new launchpad effort on Robinhood Chain, Strategy’s latest capital moves, the idea of crypto as a new financial base layer, and Cloudflare’s stablecoin wallet system for AI agents.
Three answers to blockchain privacy
In a weekly note, IOSG Ventures argued that the real dividing line between privacy systems comes down to a simple question: where is the secret kept. The firm said no single design wins across the board, and real products often combine several methods. It gave examples such as using secure hardware for speed and then proving with zero-knowledge proofs that the chip did not cheat, or storing hidden state with fully homomorphic encryption and validating inputs with zero-knowledge proofs. It also pointed to a fifth path that skips cryptography entirely by sending data only to parties allowed to view it, which it said helps explain why banks favor Canton.
The earliest privacy efforts tried to make the network private from the start rather than adding privacy tools to an existing chain. IOSG used Zcash and Monero as the main examples, while stressing that the two projects took very different paths.
Zcash is an independent blockchain that works in many ways like Bitcoin-style digital cash, but adds a private mode. It supports both transparent and shielded addresses, letting users decide whether to hide the counterparties and the amount in a transaction. That shielding process relies on zk-SNARKs. When a user spends shielded ZEC, the wallet creates a short proof showing the transaction is valid and does not double-spend funds, without exposing the transaction details to the network.
IOSG said one of Zcash’s key design choices also became a long-term weakness: privacy is optional. Users can enter the shielded pool or stay with transparent addresses, which means the privacy set depends on a large enough group of users. If only a small number choose the private mode, those transactions can become easier to identify. The report said most ZEC has remained in the transparent pool over the past decade, limiting the real size of the privacy set. At the same time, shielded ZEC has an important property — fungibility. Once a coin enters the shielded pool, its earlier transaction history drops away and it no longer carries source information from the past.
Monero took the opposite route. Privacy is on by default and there is no public mode. Every transaction is private from the outset, so there is no separate step where users join a privacy pool. The whole chain functions as the privacy set. IOSG said Monero historically relied on three main tools: ring signatures to hide the sender, stealth addresses to hide the receiver, and RingCT to hide transaction amounts. An outside observer cannot determine the real source of payment or see the exact amount.
The note added that Monero completed a major FCMP++ upgrade in early 2026. That change replaced traditional ring signatures with zero-knowledge membership proofs, expanding the anonymity set from a limited group of decoys to the full history of transactions.
Canton was presented as a third model. Rather than hiding data that has already been broadcast, it avoids sharing the data in the first place. Zcash and Monero accept public verification as a requirement of blockchains and then use cryptography to conceal the contents. Canton removes the shared ledger model, with each participant receiving only the transaction data relevant to that participant. It uses the Daml smart contract language and a Global Synchronizer to order and confirm transactions, without exposing the transaction content to unrelated parties.
Under that setup, privacy comes from access control rather than zero-knowledge proofs, FHE, or MPC. The system sends data only to those who should see it. IOSG said that design fits banks and other institutions because privacy is on by default, does not depend on anonymity sets, and avoids the fund-freeze issues that public chains can face. But it gives up some openness. There is no permissionless composability, liquidity stays inside a permissioned environment, and the privacy model depends on participant permissions rather than a purely mathematical guarantee. The report added that Canton Coin launched alongside the Global Synchronizer mainnet to pay for application and infrastructure fees, making it an important case in institutional privacy finance.
Why pools.trade looks tied to Uniswap and Robinhood Chain
A separate long post from SliKᵍᵐ said Pools appears to be Uniswap’s user-facing launchpad for Robinhood Chain. The base layer infrastructure is already on-chain, but the product itself is not yet formally open. The post also said there is no official $POOLS token at this stage.
The pools.trade landing page is minimal. It displays the message “Coming soon from Uniswap” and plays a night-vision clip of a frog by a pond that jumps into the water and then returns to the frame. The domain was registered on July 17, 2026, and the video file carries a July 28 timestamp. The site is also set to noindex, which the post read as an attempt by Uniswap to keep the page out of search results for now.
