edgeX (EDGE) Surge

The article "Harness Arbitrage Era: Rescuing DeFi from the SaaS Edge" explores the convergence of AI and decentralized finance (DeFi), arguing that AI-driven organizational models and token economies are surpassing DeFi and SaaS paradigms in efficiency. It traces how AI, particularly through agents and harness engineering, is quantizing human labor and organizational structures into scalable, automated systems ("Skill" extraction). Token consumption has exploded, driving data and content production costs toward zero, while SaaS models crumble as AI sells "work capability" rather than mere information. DeFi, though pioneering, has stagnated into a SaaS-like state—charging for transactions but failing to innovate fundamentally. The piece proposes that AI can reboot DeFi by enhancing security (e.g., via Mythos), optimizing capital efficiency, and redesigning token economics around verifiable, real-time returns rather than speculative value. Ultimately, AI agents learning from human behavior could autonomously manage DeFi protocols, making crypto tokens certificates of capital回报率 (return on capital). While AI reduces the value of data and repetitive labor, it opens new economic opportunities for individuals, reshaping finance through scalable, agent-driven automation.

Charles Hoskinson warns that Cardano risks losing its identity as the "science coin" if DReps fail to approve a key research funding proposal. He explains the ecosystem is in "treasury season" with a tougher funding environment, requesting $52 million this year versus $98 million last year, leading to cuts and layoffs. Hoskinson argues Cardano's research program is its "spine and backbone," responsible for its differentiation through work on proof-of-stake, extended UTXO, Plutus, and academic collaborations. He rejects calls to selectively fund parts of the research group, warning that treating researchers as commodities could lead to an irreplaceable talent drain to better-funded rivals. Hoskinson ties research funding to Cardano's long-term investment case, stating that without it, the project would lose a core value proposition. He ends with a direct appeal to DReps to vote for the proposal, calling it a necessary foundation for Cardano's future.

On June 2, the native token EDGE of the edgeX protocol experienced a violent flash crash, plummeting over 77% from around $1.14 to a low of $0.32 before partially recovering to approximately $0.64. edgeX officially ruled out a hack or platform security flaw, attributing the event to suspected market manipulation by external actors. The incident reignited community concerns regarding the project's transparency and past controversies. Prominent on-chain investigator ZachXBT highlighted that EDGE supply has long been controlled by a small group, with low circulating liquidity. This structure traces back to a disputed airdrop in April 2026, where the majority of tokens were allocated to partner wallets instead of the community, leading to accusations of a "well-designed rug pull." Despite having notable backers like Circle Ventures and Amber Group, and generating substantial protocol fees, edgeX faces criticism for a recurring pattern of "low float, high control, and opaque market making." While the team investigates the crash and emphasizes its core contracts are operational, user sentiment remains severely damaged, with many expressing a lack of trust in the team's integrity. Meanwhile, prediction markets have already opened betting pools on EDGE's future price.

The discussion, triggered by Anthropic's Claude Fable 5 model controversy, debates centralized control vs. decentralized distribution of cutting-edge AI. Participants explore the risks of models with hidden functionalities, government restrictions (like the Fable 5 block and Project Glasswing for Mythos), and the potential security chaos if powerful models like Mythos become widely available. The core tension lies between the need for safety/control (comparing AI to weapons) and the demand for open, permissionless access. The debate then shifts to the feasibility and future of decentralized AI (DeAI). While skeptics argue that decentralized training is inefficient, lacks data, and lags behind centralized R&D, proponents believe algorithm innovations and distributed computing (e.g., on consumer-grade hardware) can lower costs and democratize access. They argue DeAI's value lies in censorship resistance, privacy (e.g., Venice AI), and creating new economic models, not necessarily in outperforming frontier models. The conversation concludes by questioning whether DeAI's main opportunity is in challenging centralized model training or in building crucial supporting infrastructure and novel applications at the intersection of DeFi and AI.

Beyond the familiar performance charts like MMLU-Pro and MMMU, which major AI models strive to ace, stands a key "examiner": Chinese-Canadian researcher Wenhu Chen. An assistant professor at the University of Waterloo and founder of TIGERLab, Chen addresses the crucial need for more rigorous AI evaluation. As models like GPT-4 began scoring near-perfect results on older benchmarks like MMLU, it became difficult to distinguish their true capabilities. In response, Chen introduced MMLU-Pro in 2024, featuring harder, more reasoning-focused questions with more answer choices, successfully reintroducing meaningful performance gaps. His work extends to multi-modal evaluation with MMMU and its enhanced version, MMMU-Pro. These benchmarks test a model's ability to understand and reason with complex information from images, charts, and text across diverse academic subjects, exposing the significant challenges even top models face in genuine comprehension. Chen's background in complex QA, table reasoning, and his experience at Google DeepMind on projects like Gemini inform his approach. He understands that effective benchmarks must anticipate how models might "cheat" by memorizing data or avoiding visual analysis. His lab also actively researches video understanding and generation models (e.g., UniVideo, Vamba), ensuring his evaluation work is grounded in practical model-building challenges. Now at Meta's Super Intelligence Lab, Chen continues his focus on multi-modal data and evaluation, representing the deep yet often unseen contributions of Chinese talent in shaping the fundamental tools of the AI industry.