137Labs Global

On the evening of August 6, Unitree Technology set its STAR Market IPO issue price at 150.80 yuan per share, locking in a post-IPO market capitalization of approximately 61 billion yuan.
Within hours, the $UNITREE perpetual contract on Hyperliquid surged above $80, with implied valuations reaching $30–36 billion — a full four times higher.
The same company. Two markets. Completely different prices in a single night.
The A-share market delivered the official anchor. The crypto market front-ran the “first pure-play humanoid robotics stock” narrative.
This may be the most significant price-discovery event in the embodied intelligence sector in 2026.
Fundamentals: A Hardware Player That Has Already Achieved Scaled Profitability
In 2025, Unitree recorded approximately 1.708 billion yuan in revenue, up 335% year-over-year, with non-GAAP net profit attributable to the parent of about 600 million yuan and gross margins holding around 60%. These figures remain rare in the current humanoid robotics sector. The company shipped more than 5,500 pure humanoid robots in 2025 and has cumulative quadruped sales exceeding 30,000 units, maintaining a long-term leading global share. Revenue structure shifted decisively: humanoid robots accounted for over 50% of revenue for the first time, becoming the largest contributor.
Product strategy is clear. The G1 starts at around 99,000 yuan, while the R1 has been further reduced to the 29,900–39,900 yuan range. Through full-stack self-developed joint modules and motion-control algorithms, the company continues to drive down costs. Application scenarios remain dominated by research and education, while commercial and industrial orders are expanding in parallel. For the first half of 2026, the company expects revenue of 1.052–1.128 billion yuan, still growing 35–45% year-over-year. Growth has moderated, and R&D plus brand investment have increased — a natural stage of scaling and a key variable for whether current valuations can be sustained.
Sector Narrative Shift: From Hardware Performance to Embodied Models, From Demos to Real Deployment
The core narrative in the robotics sector has clearly shifted in 2026. In previous years, industry heat was driven mainly by hardware makers competing on motion performance, balance capability, and stage demonstrations. Entering 2026, the competitive focus has rapidly moved to embodied large models — world models, scene generalization, autonomous decision-making, and reinforcement learning. Capital continues to pour into model-related companies, with financing scales already far exceeding full-year totals of previous years.
Industry consensus is gradually clarifying: hardware first solves “can move, can sell, can scale,” while models solve “can understand environments and autonomously complete complex tasks.” At the same time, the narrative has moved from laboratory validation to real-world deployment. Official statements from the Ministry of Industry and Information Technology indicate that China’s full-year humanoid robot production in 2026 is expected to exceed 100,000 units. Chinese companies account for more than 80% of global shipments, forming a leading pattern represented by Unitree and AgiBot. Prices continue to fall into the ten-thousand-yuan range, while supply-chain maturity and capacity ramp-up have become high-frequency keywords. The overall stage assessment is that 2026 marks the first year of small-batch commercial use, with larger-scale mass production expected around 2028.
Founder Wang Xingxing: A Decade-Long Path of a 1990s Tech-Driven Builder
Wang Xingxing was born in 1990 in Yuyao, Ningbo, Zhejiang. He studied mechatronics at Zhejiang Sci-Tech University for his undergraduate degree and mechanical engineering at Shanghai University for his master’s. During college he started building robots with a 200-yuan microcontroller; in graduate school he self-developed the quadruped robot XDog, which attracted overseas attention. After graduating in 2016, he briefly joined DJI, then left within months, founding Unitree with roughly 2 million yuan in angel funding. In the early days the company consisted of almost only him; when funding dried up he paid employee salaries out of his own pocket.
He continues to serve as both CEO and CTO, maintaining a highly flat management style, directly tracking R&D progress, and personally participating in product definition and recruitment. Through dual-class shares with special voting rights, he will still control more than 65% of voting power after the IPO. At the current issue price, his stake is valued at approximately 20 billion yuan. This strong founder-led character and technology-driven DNA form Unitree’s clear distinction from purely capital-driven companies.
Key Points from the August 7 Online Roadshow
On the afternoon of August 7, Unitree held its STAR Market IPO online investor exchange meeting. Wang Xingxing and the management team directly addressed investor questions. Main points included:
Performance review: The company focuses on high-performance general-purpose humanoid robots, quadruped robots, components, and embodied intelligence models. In 2025, while non-GAAP net profit grew rapidly, the company achieved the world’s highest humanoid robot shipment volume. Since the mass production of the H1 in 2023, humanoid robot sales revenue rose from less than 3 million yuan to 868 million yuan, with its share of main business rising from 1.88% to 51.78%.Industry positioning: Current achievements are only a starting point. The embodied intelligence industry as a whole remains in an early development stage, similar to the early days of home computers. The generalization ability of devices in unfamiliar environments still requires joint improvement across the entire industry.Future direction: The company will continue to tackle key soft and hard technologies including embodied large models, scene data collection and analysis, reinforcement learning, embodied body models, self-developed core components, and high-performance actuators. It will actively explore more product forms such as humanoids, quadrupeds, and mecha, pushing robots to take on high-intensity, high-risk physical labor.Message to investors: The company hopes investors who buy its shares do so because they recognize its long-term value, rather than for short-term speculation.
The overall tone of the roadshow was pragmatic — showcasing existing commercialization results while clearly stating that the industry remains early-stage.
Humanoid Robot Shipments and Application Scenarios
Global humanoid robot shipments in 2025 reached approximately 13,000–18,000 units, with China accounting for 80–90%. Unitree shipped more than 5,500 units, with AgiBot following closely; the two companies together hold a significant share. China’s full-year production in 2026 is expected to exceed 100,000 units, and Unitree’s target is at least a doubling to the 10,000–20,000 unit range.
Current main application scenarios remain highly concentrated: research and education account for the highest proportion (40–70% at some companies), used by universities and research institutes for algorithm validation, data collection, and teaching; interactive services and data-collection scenarios rank next; entertainment performances and exhibitions take a certain share; industrial logistics, production-line operations, and inspection are accelerating but still represent a relatively low overall proportion. Truly large-scale entry into factories and homes still requires further cost reduction and improved model generalization capabilities.
Cross-Market Price Discovery and Observation Points
The A-share valuation of 61 billion yuan corresponds to a non-GAAP P/E of approximately 219x. Institutions fully accepted it with offline subscription multiples exceeding 2,600x. The Hyperliquid contract, meanwhile, delivered roughly a fourfold premium, pricing the scarcity of the “world’s first listed company whose core business is humanoid robots” together with China’s supply-chain scale advantages. The valuation gap between the two markets reflects differences in time horizon and risk preference.
Around the August 10 subscription date, three data sets are worth close tracking: concentration of holdings and large limit buy orders for $UNITREE on Hyperliquid; final online and offline allotment rates plus strategic placement lock-up arrangements; and the opening premium and trading volume structure on the first day of listing.
Unitree’s listing has formally moved embodied intelligence from laboratories and demonstration videos into the stage of tradable, benchmarkable assets. The A-share market provided the official 61-billion-yuan anchor, while Hyperliquid front-ran the industry’s imagination with a higher implied valuation. The pricing contest between the two markets itself constitutes one of the most noteworthy price-discovery processes of 2026.
Disclaimer: This article is for informational reference only and does not constitute any investment advice. The crypto market is highly volatile; investing involves risks. Please conduct your own research and bear the consequences independently.

