Orderly (ORDER) Plunge

Cerebras Systems (CBRS) released its first quarterly report since going public. Q1 core revenue of $191.3M beat expectations, surging 92% YoY, with full-year revenue guidance raised to $855-$865M. However, the company's Q2 gross margin guidance of 36%-38% represents a sharp drop from Q1's 47%, sending shares down over 10% after-hours. Management attributed the margin pressure to temporary costs from leasing back hardware to deploy capacity for OpenAI, highlighting a business model shift from selling chips to selling compute power. While growth is strong, key investor debates center on valuation, customer concentration, and long-term contracts. Revenue remains heavily concentrated (86% from two UAE-linked entities in FY25), though major deals with OpenAI (a $20B+ contract) and AWS provide a long-term growth narrative. Analysts are broadly bullish, citing Cerebras's unique wafer-scale chip architecture as a potential advantage in the AI inference market. However, skeptics point to the narrow moat of its speed advantage, uncertainties around margin recovery amid its business transition, and a high valuation (~50x forward sales) that prices in the flawless execution of its large, long-dated contracts.

**Summary** Rialto is positioning itself as an on-chain spot exchange focused on order routing and execution quality for a range of assets, including cryptocurrencies and tokenized stocks/ETFs. It operates primarily on the Robinhood Chain network. Its core mechanism revolves around **propAMMs** – on-chain market makers that use proprietary inventory and pricing logic, referencing external markets (like underlying stock exchanges for tokenized equities) rather than relying solely on automated market maker (AMM) formulas. For each user order, Rialto's aggregator solicits real-time quotes from multiple liquidity sources, including its own propAMM (Rivo Altus) and traditional DEX pools. It then selects and atomically executes the path offering the best net output after costs, potentially splitting orders across sources. Rialto's key competitive claim is turning **order routing control into a service for better execution**. Instead of users manually finding the best liquidity, the system automatically compares and routes to the optimal source(s) for each trade. It has launched a partner program for integrations and its swap API is already used by several protocols. Challenges ahead include expanding the asset base on Robinhood Chain, ensuring Rivo Altus's price competitiveness, attracting diverse external liquidity, and consistently delivering superior net execution compared to manual routing. Success in these areas would see Rialto competing for the crucial "order routing right" in the emerging on-chain market for tokenized real-world assets.

The article explores the evolution of order in the AI & Web3 era, framing the competition around establishing new foundations for human-AI collaboration. It argues that as AI Agents become core economic participants, the existing Web3 infrastructure—focused on assets, identity, and decentralized governance—is insufficient. The core challenge shifts to building trust, assessing complex contributions, and creating fair value distribution among diverse human and AI actors. The analysis positions the m&WDAO project within this landscape by comparing it to existing paradigms: Colony (human-centric DAO governance), SingularityNET (AI service exchange), Gitcoin Passport (Sybil-resistant identity), and Farcaster/Lens (open social graphs). m&W's distinct path, termed EcoFi (Ecological Finance/Order), aims to integrate these layers. Its three-phase evolution is outlined: 1) **m&W 1.0: Credit Anchoring** - filtering high-quality "Builder" nodes and converting their contributions into non-transferable SBTs (Soulbound Tokens) as a bedrock of trust. 2) **m&W 2.0: Collaborative Economy** - deploying the EcoFi protocol to enable verified, complex task collaboration with a hybrid AI/human judgment system for valuation and dispute resolution, creating a closed-loop value system. 3) **m&W 3.0: Intelligent Order** - where human-originated SBT credit enables trusted AI Agent "digital twins" to participate in a mature human-AI co-creation economy. The conclusion asserts that while other projects solve discrete problems (proving identity, governing contributions, exchanging AI capabilities), m&W's ultimate mission is to address the foundational question of a new socio-economic order for the coming era of human-AI collaborative networks.

Bernstein revisits long-term agreements (LTAs) in the memory industry, highlighting new contracts with purchase commitments, minimum prices, and financial guarantees signed by Micron and SanDisk. These aim to provide an earnings floor for the coming years. Micron has 16 strategic customer agreements, with 14 representing approximately $100 billion in minimum revenue and about $22 billion in cash deposits/commitments. SanDisk has contracts for around $42 billion in minimum revenue and over $11 billion in guarantees. Combined, these ~$33 billion in guarantees make it more costly for major clients to walk away. However, Bernstein models that the potential revenue needing protection over 3-5 years is around $5.2 trillion. The existing guarantees thus cover only about 0.6% of that scale. While LTAs provide a cushion, they cannot fully shield profits in a severe downturn, as clients may still find it cheaper to breach contracts if spot prices fall deeply below floor prices. LTAs are most suitable for large, credit-worthy customers like U.S. cloud service providers with stable, high-volume AI infrastructure needs. Consumer segments (phones, PCs) and some Chinese clients are less likely to adopt them, leaving an estimated 30-50% of the DRAM/NAND market exposed to spot price volatility. AI demand (e.g., HBM for training, storage for inference) supports higher valuations and makes LTAs more attractive for locking in high-demand customers. Yet, Bernstein stresses that LTAs soften, but do not eliminate, the memory cycle. Their true test will come in the next downturn, revealing whether clients honor contracts and whether guarantees provide sufficient pain to maintain supplier discipline.

To purchase the experimental weight-loss drug retatrutide, Americans like Zach are turning to Bitcoin despite initial skepticism. Retatrutide, an unapproved triple-agonist from Eli Lilly, offers significant weight loss but is inaccessible through U.S. channels. Buying from domestic resellers costs around $3,600 annually, whereas Zach found a Chinese seller offering a year’s supply for $240—with Bitcoin as the only payment option. This price disparity led him to learn cryptocurrency for the transaction. Online sellers of such “research chemicals” or peptides often use Bitcoin because traditional payment systems avoid the regulatory and legal risks. Chainalysis data shows crypto payments to grey-market peptide sellers reached $32 million in Q1 2026, suggesting an annualized volume exceeding $100 million, though this remains a fraction of the estimated $1-3 billion black market. The trend intersects with communities like DeSciNYC, where events on peptides draw crowds interested in biohacking and longevity, indicating shared interests between crypto and wellness circles. However, potential FDA moves to legalize certain peptides could shift demand away from the black market and reduce the need for crypto payments.