Anthropic has completed its first strategic acquisition, spending $1 billion to acquire Bun, a high-performance JS/TS runtime and full-stack development tool. This move strengthens the underlying foundational capabilities of Claude Code for enterprise-grade AI code generation and agent applications. Bun has committed to remaining open-source, bolstering its ecosystem. Claude Code has reached $1 billion in annualized revenue within six months of launch, signaling explosive growth for AI code agents.

Harvey, an AI platform focused on the legal industry, has completed a $160 million Series F funding round led by prominent investors including a16z. Its valuation has rapidly increased to $8 billion, and it now serves over 700 clients worldwide (including half of the Am Law 100 firms). The platform emphasizes security, compliance, and a multi-model architecture, signifying that AI has evolved from an experimental tool to core infrastructure for the legal and professional services industries.

Databricks, a company spanning large models and data platforms, is in discussions for a new round of $5 billion in funding (partially from Insight Partners). Its valuation has reached $134 billion, with 2025 revenue projected at $4.1 billion. Revenue from its AI products already exceeds $1 billion annually, demonstrating the strong trend of AI data infrastructure becoming the new core competitiveness in enterprise software.

NVIDIA has invested $2 billion in a stake in chip design tool giant Synopsys and is promoting deep collaboration between the two in EDA tools. They will integrate CUDA-X, Omniverse, and NeMo AI microservices into chip design simulation, accelerating processes from several weeks to mere hours. This significantly boosts R&D efficiency across various fields from semiconductors to automotive and digital twins.

The Chinese team DeepSeek has released the new generation open-source reasoning models V3.2 and V3.2-Speciale. With 685 billion parameters and licensed under MIT, they compete with GPT-5/Gemini 3 Pro in areas like reasoning and mathematics. The Speciale model ranked 10th globally in the 2025 International Olympiad in Informatics. Their pricing is far lower than US closed-source products, driving a reshaping of the global large model pricing landscape.

At the AWS re:Invent conference, Amazon launched multiple Nova 2 series large models (covering base, speech, and multimodal) supporting enterprise hybrid customization. It also introduced the Trainium 3 chip, Nova Act, and AgentCore platform, enhancing integrated AI model customization and agent development, marking a comprehensive entry into the AI ecosystem and enterprise application scenarios.

Tesla released a new agile version of its Optimus robot, while Figure AI simultaneously unveiled its high-speed humanoid Figure 03, becoming a social media sensation. UK company Humanoid launched the HMND 01 Alpha, achieving stable walking within 48 hours of assembly, utilizing NVIDIA Isaac Sim for efficient training. This reflects the accelerating pace of commercializing humanoid robot technology.

The US Department of Commerce is promoting a new round of robotics strategy, including tax incentives, federal investment, strengthening the domestic supply chain, and addressing Chinese subsidies and intellectual property policies. Currently, the US has only 440,000 operational industrial robots, far behind China's 2 million, highlighting an industrial policy shift aimed at catching up with China.

Global RPA leader UiPath has launched version 2025.10, enhancing its Maestro platform's agent orchestration and multi-agent collaborative automation. It also improves AI processing capabilities and orchestration efficiency in document and testing fields, accelerating the iteration of tools like Studio and Autopilot, further helping enterprises achieve end-to-end process AI automation.

NVIDIA has released the Alpamayo-R1 vision-language reasoning model and simultaneously launched several AI tools for autonomous driving and robotics R&D. This promotes the further deployment and integration of vision, perception, and reasoning capabilities in scenarios like autonomous driving and robotics, enhancing AI autonomy in complex environments and accelerating the efficiency of innovative engineering trials.