24H Flash Updates

According to reports, Dell Technologies' stock has surged nearly 100% over the past 30 days, and despite a pullback of approximately 10% over the last five trading days, the year-to-date gain remains at 230%. Given the extreme volatility and significant appreciation, Seeking Alpha consulted with professionals such as Deep Value Investing to assess whether Dell’s current stock valuation may be overheated. This market performance reflects investors’ strong expectations regarding Dell’s prospects in AI servers and infrastructure, while also raising concerns about the alignment between its short-term gains and underlying fundamentals.

According to market reports, the latest earnings guidance released by Broadcom and Ciena failed to meet investor expectations, triggering a broad sell-off in technology stocks. Broadcom's stock plunged 13% during midday trading on Thursday, dragging down the entire semiconductor sector. Meanwhile, Micron Technology's shares declined by approximately 7%. This significant price volatility directly reflects growing market concerns over the semiconductor industry’s future growth momentum. As a dominant player in the sector, Broadcom’s weak guidance is seen as a signal of slowing demand, prompting investors to reassess the profitability outlook and supply chain risks across related enterprises.

According to the official announcement, Google Labs has launched an experimental application called Dreambeans, leveraging artificial intelligence to integrate user data across multiple Google products such as Gmail, Calendar, and Photos, generating personalized narrative content on a daily basis. The app is currently available to Google AI Ultra subscribers aged 18 and above in the United States, supporting both Android and iOS platforms, while other users can join the waitlist via the official website. Google emphasizes that privacy settings within Dreambeans are independent of Personal Intelligence configurations in other products like Gemini, and do not influence one another.

It is reported that the upcoming version of Siri, set to launch as early as next week, will have certain functionalities running on Nvidia Blackwell B200 GPUs provided by Google. According to reports, Apple is accessing Nvidia’s latest Blackwell B200 chip via Google Cloud services to enhance the processing capabilities of the new Siri.

This collaboration indicates that beyond its in-house silicon, Apple is actively seeking external high-performance compute resources to meet the demanding workloads associated with AI model inference. For developers, this means AI services within the Apple ecosystem will gain stronger backend computational potential, potentially accelerating the deployment of more complex AI features on devices and improving response speed.

Reports indicate that the cybersecurity community has once again witnessed the challenge posed by cutting-edge AI to established security assumptions. When Anthropic released the Claude Mythos model and provided a technical preview to select organizations, unauthorized entities claimed to have gained access within hours. If verified, this incident not only suggests potential data breaches but also exposes the vulnerability of current IT infrastructure when confronted with high-intelligence AI agents.

As Agentic AI becomes increasingly embedded in defensive operations, maximizing its effectiveness while preventing adversarial exploitation hinges on building secure IT infrastructure. This trend is driving the industry to re-evaluate pre-deployment security assessment protocols, emphasizing that the introduction of autonomous planning-capable agents must be accompanied by parallel upgrades to underlying defense systems to mitigate systemic risks.

According to the official announcement, NVIDIA has officially open-sourced Nemotron 3 Ultra, its flagship large language model with 550 billion parameters and 55 billion activated parameters, optimized for long-haul agent tasks such as complex planning, reasoning, and tool calling. The model employs a Mamba-Transformer hybrid expert architecture (MoE), alternating between Mamba-2 state space model layers and Transformer self-attention layers, enabling a context window of up to 1 million tokens with minimal memory overhead.

Compared to dense models of comparable scale, this architecture achieves fivefold higher throughput and reduces inference costs by 30% in agent workloads. NVIDIA simultaneously released the Agent Toolkit, an agent development suite featuring NemoClaw orchestration blueprints and OpenShell runtime, with all open-source assets including model weights, datasets, and training recipes.

Security researchers have disclosed a vulnerability in Anthropic's Claude Code GitHub Action, allowing attackers to take over vulnerable public repositories running the action by simply opening a GitHub Issue.

Since Anthropic's own operational repository uses the same workflow, a successful exploit could push malicious code directly into the action itself, affecting all downstream projects that pull this repository. The vulnerability was discovered by GMO researcher RyotaK.

