137Labs Global

Em julho de 2026, a Robinhood revelou uma das expansões de produtos mais ambiciosas de sua história no evento "The World is Flat" em Londres. A empresa lançou oficialmente a Robinhood Chain, sua blockchain de Camada 2 construída sobre o Arbitrum Orbit, juntamente com uma série de novos produtos, incluindo ações tokenizadas, empréstimos descentralizados, agentes de negociação com IA, futuros perpétuos e uma estratégia de expansão global acelerada.
À primeira vista, pode parecer apenas mais uma plataforma financeira lançando sua própria blockchain. Mas, ao olhar com mais profundidade, a Robinhood está tentando algo muito maior: se transformar de uma corretora online na camada de infraestrutura para a próxima geração das finanças globais.

Introduction
In early July, reports that Meta was building a cloud computing business and preparing to sell AI compute capacity to external customers triggered an unusually sharp reaction across the AI infrastructure sector. The market response was strikingly asymmetric: Meta's shares surged, while AI compute rental companies such as CoreWeave and Nebius suffered significant losses. At the same time, nearly the entire AI hardware ecosystem—including AMD, Micron, SanDisk, ASML, TSMC, Samsung Electronics, and SK hynix—came under broad selling pressure. On the surface, this appeared to be nothing more than another technology company expanding into a new line of business. In reality, however, what the market was pricing was not whether Meta intended to commercialize its GPU resources, but whether one of the fundamental assumptions that has underpinned the AI industry over the past two years was beginning to change.
For the past two years, investors have largely embraced a single narrative: compute has been the primary bottleneck of the AI era. The companies capable of securing the most GPUs, building the largest data centers, and committing the highest levels of capital expenditure were widely regarded as the future winners. As this narrative became increasingly entrenched, valuations across the AI ecosystem came to reflect the same underlying assumption—that AI compute would remain structurally scarce, hyperscalers would continue expanding capital expenditure at an unprecedented pace, and upstream suppliers of GPUs, high-bandwidth memory (HBM), enterprise SSDs, servers, power infrastructure, and networking equipment would all enjoy sustained secular growth. Meta's latest move, however, introduces an entirely different question. If data centers are no longer built solely for internal consumption, but can also be commercialized as external infrastructure, is the industry beginning to shift its focus from continuously expanding capacity toward maximizing asset utilization? If that transition is indeed underway, then the defining competitive advantage of the AI industry may no longer be the ability to build infrastructure, but the ability to operate it efficiently.

I. Why Does Meta Need a Second Commercial Path for Its AI Investments?
Viewed in isolation, the news could easily be interpreted as Meta making a late entry into the cloud computing market. In reality, however, cloud computing itself is not the primary driver behind this strategic shift. Rather, the underlying catalyst is the unprecedented scale of capital expenditure required by the AI era.
Over the past several years, Meta has become one of the most aggressive investors in AI infrastructure globally. From continuously expanding hyperscale data centers and purchasing tens of thousands of high-end GPUs to repeatedly raising its annual capital expenditure guidance, the company has committed virtually every available resource to AI. Unlike Microsoft, which owns Azure, Amazon, which operates AWS, or Google, which has Google Cloud, Meta has never possessed a large-scale enterprise cloud business capable of directly monetizing infrastructure. Its data centers have historically served internal workloads, including advertising recommendation systems, social media platforms, content distribution, and the training of Llama models. In other words, these assets have functioned primarily as internal production infrastructure rather than commercially monetizable products.
As capital expenditure has grown from tens of billions of dollars to well over one hundred billion dollars annually, Meta has found itself facing not only technological challenges but also increasing pressure from capital markets to demonstrate an acceptable return on investment. AI undoubtedly continues to improve the efficiency of Meta's advertising business, but whether incremental gains in advertising performance alone can justify such extraordinary levels of fixed-asset investment has remained an open question for investors. Against this backdrop, Meta's decision to commercialize AI compute should not be interpreted as a sudden ambition to become another cloud provider. Rather, it represents an effort to establish a second monetization pathway for AI-era capital expenditure. Once an asset that was previously dedicated solely to internal operations acquires the ability to generate independent cash flow, its role within the company's financial profile fundamentally changes—from a cost center into a revenue-generating asset.
Viewed from this perspective, what Meta is ultimately commercializing is not its GPUs themselves, but the enormous capital investment that those GPUs represent.
II. What Is Meta Really Selling?
Many initial market interpretations reduced this development to a simple conclusion: Meta is beginning to rent out its GPUs. In reality, however, GPU rental is likely to be only one component of a much broader strategy.
Based on the information currently available, Meta appears to be building a comprehensive AI infrastructure offering rather than simply providing raw compute capacity. This ecosystem would likely include GPU computing resources for developers and enterprise customers, fully managed large language model inference services, enterprise-grade model hosting, model fine-tuning capabilities, and, over time, an integrated runtime environment designed to support AI agents. From a business model perspective, Meta's positioning appears to fall somewhere between AWS Bedrock, Azure AI, and AI-native cloud providers such as CoreWeave, rather than representing a straightforward attempt to replicate a traditional public cloud platform.
This distinction is important because it suggests that Meta's competitive advantage does not lie in enterprise IT infrastructure or decades of accumulated cloud services expertise. Instead, its strength comes from the hyperscale AI infrastructure it has already built for its own products. Over the past several years, Meta has invested heavily in optimizing AI training, recommendation systems, and inference deployment across Facebook, Instagram, WhatsApp, and the Llama ecosystem. These engineering capabilities, once developed exclusively for internal use, now have the potential to become commercial products available to enterprise customers. Put differently, Meta is not merely selling GPU capacity—it is commercializing a mature AI infrastructure platform that has already been tested and validated at internet scale.
If this model ultimately succeeds, Meta's data centers will no longer function solely as the company's internal backend infrastructure. Instead, they could gradually evolve into infrastructure assets capable of generating recurring revenue in their own right.