The more useful clues sit below the surface. The video path includes rh-cca/frong.mp4, and both public development and testing environments point to load balancers that also use rh-cca in their names. According to the post, RH almost certainly refers to Robinhood, while CCA stands for Uniswap’s Continuous Clearing Auction. Taken together with other public client-side material, that naming pattern is hard to dismiss as coincidence.
The write-up separated three ideas that could otherwise be mixed together: Launches, Pools, and CCA. Launches is already live on Uniswap pages and aggregates tokens created through third-party launchpads. Pools appears to be Uniswap’s own venue on Robinhood Chain for creating, issuing, and trading tokens. CCA is the on-chain engine underneath it, handling the auction mechanism, price discovery, and eventual migration into Uniswap v4 liquidity pools.
Public client code revealed a long list of product details. The current Uniswap App client contains many internal strings beginning with rhcca.* that describe Pools pages, token economics, and user flows. The internal working title is direct: “Pools - create a token on Robinhood Chain.”
The material suggests two issuance modes.
- Crowd Launch: a fixed-supply auction lasting 4 hours with a total token supply of 1 billion. Half of the supply is sold in the auction. The other 50%, along with the funds raised, goes into the post-launch liquidity pool. The interface says liquidity will be permanently locked. Graduation happens at a fully diluted valuation of $50,000. If the auction misses the target, all bidders can reclaim their funds. If it succeeds, the token moves into a Uniswap Pool and starts trading. The auction uses time-weighted bids, aiming to reduce bundle sniping while giving participants the same final clearing price.
- Instant Launch: a model closer to the bonding-curve launchpads already familiar to the market. Trading starts right away, 80% of supply goes into the curve, and the remaining 20% plus raised funds are used for liquidity. The graduation threshold is also a $50,000 FDV, but there is no time limit.
The client also references several additional features: ETH and USDG as funding assets, optional verification of a creator’s X account, a 0.25% pool fee, a 0.05% LP fee paid to the creator, an updatable fee recipient address, and modules for holder data, top traders, profit and loss, activity history, claims, and refunds.
Although Pools has not officially opened, its core infrastructure is already deployed. The post said Uniswap has publicly recorded Liquidity Launchpad contracts on Robinhood Chain. Those contracts were deployed between July 6 and July 10, before the pools.trade domain was registered. At the time of the scan, the CCA Factory had emitted roughly 485 AuctionCreated events, LiquidityLauncher had created more than 700 tokens, and the current LBP Strategy showed more than 100 migrations.
That is why the author argued Pools is not just a teaser page. It points to live on-chain infrastructure. The contracts are open and permissionless, so anyone can create auctions. The post said that helps explain the large number of test projects, spam tokens, and imitation projects already appearing on Robinhood Chain. Some tokens have started building a narrative around Pools, but the author said those should not be treated as official Uniswap launches or investment recommendations.
The clearest example involved FRONG and CHWDR. FRONG was the first token created by address 0xE195...cE58, a wallet that Uniswap data links to @zacklabadie. It used the Instant Launch model, its image matched the frog shown in the teaser, and the teaser video file itself was named frong.mp4. Twenty-two minutes later, the same wallet created CHWDR, this time using a 4-hour Crowd Launch. The API shows both projects tied to the same X identity.
Strategy sold BTC and issued MSTR without adding to holdings
A third feature, citing Cryptoslate, focused on Strategy’s latest bitcoin sale and MSTR share issuance. The report said the move was designed to push STRC back toward par and reopen an important financing channel. None of the proceeds were used to buy more bitcoin, extending Strategy’s pause in BTC accumulation to 6 weeks, its longest such break since 2024.
According to the company’s Aug. 3 filing with the U.S. Securities and Exchange Commission, Strategy sold 1,638 BTC between July 27 and Aug. 2 for $104.7 million. It also issued about 3.01 million MSTR common shares, raising $290.6 million. Together, the two transactions brought in about $395 million.
The report said the funds were used instead to pay preferred dividends, repurchase STRC shares, and complete an expansion of dollar reserves to $4 billion. It also noted that Strategy’s total BTC sales for 2026 have now reached 5,258 coins, the largest amount the company has sold in any single year since adopting its bitcoin strategy in 2020.