According to financial institution Cantor, driven by Gorilla Technology's "strong" Q1 results announced last week, Cantor has raised the stock target price from $31 to $40 and maintained its "overweight" rating. Gorilla Technology is a UK-based technology company primarily providing security, networking, and Internet of Things (IoT) solutions.

The revised target price reflects market recognition of the company's recent financial performance and growth potential. As a player in the IoT and edge AI space, the company delivers secure monitoring and data management services to enterprises by integrating network connectivity with intelligent analytics. Cantor’s positive rating upgrade underscores institutional investor confidence in the company’s ability to execute technological deployment and future revenue prospects, enhancing its visibility and liquidity in capital markets.

According to an official announcement, Google Labs has launched an experimental application called Dreambeans, leveraging artificial intelligence to integrate data from multiple Google products such as Gmail, Calendar, and Photos, generating personalized narrative content on a daily basis. The app is currently available only to Google AI Ultra subscribers aged 18 and above in the United States, supporting both Android and iOS platforms, with other users able to join a waitlist. Google emphasizes that privacy choices within Dreambeans are independent of Personal Intelligence settings in other products like Gemini.

Reports indicate that AI-centric applications on the Apple App Store demonstrated strong performance in 2025, with billing revenue from such apps increasing fourfold compared to previous years. Data shows that 40 of the top 100 apps on the App Store ranking in 2025 featured consumer-facing AI capabilities, underscoring AI functionality as a critical driver for achieving top-tier placement. This growth trend reflects a significant rise in user willingness to pay for AI-powered applications, prompting developers to accelerate the integration of AI features into consumer-grade products for commercial monetization. The App Store ecosystem is undergoing a substantive transformation driven by the widespread adoption of AI technology.

According to Liz Thomas, Chief Market Strategist at SoFi, the upcoming wave of IPOs led by SpaceX, Anthropic, and OpenAI may mark a historic inflection point for public markets.

These three companies are expected to go public at unprecedented valuation scales, with their potential multi-trillion-dollar market caps poised to reshape the current market landscape. This wave of trillion-dollar AI IPO activity is not only set to inject substantial liquidity into the market but could also redefine valuation benchmarks for tech stocks, drawing massive capital inflows toward AI and aerospace sectors, thereby exerting sector-level impacts on existing industry weight distributions.

According to reports, visual discovery platform Pinterest has entered a strategic partnership with Amazon Web Services (AWS), committing to invest $4 billion in AWS cloud services by 2031. This agreement represents Pinterest’s largest infrastructure commitment in its history, aimed at leveraging AWS’s computational power to enhance its search functionality and modernize its platform.

The collaboration will focus on deepening AI applications in areas such as computer vision and recommendation algorithms, helping the platform more effectively identify and govern AI-generated content, thereby improving overall content ecosystem quality. By migrating to AWS infrastructure, Pinterest expects to accelerate the development and iteration of new features, optimize personalized user experiences, and provide foundational compute capacity for future advertising innovation.

According to The Information, Nvidia has acquired Kumo AI, an artificial intelligence model manufacturer, in a deal exceeding $400 million. The transaction has been publicly confirmed by a Nvidia executive on LinkedIn. Founded by Vanja Josifovski, Jure Leskovec, and Hema Raghavan, Kumo AI specializes in enterprise data forecasting using graph neural networks and large language models. This acquisition will bolster Nvidia’s technological positioning in enterprise-grade AI forecasting and data analytics, integrating Kumo AI’s graph machine learning capabilities to enhance the performance of Nvidia’s AI software stack when handling complex relational data, further expanding its application scenarios in the enterprise services market.

According to The Hacker News, the ThreatsDay Security Weekly Report reveals multiple emerging threat trends in the current cybersecurity landscape. The report indicates that AI agents are increasingly being abused to compromise real-world systems, while hacker forums, after experiencing downtime, have resumed operations but with an even more deteriorated security environment. On the attack technique front, ClickFix exploitation techniques, JavaScript backdoors, and suspicious C2 (Command and Control) tools are being widely leveraged. Additionally, legacy vulnerabilities, malicious plugins, and fraud schemes impersonating legitimate applications remain rampant. This series of developments underscores that as the barrier to entry for cyberattacks lowers, low-cost hackers are gaining access to more powerful attack toolkits, shifting the risk to network infrastructure from traditional software flaws toward AI-driven automated attacks. Enterprises must therefore enhance monitoring and defense mechanisms against novel attack vectors and anomalous behaviors exhibited by AI agents.

