A new episode of Hivemind is live! This week the team breaks down whether the latest crypto rally has staying power, which assets could lead, and why the return of the debasement trade may signal a shift in the market. Timestamps: 00:00 Intro 00:35 Is the Crypto Bull Market Back? 08:55 HYPE & the Next Alt Season 15:35 What Wins This Cycle? 21:05 Consensus Trades vs. Hidden Alpha 29:20 Zcash & the Next Rotation 52:20 NFTs & Robinhood Chain 01:04:35 Positioning for the Next Leg 01:12:30 The Debasement Trade Is Back

China’s AI hardware disadvantage is turning efficiency into a competitive advantage. Chinese labs have spent the past several years trying to keep pace with the frontier under tighter restrictions on Nvidia’s most advanced chips. Restricted access made it harder to improve models by simply adding more compute. Labs had to look for gains in how models were trained and run. Much of that work focuses on improving chip utilization during training and designing models that require less compute and memory. Post-training methods are also extracting more reasoning capability from existing base models. Chinese AI progress cannot be reduced to distillation. These systems change how models are built and run while lowering training and serving costs. That cost advantage changes how models compete. Many workloads do not require the strongest model available. A lower-cost option can win once it performs well enough for the task. Several Chinese models already combine competitive performance with substantially lower prices. China is much closer to running competitive models on domestic chips than it is to training them from scratch. Frontier training still depends on advanced hardware China cannot yet fully replace. The next generation of Chinese models will show how far better software and more efficient architectures can narrow that gap.

NEAR is building the infrastructure for AI agents to hire one another. Most agent commerce today is built around fixed services that can be priced per call. Hiring another agent creates a different kind of transaction because the buyer is paying for an outcome that may take several steps to produce. NEAR is building a single stack around the full process. Agent Market turns a task and budget into an open job that agents can bid on. The selected provider executes the work and is paid through escrow. Delivery can run through IronClaw with private inference supplied by NEAR AI Cloud. NEAR Intents supports settlement. Completing the transaction requires one more step: deciding whether the provider delivered what was promised. That becomes difficult when the work requires expertise the requester does not have. With reliable verification, the model NEAR is building could take agentic commerce beyond pay-per-call services and into markets for completed work.

Our new report "Task Markets: the Final Form Of Agentic Commerce" is live! The next phase of agentic commerce depends on whether agents can coordinate at scale. In a task market, an agent can hand off parts of a workflow to other agents or humans and release payment once the work is verified. The need for verification grows with every extra step because errors compound across the workflow. A ten-step workflow with 95% accuracy at each stage has only a 60% chance of finishing without an error. Task markets create checkpoints at each handoff so mistakes can be corrected before they spread. Agent-to-agent hiring already exists on Virtuals, NEAR, and Daydreams. Activity remains concentrated in crypto-native use cases and simple services. x402 and MPP now give agents a direct way to purchase fixed services such as data or inference. Task markets build on this payment layer by coordinating several providers across a longer workflow. An internal onboarding agent might use pay-per-call to purchase a background check. The rest of the process could still depend on document review and human approval. A task market gives the agent a way to coordinate those steps without integrating every provider in advance. Enterprises are the most likely source of early demand because companies already use agents to run internal workflows. Task markets become useful when those workflows depend on providers outside the organization. Reliable handoffs could allow agent-to-agent hiring to scale.

Ceteris explains why speculation can set the stage for new crypto experiments. “Even if it’s a lot of these dumb meme coins... it shows that there is a level of appetite that is returning. Sometimes you get the net-new experiments and projects once you have a little bit of that momentum.”

Plasma One aims to become the primary financial account for stablecoin users. Many people use stablecoins to protect against weakening currencies when dollar banking is hard to access. Their financial activity is often split across exchanges, P2P brokers, wallets and separate card programs. Plasma One brings savings, everyday spending and free USDT transfers into one account. Monthly stablecoin card volume rose from roughly $230 million in January 2023 to $1.5 billion in August 2025. P2P stablecoin payments increased only slightly over the same period, from $1.4 billion to $1.6 billion a month. By August 4, Plasma One had processed $32 million in card volume. Nearly half came in July as monthly volume grew 69% to $15.2 million. That card activity brings users onto Plasma and gives them a reason to keep more of their money on the network. Each new market requires local payment and compliance infrastructure that other products on Plasma can reuse. Over time, its moat could come from those users and local integrations.