III. Why Is the Market So Sensitive to This Move?
The most significant implication of Meta's announcement is not whether the company can become the next AWS. Rather, it is that Meta has, for the first time, openly framed AI infrastructure as a commercial asset that can be operated—not merely built.
Over the past two years, the valuation framework for the AI industry has rested on a relatively straightforward assumption: demand will continue to expand at such an extraordinary pace that building more infrastructure is inherently the right strategy. More GPUs were always viewed as better. Larger data centers were always considered an advantage. Higher capital expenditure was interpreted as a signal of future growth because investors believed that ever-increasing training and inference demand would eventually absorb all available computing resources.
Meta's willingness to discuss selling excess AI compute introduces a fundamentally different possibility. The industry's most important question may no longer be whether companies possess enough GPUs, but whether those GPUs can maintain sufficiently high utilization over time.
These represent two very different economic models.
In the construction phase, success is measured by capital deployment.
In the operational phase, success is measured by asset returns.
During the construction phase, companies compete on procurement capability.
During the operational phase, they compete on utilization efficiency.
During the construction phase, investors ask how many GPUs a company owns.
During the operational phase, they ask how much revenue each GPU can generate over the course of a year.
This shift should not be interpreted as evidence that AI demand has peaked, nor does it imply that GPUs have become oversupplied. Instead, it reflects a natural evolution of the industry. As AI infrastructure reaches unprecedented scale, capital markets are beginning to demand proof that these increasingly expensive assets can produce sustainable cash flow, rather than relying indefinitely on the assumption that future demand alone will justify continued investment.
From this perspective, Meta's decision to commercialize AI compute may ultimately be remembered less as a cloud computing initiative than as a symbolic milestone marking the AI industry's transition from an era of infrastructure expansion to one of infrastructure operation.
IV. Why Did the Entire Market Sell Off?
To understand the broad sell-off that followed Meta's announcement, it is essential to distinguish between direct and indirect impacts.
The companies most directly affected were AI-native cloud providers such as CoreWeave and Nebius. Over the past several years, their competitive advantage has largely been built around a straightforward business model: acquiring large quantities of GPUs, building specialized AI infrastructure, and renting compute capacity to AI companies at a premium. Meta, however, possesses data center capacity on a comparable scale, significantly stronger financial resources, and, perhaps more importantly, substantially lower procurement costs. Once one of the world's largest buyers of AI infrastructure begins positioning itself as a potential supplier of AI compute, the core investment thesis behind the Neocloud model inevitably comes under scrutiny. As a result, companies operating within this segment became the most heavily impacted names following the announcement.
By contrast, the declines seen across GPU manufacturers, HBM suppliers, and the broader semiconductor sector were driven less by immediate changes in business fundamentals than by a reassessment of future expectations. Investors began to question whether hyperscalers might eventually shift their focus from continuously expanding data center capacity toward maximizing the utilization of infrastructure they had already built. If that were to happen, future GPU procurement, HBM demand growth, and overall AI infrastructure capital expenditure could all prove less aggressive than the market's most optimistic assumptions had previously implied. Such concerns are unlikely to affect NVIDIA's, AMD's, TSMC's, or Micron's order books overnight. What they affect first is valuation, particularly for companies whose multiples have been built upon expectations of sustained hyper-growth.
The storage industry deserves separate consideration.
Over the past year, HBM has been one of the biggest beneficiaries of the AI training boom, leading many investors to group all memory and storage companies under a single AI investment narrative. In reality, however, the hardware requirements of AI training and AI inference differ in important ways. During the training phase, the primary bottlenecks are computational throughput and memory bandwidth, making GPUs and HBM indispensable components. During the inference phase, by contrast, technologies such as retrieval-augmented generation (RAG), AI agents, long-context models, vector databases, and KV cache management require vast amounts of data to be accessed continuously, efficiently, and at extremely low latency. Under these workloads, high-performance enterprise SSDs become increasingly important.
From this perspective, the inference era should not necessarily be viewed as one in which storage demand declines. Instead, it is more accurately understood as a period in which the composition of storage demand evolves.
Consequently, the simultaneous sell-off in companies such as Micron, SanDisk, Samsung Electronics, and SK hynix reflected not only the broader correction across AI-related equities, but also the possibility that investors were applying a training-era investment framework to an industry that is gradually transitioning toward inference. Whether these companies ultimately prove to be fundamentally weaker or merely become victims of indiscriminate selling will depend largely on how rapidly inference workloads continue to expand and whether enterprise SSD demand materializes at the scale many industry observers now anticipate.

