Welcome to Jumble, your go-to source for AI news updates. This week, Google fired off the biggest AI announcement barrage in its history at I/O 2026. Meanwhile, Anthropic, SpaceX and OpenAI make major IPO news. Let’s dive in ⬇️
In today’s newsletter:
🌊 Google unleashes a wall of AI at I/O 2026
💸 Three IPO bombshells land on a single day
🔀 An OpenAI co-founder jumps ship to Anthropic
📜 The White House delays its AI order yet again
🥊 Weekly Challenge: Stress-test Google's newest model
🚀 Google I/O 2026 Was an AI Tidal Wave
At I/O on May 19, Google unveiled Gemini 3.5 Flash, Gemini Spark, Antigravity 2.0, and a Search rebuild all in one keynote. CEO Sundar Pichai said the Gemini app now serves 900M monthly users, up from 400M a year ago.
⚡ Gemini 3.5 Flash Is the New Default
The new model beats Gemini 3.1 Pro on coding and agent benchmarks while running 4x faster than rival frontier models. It's already powering AI Mode in Search and the Gemini API.
🛠️ Antigravity 2.0 Goes After Cursor
Google's agentic dev platform now ships with a desktop app, a new Go-based CLI, and an SDK for building custom agents. Gemini CLI shuts down on June 18.
🤖 Spark Is a 24/7 Personal Agent
Gemini Spark runs continuously in Google's cloud, monitoring tasks and taking actions across Gmail, Calendar, and Docs. It's included in the new $100/month AI Ultra plan.
What are you most excited to try first?
💸 AI's Race to Wall Street Begins
On May 21, Anthropic told investors it expects $10.9B in Q2 revenue and a $559M operating profit, its first quarterly profit ever. The same week, OpenAI prepped its IPO filing and SpaceX's S-1 dropped a bombshell about Anthropic's compute bill.
🚀 SpaceX Reveals $45B Anthropic Compute Deal
Buried in SpaceX's IPO filing: Anthropic is paying SpaceX $1.25B per month through May 2029 for GPU access at Colossus 1 and 2 in Memphis. That's $15B a year flowing to Elon's data centers.
📑 OpenAI Files Confidentially for IPO
OpenAI is working with Goldman Sachs and Morgan Stanley on a confidential S-1 targeting a September listing at up to a $1T valuation. Anthropic is reportedly aiming for October.
📉 The Cheap-AI Catch
CNBC reported the same week that Chinese models now serve 60% of OpenRouter traffic, up from 1% in 2024. Claude costs nearly 9x more than Zhipu for the same workload, putting pricing power at risk.
Weekly Scoop 🍦
🎯 Weekly Challenge: Put Gemini 3.5 Flash on Trial
Challenge: Google says 3.5 Flash runs 4x faster and roughly half the cost of rival frontier models. This week, see if that holds up for your actual workflow.
Here’s what to do:
📋 Step 1: Pick your daily driver Whatever model you reach for most, Claude, GPT-5.5, or Gemini 3.1 Pro, that's your incumbent.
⚡ Step 2: Run the same prompt twice Take a real task from this week and run it through both your default and Gemini 3.5 Flash in AI Studio or in your regular Gemini chat interface. Compare output, speed and reliability.
🔄 Step 3: Swap one workflow If 3.5 Flash held up, commit to it for a full week on that task and see whether the speed and price drop is worth the switch.
Is Google finally pulling ahead with distribution, or is Anthropic's first-ever profit the real story this week? See you next time! 🚀
Stay informed, stay curious, and stay ahead with Jumble!
Zoe from Jumble
