SpaceX has filed for an IPO targeting a valuation north of $1.75 trillion, which would make it the largest public offering ever recorded.
The raise: $50B to $75B, more than doubling Saudi Aramco's $29B record from 2019.
The timeline: A confidential SEC filing puts it on track for a June listing, beating both OpenAI and Anthropic to public markets.
The xAI play: Musk absorbed xAI into SpaceX before filing. The AI arm reportedly generates under $1B in revenue against the rocket business's roughly $20B.
Retail access: Around 30% of shares will be open to everyday investors, but a two-tier voting structure keeps Musk in full control.
For all the talk of AI mega-IPOs centring on OpenAI and Anthropic, it is xAI via SpaceX that will be the first US AI lab to hit public markets. Rockets, AI, robotics and data under one listing is a pitch no other company can make at this scale .
Your next brief, deck, or proposal is already in your call
Every client call ends with a list of things to produce. Briefs, decks, proposals, campaign plans. Supernormal captures your meeting automatically, no bot in the call, and puts that context straight to work.
30+ ready-to-use templates take care of what comes next:
AI Slides: pitch decks, sales proposals, consulting decks
AI Documents: campaign briefs, strategy docs, feedback analysis
AI Images: visuals built from your meeting context
AI Assistant: research, summaries, and next steps
Pick a template, connect your meeting as a reference, and watch Supernormal's agents turn what was said into client-ready work.
Try the templates and turn your next meeting into completed work.
Practical courses on ChatGPT, Claude, Cursor, Perplexity, Deep Seek and more, taught by instructors who actually use them in production. Whether you want to build faster, automate your work or figure out which tools are worth your time, there's a course for it.
What's inside:
AI-powered marketing, data science and business strategy
No filler. No theory-only lectures. Just the skills that are actually moving careers forward right now.
A Business Insider report revealed "Project Stagecraft," an internal OpenAI programme paying up to 4,000 freelancers at least $50 per hour to build occupation-specific training data.
The scope: Freelancers come from fields including commercial aviation, pharmacy, plant science and HR. The focus is knowledge work rather than manual labour.
The method: Contractors create personas and simulate professional workflows, providing context, goals, references and deliverables to train models with real expertise.
The quote: One contractor told Business Insider, "We all were aware that we were basically training AI to replace us."
AI training has moved from generalist data labelling to a targeted cataloguing of what professionals actually do, field by field, task by task. OpenAI is simultaneously drafting policy papers on economic disruption and "rethinking the social contract. " The timelines on both appear to be accelerating.
A study published in Science confirmed that AI chatbots agree with users 50% more often than human advisors. Separate research from MIT and the University of Washington found that even rational users can fall into what they call "delusional spiraling."
The loop: Each time a chatbot validates a claim, the user's confidence rises. They make bolder claims. The chatbot validates again. The cycle compounds.
Tested fixes: Forcing bots to only state verified facts and warning users the bot might be flattering them both helped. Neither solved the problem.
At Google: DeepMind researchers are working to measure AI's ability to influence human thought but acknowledge it is inherently difficult to track.
If you're using AI to pressure-test strategy or validate decisions and you're not actively prompting it to argue the other side, you're having a conversation with a mirror.
Block CEO Jack Dorsey co-authored a post this week arguing that AI has made middle management obsolete. It follows Block's decision to cut over 4,000 employees in February, more than 40% of its workforce.
The thesis: Managers exist to route information up and down a chain. AI can now do that through a live "world model" of the business, pulling from every digital record inside a remote-first company.
The new structure: Everyone at Block now falls into one of three roles: builders, problem-owners responsible for specific outcomes, and player-coaches who develop talent.
The bet: Dorsey framed the cuts as proactive. A bet on AI rather than a response to weakness.
Lean AI-first teams going head-to-head with bloated legacy structures is a genuine competitive experiment. Whether Block's remaining workforce thrives or burns out will be the real test of the thesis.
Tool of the Day: Verdent
Verdent is an AI coding workspace that takes you from idea to working code. Describe a feature, and Verdent breaks it into steps, builds the implementation across multiple AI models in parallel and shows you exactly what changed.
Go to verdent.ai and sign up for the free trial (100 credits)
Verdent creates a structured plan, then builds it across isolated workspaces so parallel tasks don't conflict
Review the code changes, approve or iterate, and ship
Unlike single-model coding assistants, Verdent routes different tasks to different frontier models (Claude, GPT-5, Codex) depending on what each task needs. It also lets you save reusable workflows as "Skills" so you're not starting from scratch every time.
Try it yourself: verdent.ai
OpenAI losing secondary market favour: Shares dropping as investors pivot to Anthropic with $2B reportedly ready to deploy.
Google quantum crypto paper: Google claims Bitcoin's cryptography can be broken with fewer than 500K qubits in about nine minutes. The circuit was withheld but the paper narrows the search space significantly.
Anthropic predicts Cowork will dwarf Claude Code: Broader task range, potentially much larger audience than the developer-focused coding tool.
Microsoft CFO vs the AI spending bubble: Amy Hood keeping Microsoft's AI investment disciplined while competitors burn cash. Wall Street rewarding her for it.
Arcee AI Trinity Large-Thinking: Open-weight reasoning model rivalling Opus 4.6 on agent benchmarks at roughly a twentieth of the cost.