🗞️ Opening Line
Happy Friday! Google is in talks to launch AI data centers into orbit. Cisco cut 4,000 jobs and its stock surged 15%. And Google quietly dropped a research paper that could reshape AI the way one paper did back in 2017. Let's close the week strong.
💡 The Big Update
Google Wants to Put AI Data Centers in Space. This Isn't Science Fiction.
The Wall Street Journal reported this week that Google is in advanced talks with SpaceX to launch AI data centers into orbit. The pitch: orbital infrastructure is the lowest-cost option for AI compute in the coming years — cheaper to run, free from power grid constraints, and immune to the local opposition that's been slowing data center construction on Earth.
SpaceX is positioning this as a core part of its infrastructure strategy ahead of what could be a $1.75 trillion IPO later in 2026. Google already has history here — it invested $900 million in SpaceX back in 2015. The talks build on Google's Project Suncatcher, which aims to deploy prototype satellites by 2027.
This isn't as far-fetched as it sounds. The real bottleneck for AI right now isn't models — it's compute. Data centers require massive amounts of power, water for cooling, and land. All three are becoming scarce and politically contested. Orbital infrastructure sidesteps all three problems at once.
The broader signal: the AI infrastructure race has moved beyond servers and cables. The companies that control compute at scale — wherever it lives — control the AI economy. If SpaceX and Google pull this off, the implications stretch from cloud pricing to geopolitics to who gets access to the most powerful AI systems in the world.
⚡ Quick Commits
Cisco Cut 4,000 Jobs. Its Stock Jumped 15%. Cisco announced it's cutting nearly 4,000 employees — less than 5% of its workforce — starting May 14, while reporting better-than-expected earnings driven by AI infrastructure demand. The stock surged 15%. CEO Chuck Robbins said the companies that win in the AI era will be those with "focus, urgency, and the discipline to continuously shift investment toward areas where demand is strongest." Snap, Oracle, Meta — now Cisco. The AI efficiency trade is becoming a repeatable corporate playbook. CNBC
The Paper That Made Modern AI Just Got a Sequel In 2017 a Google research paper called "Attention Is All You Need" introduced the transformer architecture that powers every major AI model today — ChatGPT, Claude, Gemini, all of them. This week Google quietly released what the AI community is calling "Attention Is All You Need V2." Details are still emerging but the discussion across the research community is significant. The original paper changed everything. It's worth paying attention to what the sequel says. TechStartups
Bonus: OpenAI Says AI Is the New Electricity. Sound Familiar? OpenAI's Chief Global Affairs Officer this week compared AI to a utility — like electricity or the internet — and argued that government and business systems need to be entirely rebuilt around the scale of change it will bring. For anyone working in the energy industry, the analogy lands differently. The grid that powers the AI buildout is straining under the load. Half of all new US electricity demand last year came from data centers. AI isn't just like electricity — it's consuming it at an unprecedented rate. The infrastructure questions are the same on both sides of that analogy. TechStartups
🛠️ The Tool Drop
Proton Mail With AI privacy concerns growing — Meta just launched Incognito Chat this week in direct response to user anxiety about AI systems training on personal conversations — it's worth knowing about Proton Mail. It's an end-to-end encrypted email service that means even Proton can't read your messages. For business owners handling client information, sensitive communications, or anyone who wants their email to stay private, it's a free and credible alternative to Gmail. Based in Switzerland, governed by Swiss privacy law. proton.me
🎯 The So What
Data centers in orbit. Layoffs funding chip purchases. A research paper that might rewrite the rules of AI. OpenAI comparing itself to the power grid.
The infrastructure layer of AI is becoming the most contested and consequential battleground in the global economy. It's not about which chatbot writes better emails anymore. It's about who controls the compute, the power, and the physical infrastructure that AI runs on.
For business owners, the practical takeaway is the same one we keep coming back to: the companies building AI into their operations now are positioning themselves ahead of the curve. The infrastructure race happening above us — literally, if Google and SpaceX have their way — is going to determine what AI costs, who can access it, and how fast it improves.
Have a great weekend.
— The Changelog
Not subscribed yet? Get The Changelog free, three times a week. Subscribe here
Know someone who'd find this useful? Forward it along.