The biggest obstacle to renewed BTC buying is now STRC, a floating-rate preferred share that Strategy wants to turn into a repeatable, sustainable financing tool for its bitcoin treasury strategy. When STRC trades near its $100 issue price, the company can issue more STRC close to par and channel the proceeds into its capital structure, including bitcoin purchases. If the stock remains well below par, that channel weakens because the company would have to accept a lower sale price or raise the dividend rate to attract buyers.
Since May, STRC has traded below par, according to the report. Strategy can support the price by adjusting the dividend, but it still chose to raise the annualized dividend rate to 12% and repurchase shares at a discount. Last week, the company used $52.30 million from bitcoin sales and $28.90 million from MSTR issuance, for a total of $81.20 million, to buy back 912,143 STRC shares. In the prior week, it spent another $25 million to repurchase 288,930 STRC shares at an average price of $86.53.
Strategy President and CEO Phong Le said on the second-quarter earnings call that repurchasing STRC below par lets the company reduce future dividend obligations at a discount while also helping demand improve as the stock moves back toward $100.
Since starting the plan in July, Strategy has spent about $106.2 million on STRC buybacks, the report said. The company still has $893.8 million of STRC repurchase capacity and a separate $1 billion MSTR common stock buyback authorization that has not been used. Its goal is to restore STRC to par by September, with the speed of future buybacks depending on share performance and market liquidity.
Crypto as infrastructure, not only as speculation
The fourth article, from imToken, argued that tokenized real-world assets, stablecoins, prediction markets, and agentic payments should not be treated as four isolated trends. In that framing, they reflect the simultaneous maturation of assets, money, identity, trading, and settlement.
The piece said that for years, much of the crypto industry’s narrative could be reduced to one near-obsession: what is the next asset that will go up. From DeFi Summer onward, NFTs, public chains and L2s, restaking, meme tokens, and AI-linked assets each took turns in the spotlight. Their technical logic and market backdrop differed, but the discussion often ended with price action. Even stablecoins, wallets, and cross-chain bridges, despite their clear utility, were often judged by how much trading and speculation they could carry.
In 2026, however, several developments from different corners of the market started appearing in the same window. imToken pointed to a stablecoin market cap of about $300 billion, DTCC completing its first tokenization conversions in a production environment and planning to formally launch related services, prediction markets moving from crypto-native products into brokerages and regulated exchanges, and AI agents beginning to use stablecoins to buy data, model access, and digital services.
Viewed together, the article said, those changes trace a broader shift. Over the past decade, crypto has built issuance, custody, trading, payments, and settlement rails that are starting to open beyond crypto assets themselves and into wider financial activity and the machine economy. The market has not outgrown speculation, but under that layer a more complete infrastructure stack is starting to take shape.
In imToken’s view, the catalyst is not a single hot narrative. Different components of a new financial system developed separately for years and are only now starting to connect. Stablecoins changed the way money can be used first. They once served mainly as exchange quote assets, on-chain hedging tools, and settlement instruments for crypto assets. Now more issuers, banks, payment companies, and fintech firms are exploring them for merchant acceptance, corporate payments, global payroll, and cross-border settlement.
The article said the core change is that stablecoins are no longer just dollars on-chain. They are a form of money that software can call directly. They can move around the clock, be embedded into programs, release automatically under preset conditions, and become the settlement asset as soon as a transaction finishes. For internet applications, sending a stablecoin payment is moving closer to calling a payments API than interacting with a complicated bank clearing system.
At the same time, RWAs address a different question: which assets can be traded and settled. Once assets can be recognized by smart contracts, they can flow directly into collateral, lending, trading, treasury management, and automated investment processes. In that sense, blockchains are no longer just venues for minting new tokens; they are starting to carry part of the operating process for traditional assets.
The article also said that treating crypto as infrastructure does not mean speculation disappears. A more useful test is whether external companies and users are starting to rely on the technology to do work that was previously too hard, too expensive, or too inefficient. By that measure, crypto and Web3 are already forming a multilayer next-generation financial stack.