V. Why Operational Excellence Will Become the Next Competitive Advantage
The greatest significance of Meta's decision to commercialize AI compute does not lie in the addition of another revenue stream. Rather, it lies in the fact that the company has prompted the market to recognize, perhaps for the first time, that AI infrastructure is beginning to transition from an era of capital deployment to one of operational management.
Over the past several years, investors have focused overwhelmingly on metrics such as GPU counts, data center capacity, capital expenditure, and model size because the industry has remained firmly in its infrastructure build-out phase. As that infrastructure matures and an increasing number of large-scale data centers become operational, however, the factors that determine competitive advantage are beginning to change. In the years ahead, technology companies are likely to compete less on their ability to acquire additional GPUs and more on their ability to maximize GPU utilization, reduce unit computing costs, generate sustainable cash flow, and build long-term commercial ecosystems around their infrastructure assets.
This transition also implies that the valuation framework for the AI industry is evolving. During the construction phase, capital markets tended to reward companies willing to invest aggressively because higher capital expenditure was viewed as a direct indicator of future growth. As the industry moves into a more mature stage, investors are likely to place greater emphasis on measures such as return on capital, asset utilization, inference revenue, enterprise customer adoption, and the efficiency with which infrastructure is monetized. Ultimately, operational excellence—not simply the scale of investment—will determine whether these extraordinarily expensive assets can generate durable long-term value.
Meta's latest move is unlikely to reshape the cloud computing landscape overnight, nor is it likely to challenge the positions of AWS or Azure in the foreseeable future. What it does signal, however, is a meaningful shift in the competitive dynamics of the AI industry. Competition is gradually moving away from a race to accumulate the greatest amount of infrastructure toward a race to generate the greatest economic value from that infrastructure.
If the past two years represented the first half of the AI infrastructure cycle—a period defined primarily by construction and expansion—then the years ahead may well represent its second half, one defined by operation, monetization, and capital efficiency. Meta may simply be the first major technology company to take that step.