According to the official announcement, Cloudflare has announced the acquisition of VoidZero, the team behind the Vite JavaScript build tool. This acquisition aims to strengthen Cloudflare's platform capabilities for AI-assisted application development. Upon completion of the transaction, core developer tools from VoidZero—including Vite, Vitest, Rolldown, and Oxc—will be formally integrated into the Cloudflare ecosystem.

This move will combine high-performance frontend build tools with Cloudflare's global edge network, delivering a more optimized AI-native development environment for developers. By integrating these tools, Cloudflare can significantly enhance the efficiency of building and deploying applications on its platform, further solidifying its competitive advantage in AI-driven web infrastructure.

According to U.S. Securities and Exchange Commission (SEC) filings, the video streaming platform Rumble has signed its largest customer commitment agreement in company history. Per the filing, a third-party cloud client has committed to deploying services on Rumble’s cloud infrastructure, with the contract described as the company’s largest customer commitment to date.

Following this announcement, Rumble’s stock surged significantly after Thursday’s market open, reaching its highest level in the past 11 months. The signing of this major contract marks substantial progress in Rumble’s market expansion within the cloud services sector, validating both the technical capability and reliability of its cloud infrastructure while also securing predictable long-term revenue growth—key factors that enhance its market share and influence in the fiercely competitive cloud services landscape.

It is reported that Snap, the parent company of Snapchat, has acquired augmented reality firm Illumix, aiming to further strengthen its AR business capabilities. The specific amount of the transaction was not disclosed. According to the acquisition agreement, Snap will formally adopt Illumix's relevant technologies. This acquisition helps Snap integrate more specialized technological resources in the AR domain, enhancing the technical depth and product performance of its platform in augmented reality experiences, providing foundational support for future AR feature development and ecosystem expansion.

According to an official announcement, Nasdaq-listed K Wave Media (KWM) has signed an agreement to formally rescind and terminate its acquisition transaction with Solaire Partners LLC, originally announced in 2023. Under the terms of the agreement, ownership of Solaire Partners will be returned to its former shareholders, while K Wave Media will reclaim approximately 9.8 million restricted common shares.

The termination of this transaction signifies K Wave Media's divestiture of previously acquired assets, resulting in a significant restructuring of its capital structure. By reclaiming shares, the company reduces its outstanding float. This adjustment is intended to rectify prior M&A decisions, refocus on core business strategy, and optimize the balance sheet to better align with future operational requirements.

According to the official announcement, Guangdong Zhidong Weilai, in collaboration with Professor Cai Panpan's team from Shanghai Chuangzhi Academy, has unveiled the MINT-4B multimodal VLA (Vision-Language-Action) large model. The model ranks among the top three in global benchmark evaluations of mainstream general-purpose robotics models, outperforming industry-standard models such as OpenVLA, GR00T, π, and UniVLA in technical metrics.

The core innovation of the MINT series lies in "replicating task intent rather than mechanically replicating trajectories." It pioneers the SDAT multiscale frequency-domain tokenization technique, enabling hierarchical decoding through cross-scale autoregressive inference. This technology is already integrated into the Xiao Zhi S2 humanoid robot, deployed across scenarios including scientific research, education, and commercial exhibitions, with the model having been commercially rolled out in multiple regions nationwide.

According to an announcement by Eightco Holdings, the company disclosed that its total asset value has reached approximately $437 million. The investment portfolio includes around $142 million in cash and stablecoins, along with crypto assets and AI-related equity valued at approximately $246 million. Specifically, Eightco holds indirect equity in OpenAI worth about $90 million, possesses 283.5 million Worldcoin (WLD) tokens valued at roughly $156 million, and holds 16,300 Ethereum (ETH) tokens.

This asset disclosure highlights Eightco’s deep strategic positioning in AI infrastructure and cryptocurrency assets. By holding equity in OpenAI and WLD tokens, the company has tightly aligned its valuation with the growth of the AI ecosystem and the expansion of the World ecosystem. This concentrated allocation to AI and crypto assets means its net asset value will be directly influenced by market dynamics and project developments within these sectors.