Nvidia B200 chips power ByteDance AI as OpenAI launches GPT-5

Key Takeaways

• ByteDance is acquiring 36,000 Nvidia B200 chips for over $2.5 billion through Aolani Cloud in Malaysia for global AI R&D.
• ByteDance has also secured access to Nvidia H100 AI chips to enhance its AI capabilities and compete with global tech leaders.
• OpenAI launched GPT-5, featuring improved reasoning, a larger context window, reduced hallucinations, and native multimodal understanding including video and real-time audio.
• MoonPay integrated Ledger hardware wallet signing for its AI crypto agents, ensuring secure transaction approval and private keys remain on the device.
• AI and LLMs are transforming cloud security by interpreting complex data patterns, detecting anomalies, and summarizing incidents.
• A Salesforce report indicates 91% of Indian sales professionals consider AI agents essential for business success, with 90% using AI for prospecting and email drafting.
• An AI agent from CodeWall reportedly breached McKinsey & Company's Lilli platform in two hours by exploiting an SQL injection flaw.
• IM Dominator's OpenClaw training series teaches marketers to build AI agents without coding, enabling autonomous skill creation from API documentation.
• Europe aims to lead in AI by focusing on industrial applications such as robotics, manufacturing, healthcare, and energy.
• Elon Musk's xAI has experienced co-founder departures, including Guodong Zhang and Zihang Dai, due to reported dissatisfaction with the coding division's performance.

ByteDance secures top Nvidia AI chips for global projects

TikTok's parent company, ByteDance, is reportedly acquiring around 36,000 Nvidia B200 chips through a Malaysian cloud firm called Aolani Cloud. This significant hardware acquisition, estimated to cost over $2.5 billion, will be used for AI research and development outside of China. ByteDance plans to meet the growing global demand for AI services from its customers. Aolani Cloud stated it follows all export regulations and serves clients worldwide. The deal aligns with U.S. export controls, as Malaysia is not a restricted country for such shipments.

ByteDance gains access to powerful Nvidia H100 AI chips

ByteDance, the company behind TikTok, has reportedly secured access to Nvidia's advanced H100 AI chips, despite U.S. export restrictions. These powerful processors are crucial for ByteDance's plans to develop and deploy new AI applications, aiming to compete with global tech leaders. The company seeks to enhance its AI capabilities for services like advanced search and content generation. This access allows ByteDance to accelerate research in large language models and generative AI, potentially launching new products that rival competitors. Nvidia has developed alternative chips for China, but this deal suggests ByteDance found a way to acquire the most advanced technology.

ByteDance to use 36,000 Nvidia Blackwell GPUs in Malaysia

ByteDance plans to use a cluster of 36,000 Nvidia B200 GPUs located in Malaysia for its AI development. The cluster, valued at $2.5 billion and comprising 500 NVL72 GB200 systems, will be managed by Aolani Cloud. This setup is in line with U.S. export controls, as Nvidia ensures compliance with all regulations. ByteDance has been leasing Nvidia H100 GPUs from Aolani since early 2025. While ByteDance is a Chinese company, it is not on restricted U.S. lists, making this arrangement permissible under current export rules.

MoonPay integrates Ledger for secure AI crypto transactions

MoonPay has enhanced its AI crypto agents by integrating Ledger hardware wallet signing. This new feature allows users to verify and approve every transaction initiated by an AI agent using their Ledger device. Crucially, private keys remain securely on the hardware device, never leaving it. This addresses major security concerns surrounding autonomous crypto trading, enabling AI agents to execute strategies across blockchains while requiring human approval. MoonPay aims to allow AI to scale in managing digital assets without compromising user control.

MoonPay's AI agents now use Ledger for secure crypto trading

MoonPay has launched Ledger hardware wallet support for its AI-driven trading agents, marking a significant security upgrade. This integration ensures that private keys stay on a user's Ledger device while AI agents manage trades and portfolio actions across multiple blockchains. Users can approve transactions via Ledger, preventing AI misuse of funds or key exposure. The system supports various Ledger devices and multiple blockchains, allowing for efficient trading with hardware-enforced security. This positions MoonPay as a provider of institutional-grade crypto infrastructure.

OpenAI unveils GPT-5 with advanced reasoning and multimodal understanding

OpenAI has officially launched GPT-5, a significant advancement beyond previous models like GPT-4o. This new AI model features drastically improved reasoning capabilities for complex problem-solving and a much larger context window for processing extensive information. GPT-5 also claims a significant reduction in 'hallucinations' or factual errors, making its responses more reliable. A major leap is its native understanding of multiple formats, including video and real-time audio, allowing it to synthesize information from diverse sources for comprehensive analysis. The API is also faster and more stable for developers.

