Presented by
Good morning, AI enthusiasts. SpaceX just became an AI infrastructure company. Elon Musk laid out the full plan in a 30-minute IPO roadshow video: orbital AI data centers and a joint chip factory with Tesla.
The pitch includes space-based compute at a scale no land-based provider can match. We don't know if it works. We do know no one else is even trying it.
In today's recap:
SpaceX reveals orbital AI data center design
FIFA's World Cup 2026 uses AI for offside calls
Set cost ceilings and kill switches for AI agents
Anthropic CEO publishes AI governance framework
4 new AI tools, prompts, and more
SPACEX
Musk unveils SpaceX's AI satellite datacenter design
SpaceX
Recaply: SpaceX just unveiled its AI data center satellite design in a 30-minute technical briefing, describing how it plans to build, launch, and run compute infrastructure in orbit, alongside a joint chip factory with Tesla.
Key details:
The satellite design would let SpaceX build, launch, and run AI compute in orbit at a scale that land-based data centers can't match due to power and space limits.
The briefing drew 23.7M views on X. It covered three areas: the AI satellite design, Starship rocket development, and a joint Tesla chip facility called Terafab.
The video was posted on spacexipo.com as part of the IPO roadshow, making the AI satellite the central pitch to public market investors.
SpaceX's IPO was set to price June 11, with trading starting June 12 on Nasdaq under ticker SPCX.
Why it matters: Computing power has a land problem. Data centers need power, cooling, and space, and all three are running short. SpaceX's pitch is to move the servers into orbit. It also plans to make its own chips with Tesla at Terafab. That's a full vertical play, chips to rockets to orbital datacenters. No other company is attempting all three at once. Whether it works is a separate question. But the IPO is betting investors will fund the ambition.
PRESENTED BY RISE ROBOTICS
Making Hydraulics Obsolete
Every excavator, forklift, and crane on the planet runs on hydraulic fluid. It leaks. It fails. It burns through 60% of the energy you put into it. That's been true for a hundred years.
RISE Robotics built Beltdraulics™ to fix all of that. Their patented actuator swaps out hydraulic cylinders for a fluid-free electric system that runs up to 3X faster and cuts operating costs by 50%. No oil. Full digital control. Built-in sensors that hydraulic systems can't touch.
The U.S. military is already a customer. MIT-founded. $9.3M in revenue. 20+ patents protecting the core technology. Dylan Jovine of ‘Behind the Markets’ said RISE “has all the little ingredients to be one of those really big winners.” His readers have been backing it ever since.
You can invest today through the community round on Wefunder.
FIFA
FIFA deploys AI offside tech at World Cup 2026
Fifa
Recaply: FIFA just deployed Advanced Semi-Automated Offside Technology at the World Cup for the first time, automating positional offside decisions instantly and equipping all 48 teams with AI-powered match analytics.
Key details:
SAOT sends a real-time audio alert to referees when a player is more than 10cm offside, letting them flag immediately. Each of the 1,248 players was 3D-scanned in one second for accuracy.
16 optical cameras per stadium track over 150 million data points per match, supporting VAR reviews, out-of-bounds calls, and 3D broadcast replays across all 16 venues.
Football AI Pro gives all 48 teams the same analytics access as well-funded clubs. It replaces 50-60 page paper reports with a natural-language interface any coaching staff can use.
The tournament runs June 11 to July 19, 2026, in Canada, Mexico, and the US, reaching an estimated 5 billion viewers.
Why it matters: VAR made offside calls accurate but slow. SAOT fixes the delay by automating the flag instantly. That's the obvious headline. The less obvious one is Football AI Pro. For the first time, every team in the tournament gets the same analytics access, regardless of budget. A coach from a small nation can now analyze the same data as a Premier League team. If this works at World Cup scale, every major league on earth will copy it.
GUIDES
Set cost ceilings and kill switches for AI agents
Recaply: In this tutorial, you will learn how to add hard cost ceilings and emergency kill switches to AI agents in LangChain, Crew, or Claude Code, preventing runaway loops from draining your budget before anyone notices.