VI. Conclusion
In the short term, Meta's decision to sell AI compute should be viewed primarily as a catalyst for market repricing. It has reshaped investor expectations surrounding the competitive landscape for Neocloud providers while prompting a broader reassessment of AI infrastructure demand, capital expenditure trajectories, and long-term growth assumptions across the hardware supply chain. Yet viewed through a longer-term lens, the more important question is not whether Meta can become another AWS, but what this decision reveals about the future economics of AI infrastructure. As capital expenditure reaches unprecedented levels, simply owning more GPUs or building larger data centers is no longer sufficient to justify premium valuations. Investors are increasingly demanding evidence that these capital-intensive assets can generate sustainable cash flows and attractive long-term returns. In the years ahead, market attention is likely to shift away from GPU counts, data center scale, and model parameters toward metrics such as infrastructure utilization, inference-driven revenue growth, enterprise adoption, and return on invested capital. From competing on investment to competing on operations, and from competing on construction to competing on returns, Meta's decision to commercialize AI compute may ultimately prove to be more than a strategic business expansion. It may instead mark a defining moment in the evolution of the AI industry—one in which AI infrastructure begins to be valued not simply for its scale, but for its ability to generate enduring economic value.

Yesterday, a new stablecoin announcement quickly dominated discussions across the crypto community and U.S. equity investors. More than 140 companies and institutions jointly introduced Open USD (OUSD), while Circle (CRCL) shares immediately fell by approximately 17.5%. At the same time, the latest Russell index rebalancing triggered additional selling pressure from passive funds. This event goes beyond the launch of a single product. It marks the accelerating integration of stablecoins from crypto-native tools into mainstream financial payment infrastructure, while prompting the market to reassess the real competitive impact of traditional financial giants entering the space. Below is a comprehensive breakdown of the event, OUSD, and its implications from multiple perspectives.
Complete Timeline of the Event
On June 30, the Open Standard Alliance officially published its introduction to OUSD. The alliance consists of more than 140 members, including traditional banks, payment networks, asset managers, technology companies, and crypto projects. OUSD is positioned as a U.S. dollar stablecoin designed for the internet economy. It features zero minting and redemption fees, reserve yield sharing, and a partner-governed model. The stablecoin is scheduled to launch on chains including Solana and Base in the second half of 2026.
At nearly the same time, FTSE Russell removed Circle (CRCL) from five major Russell Growth Indexes during its latest annual index reconstitution, including the Russell 1000 Growth, Russell 3000 Growth, and Russell Midcap Growth indexes. These indexes serve as key benchmarks for passive investment globally and are closely tracked by numerous index funds and ETFs. Any constituent adjustment automatically triggers mechanical buying or selling by passive funds, directly affecting stock liquidity.
The combination of the OUSD announcement and the index removal led to an approximately 17.5% decline in CRCL's share price in a single trading session, wiping out roughly $3.6 billion in market value. Circle founder Jeremy Allaire responded by emphasizing USDC's existing scale and adoption, while Tether expressed an open attitude toward additional competition. The event sparked extensive discussion across social media and trading platforms, leading investors to reassess the competitive landscape of the stablecoin industry.
OUSD: Background, Reserve Structure, and Key Participants
Open Standard is an open infrastructure alliance jointly established by enterprises with the goal of building a stablecoin system better suited for commercial and internet use. The alliance brings together participants from traditional finance, payment technology, and crypto, representing a practical step toward integrating mainstream finance with blockchain technology.
Its core members include major financial institutions and technology companies such as BlackRock, Visa, Mastercard, BNY, Standard Chartered, Stripe, Google, Samsung Electronics, IBM, and Shopify, alongside crypto and Web3 participants including Coinbase, OKX, Bybit, Bitget Wallet, Ripple, Crypto.com, Fireblocks, Gemini, MetaMask, Aave, Solana, and Base.
At present, OUSD's official documentation has not disclosed detailed information regarding its reserve composition. Based on the alliance's overall design philosophy, reserves are expected to primarily consist of highly liquid, low-risk assets such as U.S. dollars and short-term U.S. Treasury securities, likely managed by professional institutions including BlackRock. This approach is similar to USDC's transparent reserve model, while placing additional emphasis on reserve yield sharing. More details regarding reserve composition and audit arrangements are expected after the official launch.
Key Features and Advantages of OUSD
OUSD's design focuses heavily on practicality.
The first feature is zero minting and redemption fees, which significantly reduce transaction costs for institutions and high-volume users.
