SpaceX Acquires Cursor for $60B, DeepSeek Raises $7.4B, and Z AI Releases GLM-5.2
SpaceX has acquired AI coding startup Cursor in a $60 billion all-stock deal. Meanwhile, DeepSeek secured $7.4 billion in funding, Z AI launched the GLM-5.2 model, and Meta is addressing internal morale issues.
Podcast В· 4 min
SpaceX Acquires Cursor in $60B Deal
SpaceX has officially acquired AI coding startup Cursor in an all-stock transaction valued at $60 billion. The deal, expected to close in Q3, follows SpaceX's recent IPO and aims to integrate Cursor's AI coding capabilities into SpaceX's internal software development stack, including rockets, satellites, and Grok. Cursor, which has raised $3.38 billion since 2022, will continue its operations as part of the SpaceX ecosystem.
DeepSeek Raises $7.4B at $50B+ Valuation
Chinese AI lab DeepSeek has reportedly raised over $7.4 billion in a new funding round, pushing its valuation above $50 billion. The deal structure, which includes a $3 billion contribution from founder Liang Wenfeng, is designed to preserve founder control. This makes DeepSeek one of the most valuable AI startups in China, signaling continued aggressive investment in the region's AI capabilities.
Z AI Releases GLM-5.2
Z AI has released GLM-5.2, an open-weights model featuring a 1 million-token context window and competitive performance against GPT-5.5 and Claude Opus 4.8. The model is released under an MIT license and is designed to handle complex long-horizon coding and reasoning tasks. It offers two effort modes, High and Max, to address varying task difficulties.
SubQ Releases 12M-Token Model Report
SubQ has published the technical report for SubQ 1.1 Small, a long-context model claiming near-perfect retrieval up to 12 million tokens. The report indicates the model achieves this with 64.5x less compute than dense attention at 1 million tokens, marking a significant efficiency milestone for long-context processing.
OpenAI Launches Deployment Simulation
OpenAI has released Deployment Simulation, a new tool designed to test candidate models on de-identified real conversation patterns. The system aims to predict and mitigate risky model behaviors before public launch, enhancing safety protocols for future releases.
Meta Addresses Internal Morale Crisis
Meta CTO Andrew Bosworth has pledged a culture reset following reports of sinking morale within the company's AI division. The memo, which surfaced after reports of forced transfers and 'gulag'-like conditions, promises manager report caps, reduced shuffling, and improved office perks. The move comes as Meta attempts to stabilize its workforce following a major AI reorganization in March.
Anthropic Pauses Token-Based Billing for Claude Agent SDK
Anthropic has paused token-based billing for its Claude Agent SDK following backlash from heavy users. The decision comes after users reported concerns over billing structures, and the company is currently reevaluating its pricing model for agentic workflows.
Trump Administration Rejects Anthropic Export Ban Exemption
The Trump administration has reportedly rejected a request from UK Prime Minister Keir Starmer to exempt G7 allies from the export ban on Anthropic's top models. This follows the recent US government restrictions on foreign access to advanced AI models, highlighting ongoing geopolitical tensions regarding AI safety and export controls.
Alibaba Launches Qwen Robot Suite
Alibaba has launched the Qwen Robot Suite, a set of models specifically designed for robot navigation, object manipulation, and world prediction. This move marks Alibaba's push to extend its Qwen models beyond chat applications and into 'physical world intelligence' for robotics.