7*24

Hot topic

Market

Activities

On June 1, NVIDIA announced that Anthropic, OpenAI, and SpaceX are among the first major users of its upcoming microprocessor, marking a significant step in its expansion within the AI data center sector by securing key customers. Jensen Huang stated that they will be among the first to utilize NVIDIA's Vera central processing unit in their data centers. This new product is set to go into full production in the third quarter of this year. Vera is NVIDIA's first standalone data center microprocessor and will compete directly with Intel's Xeon product line, AMD's Epyc chips, and Amazon's Graviton among other major operators' internal projects. Huang noted that in terms of AI-related workloads, Vera's speed is 1.8 times that of chips based on Intel's technology (x86). This marks NVIDIA's first challenge to products that have long been the industry standard in performance comparisons.

[NVIDIA Secures First Users for Anthropic, Expands AI Data Center Strategy]On June 1, NVIDIA announced that Anthropic, OpenAI, and SpaceX are among the first major users of its upcoming microprocessor, marking a significant step in its expansion within the AI data center sector by securing key customers. Jensen Huang stated that they will be among the first to utilize NVIDIA's Vera central processing unit in their data centers. This new product is set to go into full production in the third quarter of this year. Vera is NVIDIA's first standalone data center microprocessor and will compete directly with Intel's Xeon product line, AMD's Epyc chips, and Amazon's Graviton among other major operators' internal projects. Huang noted that in terms of AI-related workloads, Vera's speed is 1.8 times that of chips based on Intel's technology (x86). This marks NVIDIA's first challenge to products that have long been the industry standard in performance comparisons.

NVIDIA Secures First Users for Anthropic, Expands AI Data Center Strategy

According to Beating monitoring, Jensen Huang extensively introduced the design philosophy of the Vera CPU during his speech at GTC Taipei 2026. He pointed out that all previous CPUs were designed for humans, where human interactions occur in seconds, and cloud CPUs are rented by core count. However, the world of Agents operates in nanoseconds: each time a tool is called or a database is accessed, Agents require the fastest response, as any delay would block the next step of reasoning. "In the past, we built CPUs for a billion humans; in the future, we will build CPUs for billions of Agents." He summarized the four main design pillars of the Vera CPU. First, it has the highest single-thread performance (IPC) globally: it can complete the fetching, decoding, and execution of 10 instructions per clock cycle because Agents require extremely low latency rather than traditional throughput. Second, it boasts world-class bandwidth per core. Third, the total on-chip bandwidth breaks limits: a new interconnect architecture connects all CPU cores at light speed, achieving a cross-sectional bandwidth of 3.6 TB/s, with no chiplet boundaries or cross-chip overhead, allowing all cores to work collaboratively instead of being rented by core. Fourth, it achieves extreme energy efficiency: deploying as many CPUs as possible without encroaching on GPU token generation for power. The Vera CPU also sets multiple industry firsts: it is the world's first CPU to support PCIe Gen 6, and the first server processor equipped with LPDDR5 memory achieving 1.2 TB/s bandwidth (2 to 3 times the performance of the current highest-performing x86 CPU). Huang stated that a 5% improvement in the CPU industry is already impressive, and a 10% increase is rare, but the performance gap of the Vera CPU compared to the strongest x86 is "of an unprecedented magnitude." He also revealed that NVIDIA has sold millions of Grace CPUs (Grace Blackwell series), making it one of the largest CPU manufacturers in the world. The Vera CPU plays three roles in the system: orchestrating GPUs and managing KV caches within the Vera Rubin NVL72 rack; running model orchestration, tool invocation, and database access as the harness layer for Agents; and driving the world's fastest AI storage servers in the Vera BlueField storage system.