Step-by-step:
Set
max_iterationsandmax_execution_timewhen creating your agent. In LangChain, that looks likeAgentExecutor(max_iterations=15, max_execution_time=120). This puts a hard cap on tool calls and total run time for any single session.Build a token counter as a callback. Track tokens per call and keep a running session total. When the total crosses your budget ceiling, throw a
BudgetExceedederror to stop the agent.Handle the
BudgetExceedederror with a kill function. Log the agent's final state, send a Slack or email alert, then shut down the loop cleanly rather than mid-step.Use
recursion_limitto protect multi-agent setups. LangGraph'srecursion_limit=25stops two agents from pinging each other forever. Each back-and-forth sends the full context again, so costs grow fast without a cap.Test your ceiling in staging first. Set a very low budget limit, run the agent, and confirm it stops cleanly and fires the alert. Once that passes, raise the limit to your real production target.
Pro tip: For Claude Code and Crew, wrap your agent with a lightweight checker that calls your API provider's real-time usage endpoint before each tool call. Most providers expose /v1/usage or equivalent, letting you enforce per-session ceilings without modifying agent internals.
TOGETHER WITH NEO
Smarter browsing. Your data never leaves the room.
Most AI tools are a trade — your data for intelligence. Norton Neo breaks that deal. Powerful built-in AI, anti-fingerprinting, VPN, and ad blocking come standard. No setup. No add-ons. No compromises. Search, summarize, and write with AI that works inside your browser and stays there.
ANTHROPIC
Anthropic CEO publishes AI governance policy framework
Getty Images
Recaply: Anthropic CEO Dario Amodei just published a government-facing essay on AI policy, outlining specific plans for safety rules, job support, biomedical approvals, civil liberties, and global AI leadership.
Key details:
The essay covers five areas where Amodei says governments must act fast. These include mandatory model safety testing, job loss support programs, FDA reform for faster drug approvals, and a ban on autonomous domestic weapons.
Amodei uses AI's growth rate to frame the urgency. Models went from barely writing code in 2021 to writing most code at top AI companies today. Congress can take years to pass a single bill.
The essay calls for governments to have the legal power to block dangerous AI deployments. Amodei says Trump's current AI executive order doesn't go far enough.
Anthropic and OpenAI are both filing for IPOs in June 2026. The essay arrives as regulatory scrutiny of AI companies is rising.
Why it matters: Most CEO policy essays are PR. This one reads like a draft for actual legislation. Amodei names specific proposals, from FDA reform to chip export controls. The timing matters. Anthropic is filing for IPO while asking governments to regulate the industry. That mix of financial interest and advocacy will draw scrutiny. But the specificity makes it hard to dismiss. When a frontier lab tells governments exactly what laws to write, people tend to listen.
TOOLS
Trending AI Tools
🎨 Generative Sliders - Krea's new real-time controls for adjusting the intensity, complexity, and movement of any image generated in Krea 2
🔧 Prometheus - Firecrawl's experimental Forward Deployed Agent that writes web data collection code from a plain description, with a self-maintaining hosted option
🎥 Avatars - ElevenLabs' studio-grade talking video creator in ElevenCreative, combining AI voices, scripts, and avatars in one place
🚀 Hamster - Multiplayer AI for product teams where design, engineering, product, and business all reason on the same work and agents deliver the output
NEWS
What Matters in AI Right Now?
Prometheus closed a $12 billion Series B at a $41 billion valuation, with Jeff Bezos describing the industrial AI startup as building an "artificial general engineer" to compress design-to-manufacturing cycles 10x faster. The 150-person company, backed by JPMorgan, BlackRock, and Goldman Sachs, targets industries like aviation and medical devices.
OpenAI announced it's acquiring Ona, a cloud execution startup that lets agents continue running after sessions close, expanding Codex beyond single-device work. Codex serves 5 million weekly users, up 400% from earlier this year, with Ona bringing persistent customer-controlled cloud environments for enterprise deployments.
Figma released a Chrome extension that turns live websites into editable Figma design layers, letting designers copy entire webpages directly into their projects. The extension converts any site's layout into fully manipulable layers, removing the need to manually rebuild real web designs in Figma.
CrowdStrike reported that Chinese entities accounted for more than 58% of state-sponsored cyberattacks targeting tech companies over the 12 months through March 2026, focused on stealing AI intellectual property. The firm said Beijing is escalating espionage to acquire AI capabilities it "cannot build fast enough on its own."
OpenAI is preparing a new AI model after GPT-5.5, with CEO Sam Altman telling employees the company expects to go public "within the next year," per The Information. OpenAI has surpassed $25 billion in annualized revenue and could file as early as September.
🧡 Enjoyed this issue?
🤝 Recommend our newsletter or leave a feedback.
How'd you like today's newsletter?
Cheers, Jason
