CNBeta — 2026-05-21#

Top Story#

Nvidia concedes the Chinese high-end AI chip market to Huawei as U.S. export controls tighten, while simultaneously facing a new ban on its RTX 5090D v2 consumer GPUs entering China. This marks a critical inflection point in the global semiconductor war, as Nvidia’s absence accelerates China’s push for a self-sufficient, localized AI ecosystem and independent computing infrastructure.

Tech & AI#

Tencent’s new OS-level AI assistant “Marvis” has officially launched, capable of deep system control and cross-device coordination across Windows, Mac, and Android. In the hardware race, Nvidia’s CEO is betting heavily on “Agentic CPUs” with the new Vera chip, projecting $20 billion in CPU sales this year and opening a potential $200 billion market for AI agent execution. Similarly, Intel is strictly enforcing an “A0 to production” mandate to fix historic yield issues ahead of its 14A node launch in 2028. Meanwhile, Meta is laying off roughly 8,000 employees to funnel resources into its AI ambitions. In the startup space, Anthropic expects its first profitable quarter as it negotiates to adopt Microsoft’s custom Maia chips to reduce reliance on Nvidia. On the regulatory front, President Trump has delayed signing an AI executive order over fears it could slow U.S. AI development and give an edge to China.

CNBeta — 2026-04-05#

Top Story#

According to a cnbeta report, Anthropic has officially blocked third-party access for “OpenClaw,” a highly popular AI wrapper that allowed users to tap into Claude’s capabilities. The crackdown has sparked backlash in the AI developer community, but it has simultaneously created a massive opportunity for domestic Chinese AI models. Startups like MiniMax are seizing the moment, publicly criticizing Anthropic’s walled-garden approach while touting their own cross-platform Token plans, pushing MiniMax back into the top 5 of OpenRouter’s global usage rankings.

CNBeta — 2026-04-06#

Top Story#

Iran’s Islamic Revolutionary Guard Corps (IRGC) has released an official video explicitly threatening to completely destroy OpenAI’s planned $30 billion “Stargate” AI data center in Abu Dhabi, according to a cnbeta report. This threat comes as direct retaliation against US warnings to strike Iranian energy infrastructure, highlighting how massive, billion-dollar AI computing facilities are increasingly viewed as strategic geopolitical targets.

Tech & AI#

Meta is seeking a $3 billion loan to construct a 1GW AI data center dubbed “Prometheus” in Ohio, which will operate on an independent microgrid primarily powered by natural gas, according to a report on Meta’s data center. On the supply chain side, an analysis warns that the HBM memory shortage will persist for five years, stretching until 2030, driven by the insatiable demands of advanced AI GPUs that consume three to four times the capacity of traditional memory production. Concurrently, Intel is aggressively expanding its foundry services by seeking advanced packaging deals with Amazon and Google, hoping to capitalize on its EMIB technology for complex AI chip architectures.

CNBeta — 2026-04-09#

Top Story#

Intel has officially joined Elon Musk’s ambitious TERAFAB project, a 2nm semiconductor fabrication initiative aiming to produce 100 to 200 billion chips annually, with 80% of its massive 1-terawatt AI compute capacity destined for space orbit. According to a cnbeta report on the partnership, Intel will provide its advanced design, manufacturing, and packaging expertise to support SpaceX, Tesla, and xAI, marking a paradigm shift in global silicon manufacturing to bypass terrestrial power grid limitations.

CNBeta — 2026-04-12#

Top Story#

According to a recent CNBeta report, Samsung is undertaking a massive restructuring of its operations in China, phasing out its home appliance, TV, and monitor businesses. The consumer electronics giant will now focus almost entirely on smartphones and storage solutions. This retreat from the highly competitive Chinese home appliance market underscores the fierce price wars and the rapid ascent of domestic Chinese brands, pushing Samsung to concentrate on high-margin semiconductor and mobile sectors.

CNBeta — 2026-04-13#

Top Story#

According to a report on enterprise AI, Anthropic’s annualized recurring revenue (ARR) has crossed the $30 billion mark, surpassing OpenAI’s $25 billion for the first time since the ChatGPT boom. This shift highlights a massive surge in enterprise AI adoption, specifically driven by venture-backed startups favoring Claude’s capabilities. Consequently, OpenAI is distancing itself from Microsoft’s control and deepening its ties with Amazon Web Services to secure more computing power and regain enterprise market dominance OpenAI Distances from Microsoft.

CNBeta — 2026-04-15#

Top Story#

The United States’ lead over China in artificial intelligence has narrowed to a mere three to six months, according to a recent cnbeta report on remarks by US Treasury Secretary Scott Bessent. A recent Stanford AI Index report backs this up, showing the performance gap between top US models like Claude Opus 4.6 and Chinese models like dola-seed-2.0-preview has essentially vanished to a 2.7% margin. This highlights the intense pressure on US tech giants and the rapid maturation of China’s domestic AI ecosystem as global capabilities converge.

CNBeta — 2026-04-16#

Top Story#

Nvidia CEO Jensen Huang’s recent remarks on China’s AI capabilities and US export controls take center stage. According to a cnbeta report, Huang urged the US to strengthen AI cooperation with China, warning that current trade tensions are hindering crucial research dialogues. He further stated in another report that the notion of China lacking AI chips is “nonsense”. Despite export bans on advanced EUV lithography, China’s vast energy resources and massive data centers allow them to compensate by stacking mainstream chips together to achieve the necessary computing power.

Chinese Tech Daily — 2026-04-17#

Top Story#

OpenAI’s Frontier team, led by Ryan Lopopolo, has unveiled a radical “Harness Engineering” approach, successfully maintaining a 1-million-line codebase with absolutely zero human-written code. By relying entirely on coding agents like Codex, they argue that human attention—not code generation—is the new bottleneck in software development. This marks a massive paradigm shift in how AI-native software engineering teams might operate, treating code as entirely disposable and focusing solely on system constraints, tooling loops, and agent prompts.

CNBeta — 2026-04-19#

Top Story#

According to a cnbeta report, NVIDIA CEO Jensen Huang strongly criticized US AI chip export bans on China, calling the assumption that American companies will inevitably lose market share to local Chinese competitors a “loser mentality”. Huang argued that isolating China will force it to develop an entirely separate technology stack, fundamentally harming the global reach of the US tech ecosystem and warning that AI innovation should not be artificially fractured.