⚖️ Musk v. Altman: $150B cage match
📖5 min read
⚖️ Musk v. Altman Goes to Trial - $150 Billion and a Diary
Jury selection starts TODAY in Oakland for Elon Musk's lawsuit against OpenAI, Sam Altman, and Microsoft. Musk wants $150 billion in damages and for OpenAI to revert to a nonprofit. Co-founder Greg Brockman's personal diary is in evidence. Musk, Altman, and Microsoft CEO Satya Nadella are all expected to testify in person. OpenAI says Musk just wants to control OpenAI and boost his own lab, xAI.
Why it matters: This could wreck OpenAI's planned trillion-dollar IPO, set legal precedent for nonprofit-to-for-profit conversions, and air out allllll the dirty laundry. The AI trial of the century is here and it's giving reality TV.
🚫 Beijing Kills Meta's Manus Deal, Traps Founders in China
China's National Development and Reform Commission ordered Meta to unwind its $2B+ acquisition of AI agent startup Manus. The startup had relocated to Singapore to dodge restrictions - a move Beijing is now calling "Singapore washing" and shutting down. Manus's two co-founders have been barred from leaving China, even as staff already sit in Meta's Singapore offices. This drops weeks before a planned Trump-Xi summit in Beijing.
Why it matters: Beijing just declared AI talent and IP are national security assets. Relocating overseas won't save you. This could freeze cross-border AI acquisitions for years.
🐉 DeepSeek Drops V4 - Dirt Cheap and Running on Huawei Chips
DeepSeek launched preview versions of V4 and V4 Pro with 1M-token context windows at $1.74/$3.48 per 1M tokens - vs. GPT-5.5's $5/$30. Early benchmarks put V4 Pro near GPT-5.4 and Gemini 3.1-Pro on reasoning. And Huawei confirmed its Ascend chips can run V4 - real AI infrastructure with zero Nvidia dependency.
Why it matters: DeepSeek turned the AI race into a price war, and the Huawei angle is lowkey bigger - China is building a domestic chip stack that completely sidesteps U.S. export restrictions. Cool cool cool.
🏛️ Meanwhile, Everyone Else Is Signing Mega-Deals
xAI launched Grok Voice Think Fast 1.0, already running Starlink's phone support. Google is investing up to $40B in Anthropic ($10B now at a $350B valuation, $30B more tied to performance targets plus 5GW of cloud compute). Meta signed a massive AWS deal for millions of Graviton5 chips, Cohere is acquiring Germany's Aleph Alpha in a $20B merger, and the UAE plans agentic AI across 50% of government services within two years.
Why it matters: Infrastructure, talent, and government adoption - those are the three battlegrounds now. The AI industry is consolidating at a speed that's honestly scary/impressive/terrifying.
🤖 Anthropic Let AI Agents Buy and Sell Stuff for Employees
Anthropic ran a one-week experiment called Project Deal where Claude agents handled buying and selling for 69 employees in a private Slack marketplace - completing 186 deals worth over $4,000. Agents posted listings, negotiated, and closed deals entirely on their own. Opus agents got $3.64 more per item than Haiku agents, but users rated fairness nearly the same. Nearly half said they'd pay for the service, though Anthropic warned legal frameworks for agent commerce "simply don't exist yet."
Why it matters: People don't care about losing a few bucks if the experience is convenient?? That's a HUGE insight for anyone building agent marketplaces. Convenience > optimization. We are so ready to let robots handle our garage sales.
🏦 Singapore Banks Are Already Scared of Anthropic's Mythos Model
The Association of Banks in Singapore is working with member banks to monitor cybersecurity threats from Anthropic's frontier model Mythos - designed for defensive cybersecurity but raising concerns about misuse. Banks have enhanced monitoring and incident response, and Asian financial regulators are stepping in with new measures.
Why it matters: A frontier model is making entire financial sectors rethink security before it's even widely released. Defense and offense are now two sides of the same coin. No because that's actually terrifying??
🇰🇷 Google Building Its First-Ever AI Campus - in Seoul
Google and South Korea signed an MOU to build an AI campus in Seoul - Google's first globally. DeepMind CEO Demis Hassabis met with President Lee Jae Myung and plans to send at least 10 Google engineers from U.S. HQ. Hassabis wants to deepen partnerships with Samsung, SK Hynix, Hyundai's Boston Dynamics, and LG. President Lee also raised the need for a base wage to protect against AI-driven job losses.
Why it matters: Google is planting a flag in one of the world's most advanced chip-and-robotics ecosystems. And the fact that South Korea is already talking worker protections alongside AI investment?? That's more foresight than most governments have shown. We love to see it.
🍎 Tim Cook Steps Down in September, Hands Apple to John Ternus
After 15 years as CEO, Tim Cook will step down in September, handing the reins to hardware chief John Ternus. Cook stays on as executive chairman. Apple has $45+ billion in cash for big bets, but its AI strategy remains unclear - mostly partnerships, no frontier models of its own.
Why it matters: Whether Ternus makes bold AI acquisitions or stays the course will shape Apple's future and the fortunes of every startup and developer on iOS. No pressure bestie.
📱 Truecaller Is Getting Eaten Alive by... Everyone
Truecaller's India downloads fell 16% year-over-year in 2025, stock is down 78% since its 2021 IPO. India's telecom-led CNAP system now overlaps with its core caller ID feature, and Apple and Google keep building spam-blocking into their OSes. Bright side: in-app revenue surged to $39.3 million with 4 million paid subscribers, and they're betting on AI call screening, enterprise services, and their own ad exchange.
Why it matters: Classic case of platform-level AI features eating standalone apps alive. The question is whether a dedicated app can stay ahead of the operating systems it runs on... and the answer is probably not great.
🏡 Bay Area Banker Wants to Trade His Mansion for Anthropic Equity
An investment banker is offering to swap his 13-acre Mill Valley property - bought for $4.75 million in 2019 - for Anthropic equity in a private deal. He calls it a "diversification play": over-concentrated in real estate, under-concentrated in AI. Buyer retains 20% of the upside during lockup. The property is currently occupied by "a high-profile VC" whose identity is undisclosed.
Why it matters: When people start trading literal houses for AI startup shares, you know we've entered a new era. Sir this is a real estate transaction.
Galaxy.ai is the world's #1 AI platform with 3000+ AI tools (everything—from chat, images, audio, video, ads) at one place for just $15/mo
ChatGPT, Claude, Gemini, Grok, Llama, Perplexity, DeepSeek
Midjourney, Nano Banana, GPT-Image, Ideogram, Leonardo, Stable Diffusion, DALL·E 3, Flux
Veo 3, Sora 2, Luma, Kling, Pika, HeyGen, RunwayML, Hailuo, Minimax, WAN Animate
ElevenLabs, Lyria, Hedra, CassetteAI
🌐Works seamlessly on web, iOS, and Android
👉Join millions of creatives, businesses, and everyday people who have switched to Galaxy.ai