I. Record-Breaking Earnings, But Why Did the Market React Negatively?
On August 6, SanDisk released its FY2026 fourth-quarter earnings report. From an operational perspective, this was a record-setting quarter, with revenue, profitability, and gross margin all reaching historical highs. However, despite the strong results, the stock came under pressure after the earnings release, mainly because market expectations for AI-driven storage demand had already been elevated significantly.
In the quarter, SanDisk generated $8.97 billion in revenue, below the market buy-side consensus estimate of $9.6 billion, but still representing 372% year-over-year growth and 51% sequential growth. From a revenue composition perspective, growth was not driven solely by higher shipment volumes. Approximately one-third of the growth came from increased shipments, while the remaining two-thirds came from NAND price increases and higher ASP (average selling price), indicating that the storage industry is currently benefiting from improving supply-demand dynamics and pricing recovery.
Profitability was even more impressive. SanDisk delivered GAAP net income of $6.903 billion and achieved a GAAP gross margin of 84.6%, further improving from 78.4% in FY26Q3, representing a 6.2 percentage-point sequential increase and marking the highest profitability level in the company’s history.
However, the market’s focus has shifted from “how much the company earned this quarter” to “whether earnings can continue exceeding expectations.” SanDisk’s FY27Q1 guidance came in below market expectations:
· Revenue midpoint: $10.6 billion
· Market consensus: $12.5 billion
· Gross margin midpoint: 84%
· Market expectation: 86%
Therefore, the market reaction was not a rejection of SanDisk’s current performance, but rather a reassessment of whether current profitability levels can continue expanding.
Historically, the NAND industry has followed a strong cyclical pattern: when supply becomes tight, prices rise and margins expand; however, as manufacturers increase capacity, oversupply often leads to price declines and margin compression.
The key question for investors is therefore:
Is SanDisk’s current record profitability a temporary peak in a traditional storage cycle, or is it the beginning of a structurally higher earnings profile driven by the AI era?

