【研报精选】frxETH大涨的背后

MirrorPubblicato 2023-02-02Pubblicato ultima volta 2023-02-02

Introduzione

frxETH是个很复杂的系统,且仍然在很早期的阶段,frxETH规模只不过stETH的~1%。

相较Lido/stETH,frxETH是一个复杂很多的系统。

  1. 用户行为十分动态,因为有自由的“两种选择”,只吃ETH2.0质押收益,亦或只吃Curve LP收益。
  2. 变量更多,协议方需要同时管理调控“两个收益率”,sfrxETH质押收益率以及Curve LP收益率。
  3. 通过调控“两个收益率”,协议方可以打造“两个卖点”来获得在LSD领域的竞争优势,即更高的质押收益率以及稳定的锚定。
  4. 打造“两个卖点”是有成本的,协议方可能需要经历长期的摸索,以寻求成本、规模、收益的合适平衡点,当下还很早期。
  5. 协议方亦有“秘密武器”助力降成本,在Frax稳定币业务上早已熟练运用多时。配合该“秘密武器”,在最大化利用Curve这件事上,无出其右。

以下展开论述。先上一图,可作为下文的辅助之用。

frxETH 系统结构示意图

用户的“两种选择”:从用户角度看frxETH的机制

用户凭ETH可在协议官网1:1铸造成frxETH,但这是单向的,frxETH无法通过官网赎回ETH。拿到frxETH后,有两个去处。其一,在官网stake成为sfrxETH,即可享ETH2.0的质押收益,目前在~7.3%的水平。其二,存入Curve frxETH/ETH pool吃LP收益,目前在~8.5%的水平(通过Convex)。这“两个收益率”正是协议方需要管理调控的所在。

顺道看下当前用户的选择,用户一共铸造了~50K frxETH,其中~25K stake成为了sfrxETH,~25K在Curve组LP。这也是sfrxETH质押收益率可以达到~7.3%的原因,因为组LP的frxETH把背后ETH资产的2.0质押收益让渡给了sfrxETH stake用户。

“两个收益率”之间的关系:Curve LP收益率与sfrxETH质押收益率

协议方调控Convex贿赂力度(Votium),可以影响Curve LP收益率。比如加大贿赂力度,使得Curve LP收益率显著高于sfrxETH质押收益率,这会吸引更多的frxETH选择做Curve LP,从而产生了拉平“两个收益率”的力,Curve pool TVL增大导致Curve LP收益率降低,选择做LP的frxETH比例增高带来了sfrxETH质押收益率的提高。

同理,若从高位Convex贿赂力度掉落,使得Curve LP收益率收益锐减,显著低于sfrxETH质押收益率,那么更多的frxETH会选择stake成为sfrxETH,拉平“两个收益率”的力开始作用。

所以我判断,这“两个收益率”不会出现长期的大的偏离。协议方调控的核心操作是每一期的Convex贿赂力度。

通过调控这“两个收益率”,协议方可以实现“两个卖点”,抢占LSD市场份额。

拆解“两个卖点”:更高的ETH2.0质押收益率、稳定的锚定

卖点之一即为超过市场平均水平的质押收益率。

若要实现sfrxETH质押收益率高于同行(就以stETH为例)Delta%,那就意味着需要有1 - 1/(1+Delta%)这么多比例的frxETH选择去Curve组LP。

比如,若要高过市场水平20%,那么就要驱动16.6%的frxETH选择去做LP。这就意味着协议方需要维持一个用户铸造的frxETH总数量 * 16.6% * 2 TVL量级的Curve pool,并且这个pool的LP收益率也大约在sfrxETH质押收益率的水平(即stETH收益率 * 120%)。协议方的Convex贿赂力度需要支撑起这个量级的Curve pool TVL和相应水平的LP 收益率。

提供一个比例上的参照,stETH总发行量近5百万,Curve stETH/ETH pool内的stETH和ETH分别为44万、31万。

卖点之二为稳定的锚定水平。主要的几个手段包括:更高的Curve pool A参数、更大的池子(相较日常的净frxETH->ETH交易量而言)。

frxETH在上线早期即获得了Curve pool A=120的设置,而市场龙头stETH pool的A仅有50。背后原因未细考,但frxETH这一点一上来便占据优势。

池子要大,主要手段是提高LP收益率,吸引更多的外部ETH持有人来组Curve frxETH/ETH pool LP。

可以看到,两个卖点背后意味着维持一个不小的Curve pool TVL和LP收益率的贿赂成本,并且如果sfrxETH质押收益率越高,成本会加倍拉高(Curve pool TVL、LP收益率双升)。