Robotics has become a global industrial race. Can the US compete with China? We went inside Standard Bots, the company betting America can build them at scale. Our first Delphi Media production premieres tomorrow on @RoboStrategy.

Our new report "Closing the Loop: The Self-Driving Landscape" is out now! AI can generate scientific hypotheses faster than laboratories can test them. Self-driving labs are designed to close that gap by automating the full experimental loop. The model chooses the next experiment, and connected laboratory equipment carries it out. The results feed back into the system and shape what it tests next. The stakes are especially high in drug discovery. Lead optimization alone can consume roughly three years. Bringing a drug to market takes 10–15 years, with average out-of-pocket and time costs of $2.6 billion per drug. Self-driving labs target the earlier experimental bottleneck. They could shrink individual cycles from months to days or hours. Running more experiments can also reduce the cost of each run by spreading the upfront cost of automation further. Every completed experiment adds to a structured record of what worked and what did not. That data improves the model’s next decision, creating a continuous learning loop between AI and the physical lab. AI has accelerated the generation of scientific hypotheses. Self-driving labs could accelerate the experiments that determine which ideas are worth pursuing.

JTX wants to be the only Solana trading surface you need. No single app covers everything a Solana trader does. Spot, perps, and prediction markets are scattered across different apps. Solana isn't short on liquidity, but it lacks a single place to put all of it to work. @jtx_trade aims to fold all of it into one pro-retail trading experience that feels like a CEX or brokerage account, without needing to juggle apps or give up custody. Jito's stack helps build and order a large share of Solana's blocks, shaping how trades get filled. JTX builds on that by using DFlow for spot trades and tapping into the existing liquidity on Solana to deliver the best possible onchain execution across a wide range of assets. None of this guarantees users show up. Jupiter owns the swap, Phantom owns the wallet, and terminals own the memecoin flow. Better execution alone has rarely been enough to change those habits. If JTX manages that, it could become Solana's default trading venue.

Morpho is building the infrastructure layer for other businesses to run credit products on top of. Curators handle the risk and integrators bring the distribution. The protocol stays underneath while the participants on top earn the revenue. The model is more efficient than peer-to-pool. Morpho's isolated pools run tighter than Aave's shared pool, and top USDC vaults beat both Aave and Compound on supply rate net of fees. The Kelp exploit put Aave under real stress. The intervention to protect borrowers trapped suppliers for five days while four of its biggest markets froze. On Morpho, AdaptiveCurve would have adjusted rates automatically as utilization climbed and isolated pools would have kept the damage contained.

Ostium just announced a partnership with Nasdaq. Last month, we covered how Ostium is bringing the world's deepest markets onchain. Read the consulting report for free here.

Ostium just announced a partnership with Nasdaq. Last month, we covered how Ostium is bringing the world's deepest markets onchain. Read the consulting report for free here.

Strategy buys back $1.5B of its 2029 convertibles at a discount. Our report identifies convertible repayment and STRC dividends as Strategy's main pressures. This buyback directly reduces the repayment burden. First 30 people can read it for free here.

Months ago our markets analyst @that1618guy called the BTC/Gold ratio near the floor at the 2nd deepest drawdown in the framework's history. The ratio has recovered roughly +40% off the February lows. Gold corrected nearly -19% from its all-time high as BTC gained ground back to ~$78K. This is Scenario 1 playing out with gold declining. The primary confirmation signal is a 1W 9/21 EMA green cross, projected for early June. The three prior green crosses after the deepest red crosses ran +148%, +641%, and +148%. The macro backdrop is messier than it was in February with inflation back at 3.8%, rate cuts off the table, and an active war with Iran. Kevin Warsh has been confirmed as Fed Chair and his first FOMC lands on June 16-17, around the same time as the projected cross. The PBOC has bought gold for 17 consecutive months and continued through the $5,000 level. The next leg of the ratio recovery depends on BTC strengthening. The cross is now weeks away.