II. AI Is Reshaping SanDisk, With Data Centers Becoming the Core Growth Engine
The most important takeaway from this earnings report is not simply the record quarterly profit, but the fundamental transformation taking place in SanDisk’s business structure.
Historically, NAND demand was mainly driven by smartphones, PCs, and consumer SSD markets. These markets are highly cyclical and sensitive to consumer demand, inventory adjustments, and macroeconomic conditions, making storage companies difficult to value at premium levels.
However, the rapid expansion of artificial intelligence infrastructure is changing the demand landscape, with data centers becoming a major growth driver for SanDisk.
The company’s business mix has shifted significantly:
One year ago:
· Data center business Bit share: 12%
Currently:
· Data center business Bit share: 38%
In the latest quarter:
· Data center revenue: $2.98 billion
· Sequential growth: 103%
AI is fundamentally changing the storage industry.
Future AI training and inference workloads are not only dependent on GPU computing power but also require massive storage capabilities because AI development is increasingly:
· Memory intensive
· Storage intensive
As AI models become larger, datasets expand, and inference workloads accelerate, demand for high-capacity and high-performance storage solutions will continue increasing.
This means SanDisk is gradually transitioning from a traditional consumer storage company into an AI infrastructure supplier.
This transformation is critical because if future growth is increasingly driven by AI data centers rather than consumer electronics, SanDisk’s business model will become less dependent on traditional storage cycles and more closely linked to long-term AI infrastructure investment trends.

III. The Strategic Shift Behind 85% Gross Margin: Moving From Short-Term Profit Maximization to Long-Term Partnerships
One of the most important messages from SanDisk’s earnings call was management’s view on profitability.
The company indicated that:
A gross margin level around 85% represents a fair and reasonable return, and SanDisk does not intend to endlessly raise prices or maximize short-term profits by excessively squeezing customers.
Historically, semiconductor companies have often emphasized pricing power and attempted to maximize profitability during periods of tight supply.
SanDisk appears to be pursuing a different strategy.
The traditional storage industry model has been:
Supply shortage → Price increases → Higher margins → Capacity expansion → Industry downturn.
SanDisk is attempting to move toward a different model:
Long-term partnerships → Demand visibility → Stable pricing → Sustainable returns.
The significance of this strategy is that it aims to reduce the extreme cyclicality that has historically defined the NAND industry.
The market currently has two opposing views.
The bearish argument is that an 84%+ gross margin level may represent a cyclical peak, and profitability could decline once supply conditions normalize.
The bullish argument is that this cycle is fundamentally different because AI demand is structurally changing storage consumption, while SanDisk’s new business model improves revenue visibility and reduces exposure to market volatility.
Therefore, the key question is not whether SanDisk can continue increasing margins in the next quarter, but whether the company can convert today’s exceptional profitability into sustainable long-term cash flow.

