In partnership with
Hey Radarians, this week is about the massive infrastructure behind AI.
Custom chips, huge compute deals, smarter project planning, brain-like speech models, and a practical course side hustle.
Let’s dive in.
What’s Inside 📡
Radar News Brief
Steal This Workflow
AI Research
Tools Worth Watching
Prompt Radar
AI Side Hustle Idea
Learn AI with This Course
Radar News Brief
⚡ Amazon taps Qualcomm to build custom AI data center chips
Amazon and Qualcomm are teaming up across multiple chip generations to build customized silicon for AI inference, giving AWS another route to expand its AI infrastructure
The partnership also covers high-speed optical connectivity up to 1.6 terabits per second, while Qualcomm will use AWS and Amazon Bedrock to speed up its own chip design
Reuters says the relationship could be worth up to $60 billion over time, with Amazon also receiving warrants that could let it buy roughly $4 billion of Qualcomm stock
🏗️ Anthropic lines up $517B in AI compute
Anthropic has agreed to deals that could cost up to $517 billion over the coming decade, adding at least 14.8 gigawatts of computing capacity since October 2025
Amazon and Google account for roughly 11 gigawatts and more than $300 billion of that capacity, while Anthropic is also working with Microsoft, SpaceX, AMD and others
The expansion comes months after Dario Amodei warned that overbuying compute could bankrupt AI companies if revenue growth misses forecasts, even as Anthropic’s revenue tops $65 billion
The Future of AI in Marketing. Your Shortcut to Smarter, Faster Marketing.
Unlock a focused set of AI strategies built to streamline your work and maximize impact. This guide delivers the practical tactics and tools marketers need to start seeing results right away:
7 high-impact AI strategies to accelerate your marketing performance
Practical use cases for content creation, lead gen, and personalization
Expert insights into how top marketers are using AI today
A framework to evaluate and implement AI tools efficiently
Stay ahead of the curve with these top strategies AI helped develop for marketers, built for real-world results.
More News on the Radar
Mistral AI raised €3B at a €21B valuation, Europe’s largest private tech round, to expand AI models and research
Anthropic may launch Fable 5.1’s successor by October, with the latest model expected to surpass Astra before its IPO
A Northern Ireland victim lost £250,000 after scammers used an AI video of a financial expert to promote a fake investment
⚡ Steal This Workflow
Build an AI Pre-Mortem Workflow
Do not wait for a project to fail before figuring out what went wrong. Use AI to imagine the failure first, then strengthen the plan before you start.
Give AI the project plan
Open ChatGPT, Claude, or Gemini and provide:
The project goal
Timeline
Team and resources
Important assumptions
Dependencies
Success criteria
Include enough context for AI to understand how the project is supposed to work.
Pretend the project failed
Tell AI to imagine it is six months later and the project has failed badly.
Use this prompt:
Imagine it is six months from present and this project has failed to achieve its goal. Work backward and identify the most plausible reasons why. Look for problems involving strategy, execution, assumptions, people, resources, timing, dependencies, customers, and external risks. Rank each failure cause by likelihood and potential impact. Do not invent specific facts that are not supported by the project information.This follows the traditional premortem approach of assuming failure first and then identifying what could have caused it.
Turn risks into prevention
For every major failure scenario, ask AI to create:
Warning sign → What would tell us this problem is starting?
Preventive action → What can we change?
Checkpoint → When should we review it?
Owner → Who should watch it?
Focus first on risks that are both likely and expensive to ignore.
Add the protections to the plan
Move the strongest preventive actions into your actual project plan, task manager, or calendar.
Schedule the checkpoints before the risky moments arrive.
🔎 AI RESEARCH: An AI Speech Model Started Looking Surprisingly Similar to the Human Brain
What researchers studied:
Google researchers compared activity inside the Whisper speech-to-text model with direct brain recordings from four people during about 100 hours of natural conversations.
What they found:
Whisper’s internal speech patterns closely matched activity in brain areas involved in hearing and producing speech, while its language patterns matched higher-level areas involved in understanding words and meaning.
The timing was also similar. When people listened, speech-related brain activity appeared before language-related activity. When they spoke, the order reversed as the brain first prepared what to say and then how to say it.
Why it matters:
Whisper was built to recognize speech, not copy the human brain. Yet both appear to organize parts of speech and language in surprisingly similar ways.
Source: Google Research / Nature Human Behaviour, March 2025
🛠️ Tools Worth Watching
PureBox: AI inbox cleanup tool sorting Gmail into three categories
Perplexity Portable Computer: Local AI agent running models directly on your hardware
SpacebarX: Offline workspace for notes, tasks, writing, code, and projects
AI can build faster. Can your team decide better?
AI can draft the PRD and prototype the idea. Jira Product Discovery helps teams decide whether it belongs on the roadmap. Bring feedback and ideas together, prioritize as a team, and keep your roadmap connected to delivery in Jira.
📝 Prompt Radar: Create Clear Handoff Document
What it does: Makes it easy for another person to continue work without missing context.
Work as a project handoff specialist. Using the material I provide, create a handoff document for someone taking responsibility for the work. Include the objective, current status, work already completed, important decisions and why they were made, key files or resources, stakeholders, deadlines, dependencies, unresolved problems, known risks, outstanding tasks, owners, and immediate next steps. Identify information that exists mainly as informal knowledge and should be documented. Do not hide failed attempts or uncertainty if they may prevent repeated mistakes. Organize the document so someone can understand the current situation quickly and then review deeper context if needed. End with “Start Here,” “Do Not Forget,” and “Questions Still Unresolved.” Information: [PASTE].Best for:
Teams
Projects
Client work
💼 AI Side Hustle Idea: AI Course Content Assistant
Most experts don’t need “more course ideas.” They need a faster way to turn what they already know into structured learning material.
The problem:
Experts may have hours of recordings, notes, workshops, and presentations, but turning them into organized lessons, exercises, quizzes, worksheets, and student resources takes significant time.
Your AI service:
Turn the expert’s existing material into lesson outlines, learning objectives, summaries, worksheets, quizzes, practical exercises, and FAQs, then organize everything into a consistent course structure.
Simple workflow:
Recordings + notes → extract key concepts → build lesson structure → create activities + quizzes → expert review → course-ready materials
Best clients:
Coaches, consultants, educators, corporate trainers, creators, subject-matter experts, and course businesses.
What you can charge for:
Per-module package + full course development + ongoing content updates.
Learn AI with This Course
What you’ll learn:
Learn how to build and use AI agents for real work, automate multi-step tasks, connect AI with other tools, and create reusable workflows without needing coding skills.
Key topics:
Using advanced prompting and agentic workflows for complex tasks
Building reusable AI assistants with Projects, memory, and custom instructions
Connecting tools with integrations, Skills, Plugins, and autonomous agents
Until next week,
AI Radars