1. OpenAI completed US$110 billion in financing and was valued at US$730 billion
OpenAI officially completed the second largest venture capital round in history on February 27, with financing amounting to US$110 billion and the company’s valuation rising to US$730 billion. This scale has surpassed the sovereign wealth funds of most countries, indicating that the valuation logic of top AI laboratories is moving closer to "national economies", and also indicates that the AI infrastructure arms race will further heat up in the future.
2. Total AI financing exceeded US$195 billion in February, setting a record in a single month
According to statistics, total financing in the global AI field reached US$195 billion in February 2026, setting a single-month financing record. Among them, Anthropic completed US$30 billion in Series G financing with a valuation of US$380 billion; MatX received US$500 million to develop a training chip that claims to be 10 times more powerful than Nvidia GPU; Axelera AI raised US$250 million to focus on edge inference chips. Big Tech’s full-year AI capital spending is expected to total $650 billion.
3. Anthropic Claude 5 is expected to be released this month, with greatly upgraded reasoning and cross-system collaboration capabilities.
According to industry sources, Anthropic’s Claude 5 is expected to be officially unveiled between February and March 2026. The new version will focus on strengthening continuous reasoning capabilities and introducing cross-system coordination capabilities - allowing Claude to manage long-term tasks across multiple platforms and maintain context across sessions. If released as scheduled, it will be a key weight in the head-to-head confrontation between Anthropic and OpenAI in the first half of 2026.
4. Google Gemini 3.1 Pro scores 77.1% on ARC-AGI-2 benchmark
Gemini 3.1 Pro released by Google DeepMind achieved a score of 77.1% in the latest ARC-AGI-2 inference benchmark test, supports 1 million Token ultra-long contexts, and has full-modal reasoning capabilities for text, images, audio, video and code. ARC-AGI-2 is regarded as an important indicator to measure the model's ability to approach universal reasoning, and this result has attracted widespread attention in the industry.
5. India’s Sarvam AI releases 105B and 30B local large models
Indian AI company Sarvam released two large language models trained on local data in February: the 105 billion parameter version is oriented towards Agentic task completion, code generation and scientific reasoning; the 30 billion parameter version uses MoE architecture and focuses on real-time dialogue scenarios. This move is seen as an important step for emerging market countries in their "sovereign AI" strategy.
6. Apple is developing AI wearables, with Visual Intelligence becoming a core selling point
Apple CEO Tim Cook has made it clear that Visual Intelligence will become the defining feature of Apple's wearable device line this year, and the new low-end MacBook equipped with AI is expected to be released on March 4. Combined with the previously announced iOS 26.4 AI version of Siri, Apple is building a full-scenario AI experience closed loop from the wrist to the desktop.
7. New York’s RAISE Act came into effect on March 19, and Vietnam’s AI Law officially came into effect today.
New York Governor Hochul signed the Responsible Artificial Intelligence Safety and Education Act (RAISE Act) in December 2025, which will officially take effect on March 19 and impose transparency, compliance, security testing and reporting obligations on developers of "cutting-edge AI large models". At the same time, Vietnam’s national AI law was officially implemented today (March 1), and Asia’s AI regulatory landscape continues to expand.
8. Brain MRI AI reading system is launched, completing neurological diagnosis and triage within seconds
A research team at the University of Michigan has developed an AI system that can accurately interpret brain MRI scans in seconds, cover the identification of a wide range of neurological diseases, and determine which cases require urgent treatment. This system is expected to significantly alleviate the shortage of neuroradiologists and improve the accessibility of early diagnosis in low-resource areas.
// KEY TAKEAWAYS
The core theme of this news is the dual acceleration of "money and rules": on the financing side, the record of US$195 billion in February alone shows that capital's bets on AI have not ebbed due to "ROI anxiety", but are instead concentrated in lower-level directions such as computing chips and Agentic infrastructure; on the regulatory side, the New York RAISE Act and Vietnam's AI Law have been implemented one after another, and the U.S. federal government will announce a list of "burdened state laws" on March 11, marking the global AI Governance moves from a statement of principles to an enforcement phase. Sandwiched in the middle are major model manufacturers - Claude 5, Gemini 3.1, and the new version of Siri competing at the same time. The model battle in the first half of 2026 has already begun.