The second is reserve yield sharing. Partners receive a portion of the reserve interest income, strengthening incentives for ecosystem participation and encouraging broader institutional adoption through aligned economic interests.
The third is its extensive ecosystem support. Payment infrastructure from Visa and Mastercard, asset management capabilities from BlackRock, merchant integration through Stripe, and liquidity support from crypto platforms including Coinbase, OKX, Bybit, and Bitget Wallet provide OUSD with strong credibility and circulation potential from day one. Its alliance-based governance model also makes it attractive to developers and enterprises seeking an open architecture.
These features make OUSD particularly appealing for cross-border payments, corporate settlement, and internet commerce, especially for institutions looking to reduce costs while sharing reserve-generated returns.
The Practical Challenges Facing OUSD
Any new project ultimately requires execution to prove itself.
Decision-making across a large alliance requires coordination among many stakeholders, which may slow execution compared to a single-company structure. Balancing the interests of numerous participants will also require continuous coordination.
Meanwhile, USDC and USDT have already established strong network effects. To gain meaningful market share, OUSD must rapidly build real-world use cases and liquidity. As the project remains in its early preparation stage, the effectiveness of its post-launch execution will become one of the market's primary areas of focus.
How OUSD Could Impact CRCL
Increased Short-Term Competitive Pressure
OUSD's zero-fee model and reserve yield sharing directly challenge USDC's reserve interest revenue model. Investors have become increasingly concerned about Circle's future profitability. Combined with passive selling resulting from Russell index removal, these factors contributed to the sharp decline in CRCL's stock price.
Market Expansion Driven by Institutional Participation
The participation of traditional giants such as Visa, Mastercard, and BlackRock could significantly accelerate stablecoin adoption across mainstream payments and institutional finance, expanding the industry's total addressable market. As a publicly listed and regulated stablecoin issuer, Circle could ultimately benefit from participating in a much larger market.
Valuation and Trading Characteristics
Short-term sentiment has increased CRCL's volatility while also bringing valuations back toward more rational levels. If Circle continues expanding its payment network, blockchain infrastructure, and enterprise services, its Q2 earnings could become an important catalyst for recovery.
Long-Term Moat Under Review
The alliance model behind OUSD tests Circle's pace of innovation and ecosystem stickiness. However, USDC's existing circulation of tens of billions of dollars, mature application ecosystem, and regulatory advantages remain significant competitive buffers.
Overall, OUSD presents a combination of short-term headwinds for CRCL and long-term positives for the stablecoin industry. The ultimate outcome will largely depend on Circle's execution capabilities and continued demand growth for regulated stablecoins.
A Broader Look at the Stablecoin Industry
Stablecoins are gradually becoming one of the most important bridges connecting traditional finance with blockchain technology. The emergence of OUSD further accelerates this transition as banks and payment institutions begin actively participating in industry standards and infrastructure development.
From a business model perspective, zero fees combined with reserve yield sharing introduce a new incentive structure. This model encourages broader ecosystem participation, lowers service costs, and supports wider adoption.
Regulatory and macroeconomic conditions continue to shape the industry. Regulatory frameworks are creating clearer paths for compliant stablecoins, while increased competition is encouraging greater transparency and stronger risk management. U.S. dollar liquidity, interest rates, and global payment demand remain the key underlying macro drivers.
The participation of crypto platforms such as Coinbase, OKX, Bybit, and Bitget Wallet demonstrates that institutions increasingly prefer multi-platform strategies. The broad involvement of traditional banks and technology companies further reinforces the trend of stablecoins evolving into mainstream payment infrastructure. Rather than simply redistributing existing market share, the industry's overall market size is likely to expand as more high-quality participants enter the ecosystem.
Final Thoughts
The launch of OUSD, combined with the latest index rebalancing, highlights the beginning of a new phase in stablecoin competition. While short-term volatility has created adjustment pressure, it also reflects the normal evolution of a maturing industry.
Circle and every other participant will ultimately need to respond through execution rather than narratives.
Over the coming months, the progress of OUSD's launch, Circle's financial results, and changes in stablecoin circulation and adoption will serve as important indicators for evaluating the competitive strengths of each player. The stablecoin story is far from over. It is evolving from a crypto-native utility into a critical component of global payments and financial infrastructure. Beyond price movements, investors should pay closer attention to real business adoption and ecosystem partnerships.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. The cryptocurrency market is highly volatile and involves substantial risk. Please conduct your own research and make independent investment decisions.