IV. NBM: SanDisk’s Attempt to Break Away From the Traditional Storage Cycle
The most important strategic initiative behind SanDisk’s long-term transformation is its NBM (New Business Model).
The biggest challenge facing the NAND industry has always been demand uncertainty. Manufacturers historically struggled to accurately forecast how much capacity customers would need in future years, resulting in repeated cycles of overinvestment and price volatility.
The industry has traditionally followed this pattern:
Demand growth → Capacity expansion → Oversupply → Price decline → Production cuts → Recovery.
The purpose of NBM is to change this model by shifting storage from a “forecast-driven” business into a “demand-secured” business.
Under the NBM model, AI cloud customers provide SanDisk with long-term visibility into future storage requirements, while also offering financial commitments to support those agreements.
Currently, SanDisk has signed NBM long-term supply agreements with:
8 leading cloud data center and edge computing customers.
The agreements include:
· Minimum committed revenue: $93.9 billion
· Customer-provided cash and financial guarantees: $16.5 billion
· Average contract duration: more than 4 years
This gives SanDisk significantly greater revenue visibility and reduces reliance on spot NAND market pricing.
Importantly, NBM is not a low-margin volume strategy.
The company stated that:
· NBM contracts maintain approximately 80% gross margins.
Future capacity commitments include:
· More than half of FY2027 capacity already secured by customers
· Approximately two-thirds of FY2028 capacity already locked in by major customers
If NBM succeeds, SanDisk could transition from a traditional cyclical storage manufacturer into a supplier with long-term contracts, predictable revenue, and AI infrastructure exposure.

V. HBF Technology and Valuation Re-Rating: Can SanDisk Become a Core AI Storage Asset?
Beyond NBM, another major potential catalyst is SanDisk’s upcoming HBF (High Bandwidth Flash) technology.
The company plans to officially unveil the technology during its:
August 13 Investor Day.
The market is watching HBF because it could further enhance NAND’s role within the AI Memory Hierarchy.
Historically, NAND has primarily served as a data storage medium. However, as AI models become larger and computing architectures evolve, storage is becoming an increasingly important component of AI performance.
If HBF technology proves successful, NAND could expand beyond traditional applications such as:
· Data storage
· Enterprise storage
· Consumer SSDs
and potentially enter areas including:
· AI inference systems
· High-performance computing architectures
· Next-generation AI storage frameworks
If both NBM and HBF strategies succeed, SanDisk’s valuation framework could undergo a major shift.
Historically, investors viewed SanDisk as:
“A cyclical storage company.”
However, the market could increasingly view the company as an AI infrastructure asset supported by:
· Long-term AI storage growth
· Multi-year customer contracts
· Stable cash generation
· High free cash flow potential
· Share repurchase capability
This would represent a transition from a traditional cyclical semiconductor company toward a strategic AI infrastructure supplier.

VI. Risks and Conclusion: Short-Term Debate, Long-Term Business Model Transformation
Although SanDisk is benefiting from AI-driven storage demand, several risks remain.
1. AI Capital Expenditure Slowdown Risk
SanDisk’s long-term growth depends heavily on continued AI infrastructure investment from cloud providers.
If major cloud companies reduce AI spending, slow data center construction, or experience weaker-than-expected AI demand, storage demand and NBM execution could be affected.

2. NAND Price Decline Risk
Current earnings growth is significantly supported by NAND pricing improvement.
Revenue growth this quarter was driven by:
· One-third from shipment growth
· Two-thirds from price increases
If industry supply expands again and NAND prices decline, SanDisk’s gross margin could face pressure.

3. NBM Execution Risk
The market will continue monitoring:
· Whether customer demand remains strong;
· Whether long-term contracts are successfully executed;
· Whether high-margin economics can be maintained.

Conclusion
SanDisk’s FY26Q4 earnings can be summarized as:
Short-term neutral to slightly negative, but fundamentally positive over the long term.
In the short term, the stock faced pressure because revenue and margin guidance came below market expectations, while investors questioned whether record-high profitability had reached its peak.
However, from a long-term perspective, SanDisk is undergoing a significant transformation through:
· Rapid data center growth;
· AI-driven storage demand expansion;
· NBM long-term contracts;
· High-margin supply agreements;
· HBF technology innovation.
The key factor determining SanDisk’s future value is not whether gross margin can rise further next quarter, but whether the company can leverage the AI infrastructure revolution to transform itself from a traditional cyclical storage company into a technology supplier with long-term contracts, strong cash flow, and strategic importance in the AI ecosystem.
If the NBM model succeeds, SanDisk may not simply be benefiting from another storage upcycle — it may be redefining the future business model of the storage industry.
(Risk disclaimer: This article is for educational and informational purposes only and does not constitute investment advice. )