[Jensen Huang Details Vera CPU: The World's First Processor Designed for Agents Rather Than Humans]According to Beating monitoring, Jensen Huang extensively introduced the design philosophy of the Vera CPU during his speech at GTC Taipei 2026. He pointed out that all previous CPUs were designed for humans, where human interactions occur in seconds, and cloud CPUs are rented by core count. However, the world of Agents operates in nanoseconds: each time a tool is called or a database is accessed, Agents require the fastest response, as any delay would block the next step of reasoning. "In the past, we built CPUs for a billion humans; in the future, we will build CPUs for billions of Agents." He summarized the four main design pillars of the Vera CPU. First, it has the highest single-thread performance (IPC) globally: it can complete the fetching, decoding, and execution of 10 instructions per clock cycle because Agents require extremely low latency rather than traditional throughput. Second, it boasts world-class bandwidth per core. Third, the total on-chip bandwidth breaks limits: a new interconnect architecture connects all CPU cores at light speed, achieving a cross-sectional bandwidth of 3.6 TB/s, with no chiplet boundaries or cross-chip overhead, allowing all cores to work collaboratively instead of being rented by core. Fourth, it achieves extreme energy efficiency: deploying as many CPUs as possible without encroaching on GPU token generation for power. The Vera CPU also sets multiple industry firsts: it is the world's first CPU to support PCIe Gen 6, and the first server processor equipped with LPDDR5 memory achieving 1.2 TB/s bandwidth (2 to 3 times the performance of the current highest-performing x86 CPU). Huang stated that a 5% improvement in the CPU industry is already impressive, and a 10% increase is rare, but the performance gap of the Vera CPU compared to the strongest x86 is "of an unprecedented magnitude." He also revealed that NVIDIA has sold millions of Grace CPUs (Grace Blackwell series), making it one of the largest CPU manufacturers in the world. The Vera CPU plays three roles in the system: orchestrating GPUs and managing KV caches within the Vera Rubin NVL72 rack; running model orchestration, tool invocation, and database access as the harness layer for Agents; and driving the world's fastest AI storage servers in the Vera BlueField storage system.

Jensen Huang Details Vera CPU: The World's First Processor Designed for Agents Rather Than Humans

Market data shows that ETH has fallen below $2000, currently priced at $1999.92, with a 24-hour decline of 1.57%. The market is experiencing significant volatility, so please ensure proper risk management.

[ETH Falls Below $2000]Market data shows that ETH has fallen below $2000, currently priced at $1999.92, with a 24-hour decline of 1.57%. The market is experiencing significant volatility, so please ensure proper risk management.

ETH Falls Below $2000

On June 1, NVIDIA announced that Anthropic, OpenAI, and SpaceX will be among the first major customers for its new processor. NVIDIA continues to expand its vast business landscape in the artificial intelligence data center sector, and this time it has also secured core users for the new product. Co-founder and CEO Jensen Huang specifically mentioned these AI companies during his speech, stating that they will be the first to deploy NVIDIA's Vera series central processors in their own data centers. The new product is set to go into full production in the third quarter of this year. (Sina Finance)

[NVIDIA: Anthropic, OpenAI, and Others to First Adopt New Vera Chip]On June 1, NVIDIA announced that Anthropic, OpenAI, and SpaceX will be among the first major customers for its new processor. NVIDIA continues to expand its vast business landscape in the artificial intelligence data center sector, and this time it has also secured core users for the new product. Co-founder and CEO Jensen Huang specifically mentioned these AI companies during his speech, stating that they will be the first to deploy NVIDIA's Vera series central processors in their own data centers. The new product is set to go into full production in the third quarter of this year. (Sina Finance)

