Amazon Replaces Its AI Chief
Two weeks after launching its Nova 2 models at re:Invent, Amazon just announced its AGI leader is leaving. Rohit Prasad spent twelve years building Alexa and two years trying to catch up to OpenAI and Google. Now the company is betting that an infrastructure veteran can do what a voice AI pioneer couldn’t.
📰 The Rundown
🔄 Amazon’s AI Chief Exits as Company Reorganizes for “Inflection Point”
➡️ The move: Rohit Prasad will leave Amazon at the end of December after leading the company’s artificial general intelligence team since 2023. CEO Andy Jassy announced that Peter DeSantis, a 27-year Amazon veteran who built AWS infrastructure, will take over a newly unified organization combining AI models, custom chips, and quantum computing. The shake-up comes after persistent reports that Amazon’s AI efforts have lagged behind competitors.
⚡ Why it matters: This is Amazon’s clearest admission that building great AI requires more than great models. By putting an infrastructure executive in charge of AI, Jassy is betting that the race will be won by whoever can best optimize the full stack: chips, models, and cloud services working together. It’s a very different strategy than throwing more money at frontier model research.
🎯 Your takeaway: Amazon’s bet on infrastructure over pure research is a signal for everyone: the AI advantage may shift from “who has the smartest model” to “who can deploy AI most efficiently across their organization.”
🤖 Google Launches AI Agent That Runs Your Morning
➡️ The move: Google Labs released CC, an experimental AI productivity agent that delivers personalized daily briefings to your inbox. Built on Gemini, CC connects to your Gmail, Calendar, and Drive to synthesize your schedule, highlight key tasks, surface upcoming bills, and even draft email responses. The agent learns from your replies and can remember ideas and to-dos you send it. Early access is available now to Google AI Ultra ($250/month) and paid subscribers in the US and Canada.
⚡ Why it matters: This is Google’s clearest move from reactive chatbot to proactive assistant. Instead of waiting for you to ask questions, CC volunteers information it thinks you need. The agent operates entirely through email, meaning no new app to learn. But here’s the catch: CC runs outside standard Google Workspace privacy protections, raising questions about how your sensitive data is handled.
🎯 Your takeaway: The shift from “AI answers questions” to “AI anticipates needs” is happening faster than expected. Whether you try CC or not, start thinking about what you’d want a proactive assistant to surface for you each morning.
📉 AI Stocks Tumble on Infrastructure Funding Jitters
➡️ The move: Markets dropped sharply Wednesday after the Financial Times reported that Oracle’s $10 billion Michigan data center project hit funding snags with Blue Owl Capital. Nvidia fell 3.8%, AMD dropped 5.2%, and data center energy plays like GE Vernova plunged 10%. Oracle denied the story but confirmed Blue Owl isn’t part of the funding talks.
⚡ Why it matters: The selloff reveals growing investor anxiety about whether AI infrastructure spending can sustain its pace. Building the data centers that power AI requires massive capital. If funding partners are getting cold feet, it raises questions about the entire AI infrastructure buildout that companies like Nvidia depend on.
🎯 Your takeaway: The AI boom isn’t just about models and software. It’s a hardware and infrastructure story too. When the money behind data centers gets nervous, everyone in the AI ecosystem feels it.
🔧 Tool Spotlight: Google CC
Google CC is a new experimental AI agent from Google Labs that delivers a personalized “Your Day Ahead” briefing to your inbox every morning.
What makes it different: Unlike chatbots that wait for prompts, CC proactively scans your Gmail, Calendar, and Drive overnight and emails you a consolidated summary before you start your day. It highlights upcoming meetings, flags important emails, notes bills or deadlines, and even prepares draft emails you can send with one click. You interact with CC entirely through email: reply to teach it your preferences, forward it ideas to remember, or ask it to search your files.
Best for: Busy professionals who spend too much time each morning figuring out what to prioritize. Particularly valuable if you already live in Google’s ecosystem.
The catch: CC operates outside Google Workspace’s standard privacy protections and requires deep access to your personal data. Review the privacy terms carefully before enabling.
Pricing: Currently available to Google AI Ultra ($250/month) and paid Gemini subscribers in the US and Canada. Waitlist open for others.
👉 Try it: Join the waitlist at labs.google.com or sign up directly if you’re already a paid subscriber.
✨ Try This Today: The Blind Spot Check
Before you finalize any important decision, proposal, or analysis, ask AI to identify what you might be missing.
The technique: Instead of asking AI to validate your thinking, explicitly ask it to challenge it.
How to use it:
1. Share your decision, plan, or analysis with your preferred AI assistant
2. Ask: “What am I not considering? What assumptions am I making? What could go wrong that I haven’t thought about?”
3. Review the response for genuinely useful blind spots
4. Update your thinking or document why you’re comfortable with the risks
Example prompt: “I’m recommending we switch our team to a new project management tool. Here’s my analysis: [paste your reasoning]. What important factors am I not considering? What could make this switch fail?”
Why it works: We naturally seek confirmation of our existing beliefs. AI doesn’t have that bias about your specific decision. By explicitly asking for contrary perspectives, you surface risks and considerations that your brain conveniently filtered out.
Time required: 3 minutes. Could save hours of rework or regret.
✨ The Wire
🔗 Pieter Abbeel, the UC Berkeley robotics professor who joined Amazon via the Covariant acquisition, will now lead Amazon’s frontier model research team as part of the reorganization. GeekWire
🔗 Apple faces pressure to deliver on its delayed Siri upgrade in 2026 after punting AI features throughout 2025. The company spent just $12.7 billion on AI infrastructure compared to $380 billion collectively by Google, Microsoft, Meta, and Amazon. CNBC
🔗 Goldman Sachs CIO Marco Argenti predicts “personal agents” will define 2026, with intensifying US-China competition and rising AI infrastructure costs putting pressure on enterprise adoption. Fox Business
Neural Notes — AI that amplifies your value, not replaces it.