I. 最初の決算報告で強い成長が明らかに、しかし収益性への圧力は持続
公開上場後、最初の決算報告書が投資家に提供したのは、同社が「民間資金で支えられてきた航空宇宙のイノベーター」から「上場企業としてのテクノロジープラットフォーム」へ移行していく様子を、初めて包括的に見せる情報でした。10年以上にわたり、SpaceXは民間企業として急速に拡大し、ロケット開発、衛星の配備、商用の宇宙運用を支えるために継続的な資本注入に頼ってきました。その結果、これまで外部の投資家は、同社の実際の収益構造、収益性、キャッシュフローの実態について十分な見通しを持てませんでした。初めて公開された公的な財務結果の公表により、SpaceXの技術的優位がどのように商業価値へ転換されつつあるのか、市場はより明確な像を得られます。

最近、UniswapはWebアプリで「Launches（ローンチ）」タブを正式に公開することで、さらに大きな一手を打ちました。この機能は現在、Robinhood Chainを優先しており、複数のサードパーティのローンチパッドから新規トークンを集約し、そのままUniswapのインターフェース上で表示します。ユーザーは、異なるプラットフォーム間を切り替えることなく発見・取引できるようになります。
過去1か月を振り返ると、7月1日にUniswapがRobinhood Chainの主要なパブリックAMMになって以来、チェーンの取引量、手数料、新規トークンのローンチはいずれも大きな成長を遂げています。同時に、UNIの価格も、提携発表時の安値付近からほぼ60%上昇しており、市場はメインストリームの証券取引プラットフォームと深く結びついたこの流動性ネットワークを改めて価格付けし直し始めています。

— 長安（ChangXin）の追い上げの裏側で、HBMは次なる半導体競争の戦場になりつつある
導入：なぜDDR6の開発が、世界のメモリ業界の再価格付け（リプリシング）を引き起こしたのか？
ここ数年、人工知能の台頭は、グローバルな半導体産業の競争環境を根本から作り変えてきました。メモリはかつて、プロセッサやロジックチップの背後で支える部品だと見なされていましたが、いまやAI計算の性能を左右する最も重要な基盤の1つになりつつあります。

Alphabetが7月22日に米国市場の取引終了後に2026年の第2四半期決算を発表すると、その数値は批判の余地がほとんどないように見えました。売上高、営業利益、1株当たり利益（EPS）、そしてGoogle Cloudはいずれもウォール街の予想を大きく上回り、同社にとって近年で最も強い四半期決算の一つとなりました。しかし、ほぼ欠点のない成績表のように見えたにもかかわらず、Alphabetの株は時間外取引で3%超下落しました。これは、好調な業績と投資家のセンチメントの間に拡大するズレがあることを浮き彫りにしています。

その歴史の大半を通じて、韓国の代表的な株価指数であるKOSPIは、世界経済における「炭鉱のカナリア」と見なされてきた。
韓国は輸出への依存度が非常に高く、たとえばサムスン電子、SKハイニックス、現代自動車、LGケミカルなどの企業は、半導体、自動車、電池、家電製品に関する世界のサプライチェーンに深く組み込まれている。世界の製造受注が増え、電子機器の需要が改善すると、韓国の輸出や株式はしばしばいち早く持ち直す。世界貿易が弱まると、韓国は圧力を最初に感じる市場の一つでもある。

I. なぜKimiは突然、新規ユーザーのサブスクリプションを停止したのか？
7月中旬、Moonshot AIがオープンウェイトの大規模言語モデル「Kimi K3」をリリースしてからわずか2日後、同社はすぐに中国のAI業界で最も注目を集める話題の一つとなった発表を行いました。計算資源の高い稼働率が継続していることから、Moonshotは一時的に新規の一般向けサブスクリプションを停止するとし、現行の支払い会員向けのGPU容量を優先しつつ、追加の計算クラスターの導入を前倒しすると説明しました。従来のインターネット時代に見られた、サーバ障害やデータベースの問題などでサービスが停止するような障害とは異なり、今回の説明には生成AI時代を象徴する特徴が反映されていました。つまり、製品自体は意図どおりに機能している一方で、モデルへの需要が、それを提供するために必要なインフラを上回ってしまったのです。