NVIDIA: Anthropic, OpenAI, and Others to First Adopt New Vera Chip

According to Beating monitoring, Jensen Huang officially confirmed during his speech at GTC Taipei 2026 that "Vera Rubin is in full production." He expressed special thanks to the Taiwanese supply chain partners present, stating that the supply chain scale for Vera Rubin is twice that of the previous generation Grace Blackwell, and the assembly time for a single rack has been reduced from 2 hours during the Grace Blackwell era to just 5 minutes. Huang showcased a video of the entire manufacturing process for Vera Rubin. The system begins with TSMC's 3nm process and includes 7 new chips, with a system-level integration of over 60 trillion transistors and more than 18,000 components on a single board, comprising a total of 1.3 million parts (third-generation MGX rack design). HBM4 memory is sourced from Micron, SK Hynix, and Samsung. The system features a wireless PCB mid-board design, with all ConnectX-9 SuperNIC and BlueField-4 DPU integrated on board to ensure AI factory-level reliability. The liquid-cooled bus can handle over 5000 amps of current, equivalent to 20 electric vehicles accelerating at full throttle simultaneously. He also confirmed that Microsoft, Dell, and CoreWeave have completed the deployment and operation of the Vera Rubin NVL72 engineering machine. Millions of square feet of capacity are now online to support Grace Blackwell deliveries and are currently being prepared for the ramp-up of Vera Rubin, with large-scale shipments set to commence in the second half of 2026. Additionally, Huang showcased the Vera CPU rack (a single liquid-cooled rack equipped with 256 Vera CPUs, responsible for model orchestration and memory scheduling) and the new Groq 3 LPX low-latency inference rack (256 Groq 3 LPUs with 40 PB/s SRAM bandwidth). The NVL72 focuses on the highest throughput token generation, while the Groq 3 LPX specializes in the lowest latency token generation, with both systems complementing each other.

[Jensen Huang: Vera Rubin in Full Production, Microsoft, Dell, and CoreWeave First to Deploy]According to Beating monitoring, Jensen Huang officially confirmed during his speech at GTC Taipei 2026 that "Vera Rubin is in full production." He expressed special thanks to the Taiwanese supply chain partners present, stating that the supply chain scale for Vera Rubin is twice that of the previous generation Grace Blackwell, and the assembly time for a single rack has been reduced from 2 hours during the Grace Blackwell era to just 5 minutes. Huang showcased a video of the entire manufacturing process for Vera Rubin. The system begins with TSMC's 3nm process and includes 7 new chips, with a system-level integration of over 60 trillion transistors and more than 18,000 components on a single board, comprising a total of 1.3 million parts (third-generation MGX rack design). HBM4 memory is sourced from Micron, SK Hynix, and Samsung. The system features a wireless PCB mid-board design, with all ConnectX-9 SuperNIC and BlueField-4 DPU integrated on board to ensure AI factory-level reliability. The liquid-cooled bus can handle over 5000 amps of current, equivalent to 20 electric vehicles accelerating at full throttle simultaneously. He also confirmed that Microsoft, Dell, and CoreWeave have completed the deployment and operation of the Vera Rubin NVL72 engineering machine. Millions of square feet of capacity are now online to support Grace Blackwell deliveries and are currently being prepared for the ramp-up of Vera Rubin, with large-scale shipments set to commence in the second half of 2026. Additionally, Huang showcased the Vera CPU rack (a single liquid-cooled rack equipped with 256 Vera CPUs, responsible for model orchestration and memory scheduling) and the new Groq 3 LPX low-latency inference rack (256 Groq 3 LPUs with 40 PB/s SRAM bandwidth). The NVL72 focuses on the highest throughput token generation, while the Groq 3 LPX specializes in the lowest latency token generation, with both systems complementing each other.

Jensen Huang: Vera Rubin in Full Production, Microsoft, Dell, and CoreWeave First to Deploy

On June 1, HYPE briefly surpassed $74 this morning, setting a new price high. According to hl.eco data, the HYPE TWAP net buying volume has shown a significant recovery, with an expected net buying volume exceeding $14.8 million in the next 24 hours. Note from BlockBeats: The TWAP order model (Time-Weighted Average Price) divides a large order into many smaller ones evenly over time, executing them at fixed intervals within a predetermined time window. It is a common algorithmic trading strategy primarily used to reduce market price impact when executing large orders.

[HYPE TWAP Net Buying Volume Shows Significant Recovery, Expected to Exceed $14.8 Million in the Next 24 Hours]On June 1, HYPE briefly surpassed $74 this morning, setting a new price high. According to hl.eco data, the HYPE TWAP net buying volume has shown a significant recovery, with an expected net buying volume exceeding $14.8 million in the next 24 hours. Note from BlockBeats: The TWAP order model (Time-Weighted Average Price) divides a large order into many smaller ones evenly over time, executing them at fixed intervals within a predetermined time window. It is a common algorithmic trading strategy primarily used to reduce market price impact when executing large orders.