The first layer is asset issuance and mapping. The set of assets reaching blockchains now extends beyond native tokens to stablecoins, Treasuries, money market funds, private credit, gold, fund shares, and equities. The point, the article said, is not simply to place a certificate in a wallet. Once assets are legible to smart contracts, they can move straight into collateralization, lending, trading, treasury operations, and automated investment workflows. Processes once split among registrars, custodians, brokers, and clearing systems may end up compressed into a more unified execution environment.
Cloudflare Wallets are built for AI agents
The fifth article, from TechFlow, said Cloudflare Wallets are not a conventional crypto wallet product for retail users. They are not directly comparable to MetaMask, exchanges, or DeFi apps. Cloudflare is building payment infrastructure for AI agents.
The article described a basic limitation facing AI agents today: they cannot complete economic activity on their own the way humans do. When an agent needs to call an API, buy data, or use a digital service, it usually has to go through account registration, payment setup, and key generation flows designed for human users. Without autonomous payment capacity, agents still stop at key points and wait for human approval, which breaks true end-to-end execution.
Cloudflare Wallets are meant to give AI agents something closer to a digital account, allowing them to make payments within pre-set rules. The design uses a two-layer wallet structure. The first layer is an Account Wallet controlled by a person or a business and used for deposits, withdrawals, and permission management. The second layer is a Virtual Wallet assigned to a specific AI agent. A business can define spending caps, approved services, and per-transaction limits, after which the agent can buy services on its own inside that authorization range.
The article said the main change here is Cloudflare’s attempt to solve one of the biggest barriers to AI agents entering the real internet economy: payment and identity. Internet service systems still revolve around human users. To access an API, a user normally registers an account, completes identity checks, links a bank card or credit card, and then receives usage rights. That flow does not fit AI agents, which may call hundreds of APIs and pay only a few cents, or less, per request. Traditional payment fees, settlement cycles, and account structures are poorly suited to that kind of high-frequency, low-value, automated activity.
Cloudflare’s answer uses stablecoins and the x402 protocol. The article described x402 as a way to embed payment information directly into an HTTP request so an AI agent can handle identity verification, payment, and resource access in the same step. Instead of a process built around “register first, then pay, then call the service,” the flow could become “request the service and pay at the same time.”
Beyond payments, Cloudflare also launched the cloudflare.pay domain service to give wallets recognizable identity markers. In that model, an AI agent is no longer just an anonymous machine request. It can carry a verifiable identifier similar to an organizational identity. The article gave the example of an agent deployed by a research institution proving to a service provider that it belongs to that organization. This is aimed at the trust problem. Websites often treat bot traffic defensively because they cannot tell whether a request comes from a real user, an automated program, or a malicious bot. Once agents have verifiable identity, service providers can assign different permissions, such as trial balances, enterprise discounts, or higher access tiers.
TechFlow said Cloudflare Wallets matter because they point to a shift in how infrastructure companies are thinking. Stablecoins are moving from an internal trading tool for crypto markets toward payment infrastructure for the broader internet economy. Cloudflare carries a large share of global website traffic, so its move into wallets and payments for AI agents sends a different signal than another standalone Web3 project. It suggests internet infrastructure is starting to adapt itself to the economic activity of the AI era.
The article also stressed that Cloudflare did not push a particular blockchain or launch a token. It treated stablecoins as a payment medium, with the underlying network left interchangeable based on need. In that view, what matters is whether payment capacity can be called by machines.
Cloudflare’s Monetization Gateway, launched at the same time, forms the other side of the market. For service providers, the gateway can help websites charge AI agents. For AI agents, Wallets supply the spending side. Put together, the article said, they create a full loop: providers offer resources and set prices, AI agents discover demand and pay for access, stablecoins serve as the settlement medium, and HTTP becomes the base transport channel for the transaction.
The implication, as the article framed it, is a possible new internet economy where machines exchange resources directly. AI agents could buy data, pay for model access, acquire compute, and use software services without a human approving each transaction one by one. Humans would set goals, permissions, and budgets. Agents would carry out the economic actions.