AI and LLMs are transforming cloud security defenses

Artificial intelligence (AI) and Large Language Models (LLMs) are revolutionizing cloud security and cyber defense strategies. Traditional security systems struggle to keep up with the rapidly evolving threat landscape and the massive volume of data generated in cloud environments. LLMs can now interpret complex patterns across various data sources, detect anomalies based on meaning, and summarize incidents quickly. This shift from simply processing logs to truly understanding them is a critical evolution for modern cloud security, enabling more effective responses to cyber threats.

Elon Musk's xAI faces co-founder departures amid performance issues

Elon Musk has reportedly led to more job cuts at his AI firm xAI, including several co-founders, due to dissatisfaction with the coding division's performance. Co-founder Guodong Zhang and Zihang Dai are among those who have recently left the company. These departures leave only two of the original twelve co-founders remaining. Staff have expressed concerns that the ongoing changes are harming morale and hindering the company's progress. Musk acknowledged past rejections and plans to reach out to promising candidates.

New course teaches marketers to build AI agents without coding

IM Dominator's new OpenClaw training series teaches marketers how to build and extend AI automation capabilities without writing any code. The course covers how an OpenClaw agent can autonomously create new integration skills by using provided API documentation. Users can access a Skills Marketplace called ClawdHub for pre-built integrations. The advanced curriculum includes safe skill installation, security auditing, and a pipeline for testing and deploying new skills. A key feature allows agents to access API documentation, write integration code, and test it in a sandbox environment, all directed by the operator.

Europe's AI opportunity lies in industry, not just chatbots

Europe has a second chance to lead in AI by focusing on embedding intelligence into the real economy, rather than just chatbots. While past digital waves favored data-rich tech giants, the next AI innovation wave will focus on robotics, manufacturing, healthcare, and energy. Europe's strengths in scientific talent, industrial expertise, and cross-sector ecosystems position it well for this shift. Unlike previous eras, this AI wave will value industry-specific data and domain knowledge more than just computation. This focus on physical and scientific applications offers Europe a unique advantage.

Navigating AI-generated images in war coverage

AI-generated images are becoming widespread and realistic, making it difficult to distinguish between real and fake content, especially during wartime. Authentic images can be wrongly flagged as fake, while fabricated ones spread misleading information. Experts advise critical thinking and slowing down before sharing content online. It's important to trust verified experts and news organizations that rigorously check information. Even verified images may not tell the whole story, so seeking context and understanding the broader picture is crucial.

Indian sales teams find AI agents crucial for success

A Salesforce report reveals that 91% of Indian sales professionals consider AI agents essential for business success. The survey found that 90% of Indian sales organizations use AI for tasks like prospecting and email drafting. Sellers using AI report deeper customer understanding and less work-related stress. Top performers are significantly more likely to use AI prospecting agents than under-performers. Data quality is a key challenge, with many prioritizing data cleansing to improve AI effectiveness. Combining human judgment with AI is seen as the best strategy for future success.

AI agent hacked McKinsey's AI platform in just 2 hours

A cybersecurity startup called CodeWall claims its AI agent breached McKinsey & Company's generative AI platform, Lilli, in just two hours. The agent exploited a publicly exposed technical documentation flaw, accessing millions of staff messages and thousands of files without credentials or human intervention. The vulnerability was an SQL injection flaw in an open endpoint that did not properly validate user input. This allowed the agent to access sensitive data and even alter internal system prompts, highlighting a significant security gap in the platform.

Sources

• China's ByteDance gets access to top Nvidia AI chips, WSJ reports
• Exclusive | China’s ByteDance Gets Access to Top Nvidia AI Chips
• China's ByteDance to access 36,000 Blackwell GPU cluster through Malaysia cloud operator — Nvidia confirms no objections, deal is in line with US export controls
• MoonPay adds Ledger-secured AI crypto agents to deal with wallet key risks
• MoonPay Adds Ledger-Secured AI Agents to Expand Institutional-Grade Crypto Trading
• GPT-5: Alles, was wir über das nächste Modell von OpenAI wissen (2026)
• How AI And LLMs Are Redefining Cloud Security and Cyber Defense
• Musk ousts more xAI founders as AI coding effort falters, FT reports
• IM Dominator Course Series Covers Self-Building AI Automation, Enabling Marketers to Extend Agent Capabilities
• Europe’s second chance on AI: building an opportunity in factories, labs, and the real economy
• AI-generated Iran images are widespread. How do we know what to believe?
• Over 90% Indian sales professionals find AI agents critical to business success: Salesforce report
• Autonomous Agent Hacked McKinsey's AI in 2 Hours