HYPE TWAP Net Buying Volume Shows Significant Recovery, Expected to Exceed $14.8 Million in the Next 24 Hours

According to Beating monitoring, Jensen Huang systematically introduced NVIDIA's DSX (the third major product line following RTX and DGX) during his speech at GTC Taipei 2026, positioning it as an end-to-end reference design and operational platform for AI factories. He stated that NVIDIA has evolved from a GPU company to a systems company and is now transforming again into an AI infrastructure company: 'Customers do not want to buy computers; they want to build AI factories.' DSX consists of four major modules. DSX SIM, based on Omniverse digital twins, completes the layout planning, power cooling simulation, and network validation of the entire factory before the first rack is deployed; DSX OS is responsible for the automatic configuration, operational monitoring, and fault recovery of the infrastructure, transforming the system into a multi-tenant, highly available AI-ready compute; DSX MAX LPS addresses the common issue of 40% power over-provisioning in current AI factories by dynamically reallocating power between racks and smoothing peak currents, allowing for the deployment of more GPUs within the same power budget; DSX Flex reads grid signals in real-time and dynamically adjusts factory power consumption when the grid needs relief, making AI factories 'flexible energy assets' for the grid. Additionally, NVIDIA's ready 45°C high-temperature liquid cooling technology eliminates the need for traditional chillers, significantly reducing water and energy consumption. Huang provided a clear economic account: 'Every token can be profitable, compute is revenue, and every watt of performance is your income.' The cost of a single 1GW AI factory has risen from $20 to $30 billion to $50 to $60 billion, and is expected to reach $80 to $100 billion soon. By the end of this decade, there will be 100GW of AI factories launched globally, which he called 'the largest infrastructure build in human history.' He also showcased several emerging AI cloud companies built on NVIDIA's full stack: CoreWeave (valued at $50 to $70 billion and rapidly growing), Nebius, N Scale (clients include BT and Google), India's Yoda, Singapore's AI Singapore, Indonesia's IndoSat, and Taiwan's GMI, emphasizing that 'AI will land in every region, and every company will be AI-driven.'

[Jensen Huang Explains DSX AI Factory Economics: 'Compute Equals Revenue', 100GW AI Compute to Launch Globally by End of This Decade]According to Beating monitoring, Jensen Huang systematically introduced NVIDIA's DSX (the third major product line following RTX and DGX) during his speech at GTC Taipei 2026, positioning it as an end-to-end reference design and operational platform for AI factories. He stated that NVIDIA has evolved from a GPU company to a systems company and is now transforming again into an AI infrastructure company: 'Customers do not want to buy computers; they want to build AI factories.' DSX consists of four major modules. DSX SIM, based on Omniverse digital twins, completes the layout planning, power cooling simulation, and network validation of the entire factory before the first rack is deployed; DSX OS is responsible for the automatic configuration, operational monitoring, and fault recovery of the infrastructure, transforming the system into a multi-tenant, highly available AI-ready compute; DSX MAX LPS addresses the common issue of 40% power over-provisioning in current AI factories by dynamically reallocating power between racks and smoothing peak currents, allowing for the deployment of more GPUs within the same power budget; DSX Flex reads grid signals in real-time and dynamically adjusts factory power consumption when the grid needs relief, making AI factories 'flexible energy assets' for the grid. Additionally, NVIDIA's ready 45°C high-temperature liquid cooling technology eliminates the need for traditional chillers, significantly reducing water and energy consumption. Huang provided a clear economic account: 'Every token can be profitable, compute is revenue, and every watt of performance is your income.' The cost of a single 1GW AI factory has risen from $20 to $30 billion to $50 to $60 billion, and is expected to reach $80 to $100 billion soon. By the end of this decade, there will be 100GW of AI factories launched globally, which he called 'the largest infrastructure build in human history.' He also showcased several emerging AI cloud companies built on NVIDIA's full stack: CoreWeave (valued at $50 to $70 billion and rapidly growing), Nebius, N Scale (clients include BT and Google), India's Yoda, Singapore's AI Singapore, Indonesia's IndoSat, and Taiwan's GMI, emphasizing that 'AI will land in every region, and every company will be AI-driven.'