Months ago our markets analyst @that1618guy called the BTC/Gold ratio near the floor at the 2nd deepest drawdown in the framework's history. The ratio has recovered roughly +40% off the February lows. Gold corrected nearly -19% from its all-time high as BTC gained ground back to ~$78K. This is Scenario 1 playing out with gold declining. The primary confirmation signal is a 1W 9/21 EMA green cross, projected for early June. The three prior green crosses after the deepest red crosses ran +148%, +641%, and +148%. The macro backdrop is messier than it was in February with inflation back at 3.8%, rate cuts off the table, and an active war with Iran. Kevin Warsh has been confirmed as Fed Chair and his first FOMC lands on June 16-17, around the same time as the projected cross. The PBOC has bought gold for 17 consecutive months and continued through the $5,000 level. The next leg of the ratio recovery depends on BTC strengthening. The cross is now weeks away.

Morpho is becoming the credit stack for institutions. What started as a yield optimizer has become the base layer other lenders build on. Morpho v2 keeps the core immutable and permissionless while adding key features for institutions to operate onchain. Apollo is acquiring up to 9% of Morpho's supply over four years. Morpho Midnight brings fixed rates and fixed terms to onchain lending. Borrower and lender intents match offchain and settle onchain as zero-coupon bonds, which opens up a real secondary market for the loans themselves. DeFi has never been able to pull this off at scale. Fixed rates are the foundation every mature credit market is built on. Pool-based lending can't deliver them because rates only adjust once utilization has already moved. Midnight gives DeFi a real yield curve that institutions can build on.

Morpho is becoming the credit stack for institutions. What started as a yield optimizer has become the base layer other lenders build on. Morpho v2 keeps the core immutable and permissionless while adding key features for institutions to operate onchain. Apollo is acquiring up to 9% of Morpho's supply over four years. Morpho Midnight brings fixed rates and fixed terms to onchain lending. Borrower and lender intents match offchain and settle onchain as zero-coupon bonds, which opens up a real secondary market for the loans themselves. DeFi has never been able to pull this off at scale. Fixed rates are the foundation every mature credit market is built on. Pool-based lending can't deliver them because rates only adjust once utilization has already moved. Midnight gives DeFi a real yield curve that institutions can build on.

Strategy buys back $1.5B of its 2029 convertibles at a discount. Our report identifies convertible repayment and STRC dividends as Strategy's main pressures. This buyback directly reduces the repayment burden. First 30 people can read it for free here.

Perp DEXs have a capital productivity problem. A venue can attract deposits with incentives, but the harder test is making that capital useful enough to stay. @DecibelTrade starts by making collateral more useful. Through DLP users can earn market-making yield from the liquidity vault while using the same position as margin for perp trades. That makes the deposit more useful because the same capital can earn yield while it supports trading. Decibel's native stablecoin (usDCBL) applies the same idea to the venue’s balance sheet. Stablecoin collateral can generate reserve revenue for the protocol. That revenue can create more room to compete on fees over time. X-Chain Accounts make that capital easier to bring in. Users can sign Aptos transactions with existing Ethereum and Solana wallets instead of setting up new infrastructure. The next step is turning productive capital into usable liquidity. Market makers update quotes far more often than they get filled. Tight spreads become harder to sustain when each update burns gas. Decibel’s bulk orders reduce quote-update gas by roughly 90%. Makers can replace the full ladder in one transaction. The cancel and replace happen together so stale quotes are less likely to remain live. Aptos’ encrypted mempool supports the same goal by keeping orders and quote updates hidden until execution. This reduces the window for stale quotes to get raced. Together this increases capital productivity. Collateral can earn while it backs trades, stablecoin reserves can fund the venue outside trading fees, and lower quote-update costs can help makers keep the book tighter. The question is whether this turns incentive driven deposits into sticky liquidity.