我想协议方在未来很长一段时间内,需要不断试探摸索,求得可接受的贿赂成本、满意的市场份额提升的一个平衡点。

最后,需要提一下协议方的“秘密武器”,这个武器可以帮助协议方降低贿赂的成本。

“秘密武器”:“凭空”铸造的frxETH

协议方是可以“凭空”铸造frxETH的。对协议方来讲,并不需要走常规的凭ETH铸造frxETH的流程,而是可以直接调用frxETH合约“凭空”铸造出frxETH。只要这些frxETH归属于协议,没有直接进入流通,就不会带来问题。事实上在Frax稳定币业务中,主要的发行量便源自这种铸造模式,我有一篇旧文解析了一下。

我们看一下frxETH的资产负债表,也有助于我们理解此事。

frxETH 资产负债表

负债一栏显示57K frxETH,但实际上用户存入的ETH只有~50K,有近6K便是协议方“凭空”铸造出来的,这6K躺在frxETH的资产栏的国库明细项内。“凭空”铸造并持有,对应到资产负债表,就是资产/负债两端同步增加相应数量,并无问题。

以上我说的~6K数字在资产负债表里并不能直接找到,这是因为协议方另留下了~3.6K ETH,也存入了Curve pool。我们看一下Curve pool的构成,会更加清晰。

Curve pool 来源构成

可以看到,协议方“凭空”铸造了5.7K ETH,此外未投入ETH2.0的~3.6K ETH,都在Curve里组LP。协议方目前拥有整个Curve pool ~15%的份额,这意味着有15%的LP收益会由协议方收取,进而摊薄了协议方的Convex贿赂成本(左口袋进右口袋的意思)。

这便是这个“秘密武器”的意义,通过“凭空”铸造frxETH存入Curve pool,降低Convex贿赂成本。

我设想了一个简化的场景,可以更加清楚地说明这个效力。

在该场景内,frxETH只打造一个核心卖点,就是稳定的锚定。Curve pool里的frxETH全部由协议方“凭空”铸造,ETH来自外部用户。用户铸造的frxETH全部stake成为sfrxETH,这也意味着在ETH2.0质押这一块,sfrxETH质押收益率只是市场平均水平。

此时,协议方可以吃走一半的LP收益,大幅降低了贿赂成本。如果再考虑到Convex贿赂的ROI经常>1,甚至有时超过2,这甚至可能是无成本的操作。这里不做展开,我曾计算过历史上两期Convex贿赂的ROI,有兴趣可参考。当然,这是一个简化的理想的局面,现实中不可能存在,因为协议方无法阻拦frxETH用户存入Curve组LP。

“凭空”铸造frxETH还有另一重意义,即可以帮助动态调控Curve pool的平衡。当frxETH -> ETH净交易量突然放大,导致池子出现不平衡的时候,协议方可以通过从pool中提出frxETH并销毁的方式,维持pool的平衡。

结语:一个很难预判走势的复杂系统

frxETH是个很复杂的系统,且仍然在很早期的阶段,frxETH规模只不过stETH的~1%。

目前只能很笼统地推断frxETH可能的破局点,也能大致地感受frxETH在Curve生态成本上的优势,但市场的未知之处相当之多,随便列举三四,质押收益率水平提高的边际效果是怎样的?品牌力弱于龙头带来的信任不足的影响有多大、需要多久能够弥补?缺少杠杆staking工具的影响有多大?从成本考量,Convex高额贿赂是否可持续?

更不要说上海升级后的新变数,很难预判退出staking的队伍会有多长,而这对Curve pool的管理有重要的影响。

大戏才刚刚开始,我们走着看。

Letture associate

Morning Post | Bitmine Plans to Raise $300 Million Through Preferred Stock Issuance; Polymarket Accuses Kalshi of Commercial Espionage