7月14日と15日の2日間、FRB議長ケビン・ウォーシュは連邦議会に出席した。彼は下院の金融サービス委員会および上院の銀行・住宅・都市問題委員会に対し、FRBの半年ごとの金融政策報告書について証言した。
ウォーシュが就任後初めて行った連邦議会での証言として、今回の公聴会は世界の金融市場から大きな注目を集めた。
投資家にとって、公聴会は米国経済の広範な見通し以上のものを提供した。さらに、金融政策の今後の方向性を検討する重要な機会にもなった。2日間にわたり、ウォーシュはインフレ、金利、FRBのバランスシート、中央銀行の独立性、そして人工知能が経済に与える影響について言及し、いくつかの注目すべき政策シグナルを発信した。

2026年7月16日、世界有数の純粋ファウンドリー（専業製造）企業である台湾セミコンダクター・マニュファクチャリング・カンパニー（TSMC）が、第2四半期の決算レポートを発表した。この一連の結果が、世界の資本市場、半導体サプライチェーン、そしてテクノロジー業界の経営幹部たちからこれほどまでに注目を集めた理由は、単にTSMCが規模の非常に大きく、戦略的に重要な企業であるからだけではない。人工知能（AI）インフラ向け支出の継続的な拡大を背景に、先端プロセス技術における競争が激化し、かつグローバルな半導体サプライチェーンの再編が加速する中で、TSMCは投資家や業界関係者が、AI需要の実力、先端チップの生産ボトルネック、そしてテクノロジー・セクターの設備投資（キャピタルエクスペンディチャー）のより広範な景況感を見極めるための、最も重要な窓の一つとなっている。

今日、長城新技術（Changxin Technology）に関するニュースに圧倒されましたか？7月15日前後に、Trade.xyzのHIP-3プロトコルに基づくHyperliquidプラットフォームがCXMT無期限先物契約をまもなくローンチするというニュースが、投資グループチャットやソーシャルメディアで瞬く間に大きな話題を独占しました。
中国最大、世界で4番目のDRAMメモリーチップメーカーである長城新技術（Changxin Technology）は、7月14日に1株当たり8.66人民元の発行価格を決定しました。7月16日にオンラインおよびオフラインの申込みを開始し、上海証券取引所のSTAR市場に7月27日前後で上場する見通しです（銘柄コード：688825.SH）。これは中国の資本市場における2026年の最も重要なIPOイベントであるだけでなく、STAR市場史上で2番目に大きく、今年のA株における最大の新規株式公開でもあります。計画調達額は2950億人民元であり、今年のアジア最大級のIPOの1つとされています。

7月14日（現地時間）に、米国労働統計局（BLS）は2026年6月の消費者物価指数（CPI）報告書を公表した。世界の金融市場で毎月最も注目されるマクロ経済指標の一つであるCPIは、米国全体の消費者物価の変化を映し出すだけでなく、連邦準備制度（FRB）の今後の金融政策方針を評価するための重要な基準としても機能する。その結果、投資家、エコノミスト、政策立案者、そして市場参加者は皆、この発表を注視していた。最新の数値は、総合インフレが前月までと比べて大幅に落ち着いたことを示しており、ヘッドラインCPIとコアCPIのいずれも市場予想を下回った。データは直ちに市場のセンチメントを押し上げ、主要な米国株式指数は上昇し、国債利回りは低下し、米ドルは弱含んだ。これは、投資家がFRBによる追加的な金融引き締めへの見通しを引き下げたためである。とはいえ、楽観が再燃する中で、より根本的な問いが浮上している。今回の改善は、米国のインフレがついにコントロール下に入ってきたことを示すのか、それとも短期要因による一時的な低下を反映しているだけなのか。

はじめに
過去2年間、人工知能に関する市場の議論は主として、モデルの能力が今後も向上し続けられるのかどうか、生成AIがインターネットやスマートフォンに匹敵する汎用技術になり得るのか、そして大規模言語モデルが検索、ソフトウェア、広告、エンタープライズ向けサービスといった産業をどのように変革し得るのかといった点に焦点が当てられてきました。しかし、モデルの性能が進歩するにつれてユーザーの採用が加速し、世界的な計算基盤（コンピューティング・インフラ）が集中的な建設フェーズに入ると、投資家の関心は次第に商業的なリターンへと移っていきました。そこで現在の中心的な問いは、大手テック企業がAIに投じている膨大な資本を、最終的に安定した収益、利益、そしてフリー・キャッシュフローへと転換できるのかどうかです。