I. Por que a Grayscale voltou sua atenção para a Solana?
Nos últimos anos, duas palavras definiram quase sempre a Solana: desempenho e memecoins.
Como uma das principais blockchains Layer 1 do ciclo de mercado anterior, a Solana ganhou destaque graças ao seu alto rendimento, baixos custos de transação e finalização quase instantânea. Ao mesmo tempo, projetos do ecossistema como BONK, dogwifhat (WIF) e Pump.fun transformaram a Solana no epicentro da febre das memecoins. Ainda assim, essa percepção também ofuscou uma transformação mais profunda que vem acontecendo em toda a rede.

No fim de junho de 2026, a Chainalysis apresentou um novo framework chamado Blockchain Tracing Ontology, com o objetivo de estabelecer uma forma mais padronizada e transparente de descrever a inteligência de blockchain. Em vez de lançar mais um produto de analytics ou uma ferramenta de investigação, a empresa está tentando algo muito mais fundamental: redefinir como os dados de rastreamento de blockchain são estruturados, interpretados e comunicados.
Embora a estrutura ainda esteja na fase de proposta, ela já provocou uma importante discussão em toda a indústria de ativos digitais. No seu cerne, há uma pergunta simples, porém de grande alcance: A inteligência de blockchain precisa de uma linguagem comum?

Em 26 de junho de 2026, a OpenAI apresentou oficialmente a família GPT-5.6, revelando três modelos distintos: Sol, Terra e Luna. Diferentemente de lançamentos anteriores centrados em um único modelo carro-chefe, o GPT-5.6 representa uma mudança significativa na estratégia de produtos da OpenAI. Em vez de entregar um único modelo “melhor”, a empresa agora oferece um portfólio completo de modelos, projetado para atender a três prioridades diferentes: inteligência máxima, desempenho equilibrado e eficiência de custo em alta taxa de transferência.
De acordo com a OpenAI, a série GPT-5.6 melhora significativamente as capacidades em engenharia de software, operação de computadores, trabalho profissional de conhecimento, pesquisa científica e cibersegurança. No lançamento, os modelos ficam disponíveis apenas por meio de uma prévia limitada via API e Codex para um pequeno grupo de parceiros confiáveis, com uma disponibilidade mais ampla no ChatGPT esperada para uma etapa posterior.

Em 22 de junho de 2026, a trade.xyz lançou oficialmente o contrato perpétuo ZHIPU-USDC no mercado Hyperliquid HIP-3. O contrato oferece até 10x de alavancagem e permite negociações 24/7.
Este é o segundo ativo listado em Hong Kong pela trade.xyz. O primeiro foi o MINIMAX (Xi Yu Technology / MiniMax Group, HK:0100), que foi lançado em 18 de junho de 2026.
Visão Geral Rápida
Este artigo começa com o histórico da empresa ZHIPU e a inovação tecnológica GLM-5.2, depois detalha a mecânica do contrato trade.xyz e seu desempenho inicial, comparando-o com o MINIMAX, analisando múltiplos fatores que impulsionam o mercado, explorando o ecossistema Hyperliquid HIP-3 e, finalmente, olhando para o potencial a longo prazo da precificação on-chain para diversos ativos.

No dia 22 de junho de 2026, a Marvell Technology oficialmente se tornou um componente do Índice S&P 500. À primeira vista, isso pode parecer um evento rotineiro de rebalanceamento de índice. No entanto, quando olhado através da lente mais ampla do ciclo de investimento em infraestrutura de IA, a reavaliação da indústria de semicondutores dos EUA e a crescente influência dos fluxos de investimento passivos, a inclusão da Marvell representa algo muito mais significativo. Serve como um reconhecimento formal da transformação bem-sucedida da empresa de fornecedora tradicional de semicondutores de comunicação para um jogador crítico no ecossistema de infraestrutura de IA. Para a Marvell, entrar no S&P 500 não só lhe confere status de blue-chip aos olhos dos investidores globais, mas também eleva as expectativas em relação ao crescimento, rentabilidade e execução a longo prazo. Assim, este marco é tanto uma conquista quanto o início de uma fase mais exigente de sua jornada corporativa.