ChainCatcher's Daily Crypto Brief: Key developments from the past 24 hours include significant funding moves, regulatory actions, and market predictions. Bitmine announced a $300 million preferred stock fundraising. Polymarket accused rival prediction platform Kalshi of corporate espionage, citing numerous suspicious coincidences in product launches, a claim Kalshi strongly denied. The U.S. Department of Justice, in a joint "Disruption Week" anti-fraud operation with companies like Coinbase and Meta, froze over $3.8 million in cryptocurrency linked to scams. In infrastructure news, Macau completed its integration with the multi-central bank digital currency bridge, mBridge, aiming to build efficient cross-border payment channels. Cosmos Labs acquired the block explorer Mintscan. Market-wise, Geoffrey Kendrick, Standard Chartered's Head of Digital Assets Research, stated Bitcoin is nearing a bottom around $63,000, maintaining a year-end target of $100,000. He noted stability in U.S. spot Bitcoin ETF holdings. Ahead of SpaceX's anticipated IPO, internal insiders at Rocket Lab (RKLB) sold over $18.41 million in stock. In tokenization, Goldman Sachs partnered with Apex and Archax to launch a tokenized real estate fund. The meme token tracker GMGN reported the top trending tokens: on Ethereum, HEX, SHIB, LINK, PEPE, mUSD; on Solana, TROLL, swarms, WORLDCUP, neet, Buttcoin; and on Base, PEPE, toby, ODDS, ELSA, SKI.

链捕手13 min fa

Morning Post | Bitmine Plans to Raise $300 Million Through Preferred Stock Issuance; Polymarket Accuses Kalshi of Commercial Espionage

链捕手13 min fa

55TB to 28TB? The Rumor and Panic Behind Rubin's Memory Being Halved

Title: 55TB to 28TB? The Rumor and Panic Behind the Potential Halving of Rubin's Memory. On June 4th, a report from SemiAnalysis suggested NVIDIA's next-gen Vera Rubin NVL72 AI rack may ship with roughly 28TB of SOCAMM DRAM per rack instead of the anticipated 55TB, primarily using 96GB modules. This sparked a market panic, causing Micron's stock to drop over 10% on fears of halved memory demand. However, the article argues this panic is misguided for several key reasons. First, SOCAMM modules are socketed and upgradeable, not soldered. Lower initial configuration doesn't mean permanent demand loss. Second, the primary driver is a severe 2026 LPDDR5X supply shortage, not diminished need. NVIDIA is likely prioritizing rack shipments with available components. Third, with fixed total LPDDR5X supply, using less per rack could allow NVIDIA to ship *more* racks, not necessarily reducing overall memory orders. Micron's sharp drop was also attributed to a broader semiconductor sell-off triggered by Broadcom's earnings, with the SemiAnalysis report providing a convenient narrative for profit-taking after Micron's massive rally. In summary: the report on lower default configurations is likely accurate, but interpreting it as a demand collapse is wrong. The real risk for Micron lies in its reportedly minimal HBM4 share for Rubin, not in potentially flexible SOCAMM demand. The sell-off appears more like a correction amplified by coinciding negative catalysts.

marsbit30 min fa

55TB to 28TB? The Rumor and Panic Behind Rubin's Memory Being Halved

marsbit30 min fa

Exclusive from Yingke | Tang Wenbin's 'Yuanli Lingji' Merges with Logistics Robotics Company, and Secures Investment from Zhipu, SenseTime, Jieyue, and Others

Exclusive report: Embodied AI company "Yuanli Lingji" recently completed a new round of financing from major AI model firms including Zhipu AI, Stepfun, and SenseTime, alongside continued investments from industrial backers like Huaqin and SAIC Hengxu. Founded in March 2025 by Tang Wenbin, former co-founder and CTO of Megvii, Yuanli Lingji is a general-purpose embodied AI model company. In a notable move, the company has merged with logistics robotics firm "Atomix" (formerly known as Yuanli Juhe) through a share acquisition. Atomix, which originated from Megvii's logistics robotics business led by Tang in 2016 and was spun off in July 2024, has grown to become the world's second-largest supplier of pallet shuttle robots, with annual revenue nearing 1 billion RMB and over 500 projects globally for clients like Uniqlo and CATL. This merger aims to break the industry's "data deadlock" by combining Atomix's extensive real-world operational data from more than 20 countries with Yuanli Lingji's model training capabilities. The company's embodied AI model "DM0" utilizes a cross-domain training approach, integrating internet semantics, autonomous driving rules, and robotics data to achieve hardware-agnostic, precise manipulation even with a compact 2.4B parameter size. The collective investment from key AI players and the strategic merger signal a shift in the competitive landscape, as major model companies pivot from language tokens to physical actions ("from Token to Action"). The industry is entering a consolidation phase where hardware, AI models, data, and application scenarios converge to scale embodied intelligence, a trend mirrored by recent moves from giants like ByteDance and Skild AI.

marsbit38 min fa

Exclusive from Yingke | Tang Wenbin's 'Yuanli Lingji' Merges with Logistics Robotics Company, and Secures Investment from Zhipu, SenseTime, Jieyue, and Others

marsbit38 min fa

U.S. Stock Market Trends: Dow Hits New High, Nasdaq Falls, Whom Did Broadcom's Slap Wake Up?

