SpaceX Acquires Cursor in $60 Billion All-Stock Deal, Integrating into SpaceX/xAI Ecosystem
On June 16, SpaceX filed an 8-K with the SEC, exercising its option to acquire Anysphere, the developer of AI coding tool Cursor. The transaction is all-stock, with Cursor’s implied equity value at $60 billion. The price of SpaceX Class A common stock will be calculated based on the volume-weighted average price over the 7 trading days before the closing date, expected to be completed in Q3 2026. Elon Musk stated that AI will achieve Stockfish-level programming capabilities and general computer operation. Over the past few months, SpaceXAI has jointly trained models with Cursor, and new models will be released in Cursor and Grok Build.
Sources:
- @SpaceX: https://x.com/SpaceX/status/2066873915717136548
- @elonmusk: https://x.com/elonmusk/status/2066880262668247091
- @WSJ: https://x.com/WSJ/status/2066844533619327452
- @MaxForAI: https://x.com/MaxForAI/status/2066875142362234927
- @cursor_ai: https://x.com/cursor_ai/status/2066875698346954891
DeepSeek Completes 50 Billion Yuan First Round Financing with Rare Structure: Five-Year Lock-Up, Liang Wenfeng’s Effective Control
According to The Information, DeepSeek has completed its first round of financing exceeding 50 billion yuan ($7.4 billion), with an extremely unusual structure. Investors are required to put funds into a limited partnership managed by CEO Liang Wenfeng rather than DeepSeek itself, and all investor shares have a five-year lock-up period. The only exception is the China National Artificial Intelligence Industry Investment Fund, which directly invested 1 billion yuan in DeepSeek and has voting rights. Liang Wenfeng himself contributed 20 billion yuan as the largest share, Tencent 10 billion yuan, CATL 5 billion yuan, and JD.com/NetEase/IDG Capital each 3 billion yuan. Liang’s team also conducted identity checks on fund LPs to ensure shares do not fall into unknown investors’ hands.
Sources:
- @MaxForAI: https://x.com/MaxForAI/status/2066752930922004579
- @jingyanghk: https://x.com/jingyanghk/status/2066751211462889521
Anthropic’s Multiple Crises: Pentagon Moves 2/3 of Classified AI Work, Users File Class Action, Opus 4/Sonnet 4 Taken Down
Anthropic’s relationship with the U.S. government continues to deteriorate. According to Inside Defense, the U.S. Department of Defense has moved over two-thirds of its classified network AI work from Anthropic to suppliers such as SpaceX, OpenAI, Google, and NVIDIA, because Anthropic insists on model usage boundaries while the Pentagon requires commercial AI to be free from usage policy restrictions. On the same day, a Washington user filed a federal class action lawsuit against Anthropic, alleging that its Max 5x/20x subscription plans deliver far less actual usage than advertised. Additionally, Anthropic took down Claude Opus 4 and Sonnet 4. The company has sent senior technical personnel to Washington to repair the dispute with the White House.
Sources:
- @MaxForAI: https://x.com/MaxForAI/status/2066745954729140704
- @MaxForAI: https://x.com/MaxForAI/status/2066554596684886123
- @MaxForAI: https://x.com/MaxForAI/status/2066585379344687280
- @MaxForAI: https://x.com/MaxForAI/status/2066551432166429154
Microsoft Hit with Shareholder Class Action, Accused of Concealing Azure Growth Slowdown After $357 Billion Market Cap Loss
According to Reuters, on June 15 several shareholders sued Microsoft in federal court, accusing the company of deceiving investors by failing to disclose the slowdown in Azure cloud business growth and massive AI infrastructure spending. The lawsuit is led by a Michigan pension fund, with defendants including CEO Satya Nadella and CFO Amy Hood. Previously, Microsoft’s quarterly earnings caused its stock to drop 10% in a single day, wiping out approximately $357 billion in market capitalization, the largest single-day decline in nearly six years. The lawsuit alleges that Microsoft attributed the growth slowdown to capacity constraints, but in reality was shifting resources to AI R&D and Copilot.
Sources:
Mistral New Model Rumors Spark Social Media Frenzy, ‘Le Chaton Fat’ Tops OpenRouter
Social media is abuzz with rumors that Mistral is about to release a Napoleon-class model named ‘Le Chaton Fat,’ reportedly with over 10 trillion parameters, surpassing Mythos on VoltaireBench. The model shot to the top of OpenRouter, sparking a wave of memes and fan creations. Many bloggers speculate this is a self-deprecating marketing craze started by French developers.
Sources:
- @MaxForAI: https://x.com/MaxForAI/status/2066563988884988231
- @shashankgoyal95: https://x.com/shashankgoyal95/status/2066534917329072181
- @fabianstelzer: https://x.com/fabianstelzer/status/2066485744605057245
ByteDance Launches Seedance 2.0 Mini, Reducing Video Generation Costs by About 30%
ByteDance has released Seedance 2.0 Mini, a streamlined version of Seedance 2.0, priced about 30% cheaper than the original, with twice the speed of the Fast version and comparable quality. API pricing is approximately $0.073 per second, making a 30-second ad video cost about $2.19. It supports text-to-video and image-to-video, and can be used via CapCut App, Dreamina web/desktop. With promotions, it can be up to 55% cheaper than the original.
Sources:
Novo Nordisk Confirms Hack, AI Model Checkpoints and Training Data Stolen
The renowned pharmaceutical company Novo Nordisk has been breached and confirmed the intrusion. Attackers stole trained AI model checkpoints (16GB), proprietary training datasets (407MB), complete source code and training pipeline logs, internal infrastructure maps, container images (53GB+), developer credentials, and private GitHub repository URLs. The stolen data includes information about Novo Nordisk’s internal AI agents.
Sources:
- @MaxForAI: https://x.com/MaxForAI/status/2066557897308545264
- @vxunderground: https://x.com/vxunderground/status/2066553174245220772
Multiple Developer Tool Updates: baoyu-design Supports PPT Export, Papr Lightweight RSS Client Released, Open Design Integrates 150+ Design Systems
Several bloggers released tool updates. @dotey’s baoyu-design skill now supports exporting locally generated slides as editable PPTX files, using Chromium to render each page and then mapping to PowerPoint native elements via PptxGenJS. @vista8 recommended the lightweight and fast open-source RSS client Papr, which supports custom API keys for AI summarization and Q&A, and can be installed via Homebrew. @tuturetom’s Open Design project integrates design systems from 150+ mature products, supporting batch extraction of web elements. Vercel launched drop.new for website deployment, allowing users to upload files/folders to generate temporary websites. Nous Research’s Hermes Agent partnered with Stripe to launch a payment skill suite, enabling agents to autonomously shop and pay per API usage.
Sources:
- @dotey: https://x.com/dotey/status/2066743786466689117
- @dotey: https://x.com/dotey/status/2066736451404923109
- @vista8: https://x.com/vista8/status/2066902712986546574
- @tuturetom: https://x.com/tuturetom/status/2066888096176570641
- @vista8: https://x.com/vista8/status/2066879306031337815
- @NousResearch: https://x.com/NousResearch/status/2066647737613832624
Statistics: Number of timeline scans=360 Number of bloggers hit=37 Total tweets hit=207 Weighted tweet score=161.25 Original tweets=105 RT tweets=49 Crawl attempts=2 Boundary coverage status=tail_confidently_crossed_target_boundary