À primeira vista, a reunião de política do Federal Reserve em junho de 2026 parecia tranquila. O Comitê Federal de Mercado Aberto (FOMC) decidiu manter a taxa de fundos federais inalterada em 3.50%–3.75%, uma jogada que foi amplamente antecipada pelos mercados e em grande parte precificada antes do anúncio.
No entanto, focar apenas na decisão da taxa de juros pode fazer você perder a mensagem mais importante da reunião. Embora o Fed tenha optado por não aumentar as taxas desta vez, suas projeções econômicas atualizadas, mudanças no gráfico de pontos, linguagem de política revisada e a estreia do novo presidente do Fed, Kevin Warsh, enviaram um sinal muito mais significativo: a conversa na política monetária mudou de "quando os cortes de taxa começarão?" para "ainda seriam necessários mais aumentos de taxa para conter a inflação?"

Introdução
Ao longo da última década, a Notion se tornou uma das empresas mais interessantes para se estudar no cenário global de SaaS. Não foi construída através de uma única funcionalidade inovadora, um hack de crescimento passageiro, ou uma máquina de vendas empresariais agressiva. Em vez disso, a Notion cresceu através de um sistema de crescimento complexo, mas altamente orgânico, evoluindo de uma ferramenta de produtividade de nicho para uma plataforma global de gestão do conhecimento, colaboração em equipe e design de fluxo de trabalho. Muitos produtos conseguem usuários iniciais através da novidade, mas à medida que o interesse dos usuários diminui, as alternativas se multiplicam e os custos de aquisição sobem, eles rapidamente atingem um teto de crescimento. O que torna a Notion diferente é que seu crescimento nunca foi baseado em um único canal. Ela conectou a experiência do produto, ecossistemas de templates, comunidades de usuários, distribuição de conteúdo e necessidades de colaboração em equipe em uma rede de reforço.

Introdução
Em junho de 2026, o Banco do Japão aumentou sua taxa de política para 1%, marcando a primeira vez desde 1995 que a taxa de referência do Japão alcançou esse nível. Em termos absolutos, uma taxa de 1% não é nada impressionante entre as principais economias. A taxa de fundos federais dos EUA permanece acima de 4%, e as taxas de política em grande parte da Europa ainda estão significativamente mais altas do que a do Japão. Visto puramente como um número, o aumento da taxa do Japão não parece significativo o suficiente para atrair tanta atenção global. No entanto, os mercados financeiros raramente se concentram apenas no nível das taxas de juros; eles se concentram no que essas taxas sinalizam sobre a direção da política e o ciclo econômico mais amplo. Para uma economia que passou décadas em um ambiente de taxas zero e até taxas negativas, a mudança de taxas negativas para 1% representa uma mudança profunda na estrutura monetária que sustentou a economia do Japão por quase trinta anos.

No último ano, a IA incorporada emergiu como um dos setores mais observados na tecnologia global. Desde a Figure AI e a Physical Intelligence nos Estados Unidos até a AgiBot e a Galbot na China, investidores, pesquisadores e líderes da indústria têm buscado a mesma pergunta: Quem vai construir a camada de inteligência que alimenta a próxima geração de robôs?
Por décadas, os robôs operaram principalmente através de regras pré-definidas, fluxos de trabalho cuidadosamente projetados e ambientes altamente estruturados. A visão de máquinas verdadeiramente inteligentes—robôs capazes de entender seu entorno, se adaptar a situações desconhecidas, prever resultados e tomar decisões de forma autônoma—continuou sendo elusiva. No entanto, hoje, os avanços em modelos de base e inteligência incorporada estão aproximando essa visão da realidade.