韓国のメモリ大手SKハイニックスが米国上場のデビューを果たしてから3日が経った。この記事では、この出来事を4つの主要な観点から振り返る。すなわち、SKハイニックスの上場概要と事業の強み、メモリー分野を押し上げるAI向け設備投資の高まりによる支援、ゴールドマン・サックスによるAIバリューチェーンの分解、そして将来を見据えた示唆とともに見るSKハイニックスの戦略的価値である。
SKハイニックスはADRを149ドルで価格設定した。取引はSKHYVで始まり、その後SKHYに切り替わった。同社は、外国企業による史上最大規模の米国IPOである最大26.5 billionドルを記録的に調達した。ADRは初日に約13%急騰し、時価総額は1.2兆ドル超に押し上げられた。調達資金は主に韓国でのウェハ工場の拡張と、HBMの増産体制を加速するための先端EUV装置の買収に充てられる。NVIDIAやその他の高性能AIアクセラレータの重要なサプライヤーとして、同社の製品はすでに世界のデータセンターのサプライチェーンにおいて重要な位置を占めている。

I. 新しい市場を生み出した利子率サイクル
2021年にさかのぼると、現実世界の資産（RWA）の概念は、まだほとんど理論的な段階にありました。RWAとは、伝統的な経済における資産をブロックチェーン上で表現し、それらをトークン化された形で発行・保有・移転できるようにするプロセスを指します。しかし当時、その考えはまだ、魅力的な商業市場へと発展していませんでした。一方では、米連邦準備制度（FRB）が長年にわたり金利をほぼゼロに保っていたのに対し、米国の短期国債の利回りは1%未満にとどまり、オンチェーンへ移るほど魅力的な、低リスクの収益が従来の金融にはほとんど存在しませんでした。もう一方では、暗号資産業界が急速に拡大しており、DeFi、NFT、GameFi、その他の投機的な分野が目を見張るようなリターンを生み出していました。投資家は、年あたり数％の収益を提供する商品よりも、10倍、あるいは100倍に値上がりし得る資産に強い関心を寄せていました。こうした状況では、RWAは実現可能な金融ビジネスというより、技術的な方向性に近く見えていました。

2026年7月8日、Binance Researchは「Stablecoins: Transforming The Financial Landscape（ステーブルコイン：金融の風景を変える）」と題した詳細なレポートを発表しました。このレポートでは、ステーブルコインが暗号資産エコシステムにおける単なる架け橋にとどまらず、グローバルな金融システムを再構築する中核インフラへとどのように進化してきたのかを、体系的に検証しています。
この記事では、レポートの主要な洞察、重要なデータ、そして新たに浮上するトレンドを詳しく分解して解説し、読者が、ステーブルコインに起きている構造的な変化と、それが個人、機関、そしてより広い金融環境にもたらす深い影響を理解できるようにします。

7月1日にパブリックなメインネットをローンチしてからわずか1週間後、Robinhood Chainの総ロック資産（TVL）は1億ドルの節目を突破しました。24時間の上昇率がある時点で最大159%に達し、ピークは約1億600万ドルでした。DeFiの貸出プロトコルによって後押しされ、ミームコイン取引の流動性によってさらに加速したこの爆発的な成長は、新たなレイヤー2チェーンをいち早く注目の的へと押し上げました。
7月8日、オンチェーンのミームコインが時価総額を一時的に1億1000万ドル超まで押し上げたものの、その後約1億400万ドルまで後退しました。24時間で13.9倍超の値上がりを記録（ピーク時にはさらに上振れ）し、日次の取引高は数億ドルに達しました。Robinhood Chainの24時間のDEX総取引高は5億ドル超に急増し、市場で最も熱い話題の1つとなりました。

連邦準備制度（FRB）の6月FOMC議事要旨が公表されたことで、投資家の関心は「利下げはいつ始まるのか」というおなじみの問いから、より根本的な問題へと移っています。つまり「米国経済の次の局面で、何がインフレを押し上げているのか？」です。
政策当局者は全員一致で金利を据え置くことを決めた一方、議事要旨は、インフレ圧力が消え去るのではなく、むしろ変化しながら続いていることへの懸念が高まっていることを示しています。賃金や個人消費に加えて、FRB当局者は今後の持続的なインフレの要因として、次の3つの新たな源泉をより強く指摘するようになっていました。すなわち、人工知能（AI）への投資、上昇するエネルギー価格、そしてより高い関税です。