U.S. Stocks Split: Dow Hits Record High as Nasdaq Slips; Broadcom's Plunge Sparks Rotation On June 4, the U.S. stock market saw a sharp divergence. The Dow Jones surged 875 points (+1.73%) to a record high of 51,561.93, while the Nasdaq Composite edged down 0.09%. The S&P 500 rose 0.41%. The primary catalyst was a sharp sell-off in AI-related chip stocks, led by Broadcom (AVGO). Despite reporting a 143% year-over-year jump in AI semiconductor revenue to $10.8 billion, the company's shares plunged about 14%. This was triggered by its maintained long-term AI revenue target, which failed to meet heightened expectations for a stock that had gained 55% this quarter and traded at a high P/E ratio. The slide dragged down the broader semiconductor sector and the technology板块. Conversely, money rotated into sectors like Healthcare (+3.14%), Financials (+2.67%), and Real Estate (+1.87%). UnitedHealth and Goldman Sachs were major contributors to the Dow's gains. The rotation was attributed to a search for value outside overheated tech names and a slight dip in Treasury yields. In other major news, SpaceX confirmed its IPO for June 12, targeting a record $75 billion raise at a ~$1.75 trillion valuation. Additionally, initial jobless claims rose to a four-month high, adding nuance to the labor market narrative ahead of the key May non-farm payrolls report. The day's action signaled that while the AI growth story remains intact, excessive valuations are prompting a market reassessment. Funds are moving, at least temporarily, from high-flying tech to more defensive and value-oriented sectors. The sustainability of this rotation hinges on upcoming economic data, particularly the jobs report, and the market's absorption of the massive SpaceX IPO.

marsbit41 min fa

U.S. Stock Market Trends: Dow Hits New High, Nasdaq Falls, Whom Did Broadcom's Slap Wake Up?

marsbit41 min fa

From 'Old Dogs' to 'New Darlings': How AI is Revaluing Old Infrastructure, from Dell to Nokia

"Old Dogs" Become AI's New Darlings: Revaluing Legacy Infrastructure The AI investment narrative is shifting. Beyond the spotlight on core chipmakers like Nvidia, a new wave of interest is rising for legacy tech companies—Dell, HPE, Nokia, Cisco, Corning, Western Digital—once labeled as slow-growth, outdated stories. This resurgence stems from AI's evolution from model development to real-world deployment, creating massive demand for physical infrastructure. As AI moves into data center construction and enterprise adoption, the focus turns to who can actually build and deliver complex systems. These established players hold decades of experience in supply chains, integration, networking, and enterprise delivery—assets now critical for scaling AI. The revaluation can be grouped into three key infrastructure areas: 1. **Servers & Integration (e.g., Dell, HPE):** They are becoming essential system integrators, transforming GPUs into full-scale AI servers with networking, power, and cooling, then delivering them to clients. Strong recent earnings and AI-specific revenue/order growth for Dell and HPE underscore this shift. 2. **Networking & Connectivity (e.g., Corning, Nokia, Cisco):** As AI clusters grow, high-speed data transfer becomes paramount. Corning benefits from fiber demand for data center links, Nokia is exploring AI-integrated wireless networks (AI-RAN), and Cisco sees surging orders for data center switches—all critical for efficient AI operations. 3. **Storage (e.g., Western Digital, Seagate):** The AI data explosion requires vast capacity. Beyond high-speed memory (HBM), there's growing need for high-capacity HDDs to store training data, logs, video, and cold/archival data cost-effectively. This revaluation, however, is not a blanket endorsement. True reassessment requires concrete proof: AI-driven orders and revenue growth, upward revisions to company guidance, and sustainable improvements in profit quality, not just top-line sales. In essence, AI is not turning all old tech firms into high-growth stocks; it is selectively re-pricing the "old assets" of companies that are mission-critical for building the new AI infrastructure, transforming their legacy capabilities into renewed growth engines.

marsbit50 min fa

From 'Old Dogs' to 'New Darlings': How AI is Revaluing Old Infrastructure, from Dell to Nokia

marsbit50 min fa

Trading

Spot
Futures
活动图片