Jensen Huang Explains DSX AI Factory Economics: 'Compute Equals Revenue', 100GW AI Compute to Launch Globally by End of This Decade

On June 1, according to official news, Huobi HTX has launched a TradFi trading competition for contract trading in U.S. stocks. From now until June 11 at 20:00, new and existing users can participate by completing designated trading pairs to share a prize pool of $50,000 based on their trading volume. For more details, please refer to the official website.

[Huobi HTX Launches TradFi Trading Competition with $50,000 Prize Pool]On June 1, according to official news, Huobi HTX has launched a TradFi trading competition for contract trading in U.S. stocks. From now until June 11 at 20:00, new and existing users can participate by completing designated trading pairs to share a prize pool of $50,000 based on their trading volume. For more details, please refer to the official website.

Huobi HTX Launches TradFi Trading Competition with $50,000 Prize Pool

On June 1, NVIDIA (NVDA.O) CEO Jensen Huang stated at the GTC Taipei 2026 conference that from an industry perspective, tokens are assets and have become revenue units that generate profit. This is because they can create profit. AI companies will want to build more tokens, generate more tokens, and produce more AI factories, which is also why Taiwan's computing demand has skyrocketed.

[NVIDIA CEO Jensen Huang: From an Industry Perspective, Tokens are Assets and Revenue Units]On June 1, NVIDIA (NVDA.O) CEO Jensen Huang stated at the GTC Taipei 2026 conference that from an industry perspective, tokens are assets and have become revenue units that generate profit. This is because they can create profit. AI companies will want to build more tokens, generate more tokens, and produce more AI factories, which is also why Taiwan's computing demand has skyrocketed.

NVIDIA CEO Jensen Huang: From an Industry Perspective, Tokens are Assets and Revenue Units

According to The Block, Cosmos-based decentralized exchange Osmosis has launched a stableswap — a protocol for trading stablecoins. The solution aims to let traders swap large amounts of stablecoins with minimum price impact. With this launch, Osmosis developers hope to become the main venue where Cosmos users can access and trade a wider range of stablecoins, both centralized and ones backed with crypto tokens. Osmosis want to support multiple stablecoins in the Cosmos ecosystem like USDC, USDT, BUSD.

Detail

【Cosmos-based DEX Osmosis launches a stableswap】According to The Block, Cosmos-based decentralized exchange Osmosis has launched a stableswap — a protocol for trading stablecoins. The solution aims to let traders swap large amounts of stablecoins with minimum price impact. With this launch, Osmosis developers hope to become the main venue where Cosmos users can access and trade a wider range of stablecoins, both centralized and ones backed with crypto tokens. Osmosis want to support multiple stablecoins in the Cosmos ecosystem like USDC, USDT, BUSD.

Cosmos-based DEX Osmosis launches a stableswap

Microsoft to Discontinue Using Claude: Too Expensive, or a Calculated Move?

NeoCloud Three Giants: NBIS, IREN, CRWV – Which One Has More Investment Value?

US Spot Bitcoin ETFs Near Yearly Outflow Territory

New York Lawsuit Takes Aim At 3.79 Million Dormant Bitcoin

Galxe: How a Quest Platform Evolved into Web3's Growth Infrastructure

DeFi Has Reached Its Most Dangerous Moment: The Real Vulnerabilities Are Not in the Code

Eric Trump Sets A “Beyond Catastrophic” Bar To Sell Bitcoin — How Far Are We From That?

Investors Frantically Snap Up AI Firms with 'No Profits': A High-Stakes Gamble on 'the Right to Define the Future'

The 'Stock Call King' Serenity: 3840% Annual Return, Beating Institutions to the Punch

Retail Investors' 'Lead Brother' Serenity vs. Newly Minted Stock God Leopold: How Are the Two Top Hunters Mining AI's 'Physical Limits'?

Article

Microsoft to Launch Self-Developed Programming Models Next Week to Counter Cursor and Claude's Encroachment on Copilot

Cosmos-based DEX Osmosis launches a stableswap

Penafian：Konten diatas tidak merepresentasikan posisi HTX.，HTX tidak memberikan rekomendasi perdagangan apa pun.。

Artikel Terkait

Semua Komentar0TerkiniHangat

Topik Trending

Artikel Terkait