De Ranking da Indústria a um Mapa do Poder Financeiro
Em junho de 2026, a Fortune revelou seu inaugural Crypto 100, um ranking abrangente projetado para identificar as empresas, protocolos e instituições mais influentes no ecossistema de ativos digitais. Diferente dos rankings tradicionais baseados apenas em receita, capitalização de mercado ou volume de negociação, o Crypto 100 tenta algo muito mais ambicioso: ele busca mapear as organizações que estão construindo a infraestrutura da próxima era financeira.
O ranking divide a indústria em dez categorias—Finanças Centralizadas (CeFi), Finanças Tradicionais (TradFi), Fintech, Finanças Descentralizadas (DeFi), Capital de Risco, Stablecoins, Serviços de Cripto, Ativos Digitais & ETFs, Mineração, e Protocolos de Blockchain. Ao fazer isso, ele fornece uma das imagens mais claras de como o cenário dos ativos digitais está evoluindo.

Em junho de 2026, a Oracle entregou o que pode ser o relatório de ganhos mais impactante de sua história. A receita trimestral alcançou US$ 19,2 bilhões, um aumento de 21% ano a ano, enquanto a receita total do ano subiu para um recorde de US$ 67,4 bilhões. No entanto, o que realmente impressionou foi a obrigação de desempenho restante da empresa (RPO), que disparou para impressionantes US$ 638 bilhões, representando um aumento de 363% em relação ao ano anterior. Esse número efetivamente significa que a Oracle acumulou uma backlog de receita futura equivalente a quase dez anos de sua receita anual atual.

Fechamento Parcial do Governo Termina, mas uma Luta Política Maior Está por Vir
Em 3 de fevereiro de 2026, a Câmara dos Representantes dos EUA aprovou por pouco um amplo pacote de financiamento do governo por uma votação de 217–214, pondo fim a um breve fechamento parcial do governo federal. O projeto, totalizando cerca de $1,2 trilhão, foi rapidamente sancionado pelo presidente Donald Trump, permitindo que a maioria das agências federais retomasse operações normais.
No entanto, o acordo ficou muito aquém de uma resolução completa. Embora a legislação financie a maioria dos departamentos do governo até o final do ano fiscal em 30 de setembro, ela fornece apenas uma extensão temporária de duas semanas para o Departamento de Segurança Interna (DHS). Essa decisão efetivamente adiou—em vez de resolver—o conflito mais contencioso no cerne do fechamento: até onde o Congresso deve ir ao impor limites à aplicação da imigração federal.

Se você aconteceu de andar por Nova York recentemente e notou uma loja de supermercado pop-up dando comida de graça, há uma boa chance de que você já estivesse dentro da narrativa dos mercados de previsão—sem perceber.
No início de 2026, a Polymarket e seu principal concorrente Kalshi lançaram ativações de 'supermercado gratuito' quase simultâneas em Nova York. Sem caixas de doação, sem carteiras de cripto, sem tutoriais de integração. Apenas uma fila, uma sacola de compras e uma presença de marca discreta.
Isso não era caridade. E não era uma armadilha.

A Fundação Ethereum diz que o ERC-8004 está indo para a mainnet em breve. Para muitas pessoas, a primeira reação é familiar: mais um novo padrão—isso realmente importa?
Desta vez, pode ser.
O ERC-8004 não se trata de blocos mais rápidos ou aplicativos mais chamativos. Ele se destina a um problema mais desconfortável—um que se torna inevitável uma vez que agentes de IA comecem a agir em nosso nome e gastar dinheiro real:
Como você sabe se o agente do outro lado é legítimo—e digno de confiança?
1. Quando os Agentes Escalam, a Confiança Quebra Primeiro

Moltbook: Os Humanos Ainda Estão no Sistema? Nas redes sociais, uma das acusações mais comuns que as pessoas fazem umas às outras é simples: “Você é um bot?” Moltbook leva essa ideia ao seu extremo lógico. Não pergunta se você é humano ou não — assume que você não deveria estar lá em primeiro lugar. Moltbook parece familiar à primeira vista. Lembra o Reddit: fóruns baseados em tópicos, publicações, comentários, votos positivos. Mas há uma diferença fundamental. Quase todos que postam e interagem na plataforma são agentes de IA. Humanos são permitidos como espectadores, mas não para participar. Isso não é “IA te ajudando a escrever uma postagem.” Não é “humanos conversando com IA.” É IA conversando com IA em um espaço público compartilhado — discutindo, formando alianças, discordando, exibindo-se e ocasionalmente se despedaçando. Humanos são explicitamente empurrados para os bastidores. Somos observadores, não participantes.