Agents, Chips, and Gaffes
· The Fluency Briefing
The Fluency Briefing
Your Guide to What's Happening in AI and Why It Matters to You
Thursday, June 25, 2026
Everyone's building AI that answers in milliseconds, but the real money just landed on AI that works for hours. A former Apple engineer grabbed $80 million to rebuild AI infrastructure from scratch, a politician accidentally left "Claude responded:" in draft legislation, and Scouts are now earning badges in AI literacy instead of fire-starting. Thursday's serving up a full buffet of AI growing pains.
Today in AI:
- $80 Million to Plumb the AI Pipes - Sail Research launched from stealth with $80 million to build infrastructure for long-running AI agents, not quick chatbot replies. Enterprise AI bills have tripled because agentic workflows consume tokens 50 to 500 times faster than simple chat. Fortune
- Amazon Drops Another $13 Billion on India - Amazon announced a fresh $13 billion investment to expand AWS data centers in Mumbai and Hyderabad through 2030. The company's total India commitments now stand at $48 billion, following similar mega-bets from Microsoft and Google. TechCrunch
- IBM Stacks Transistors Like a Block of Flats - IBM unveiled sub-1-nanometer chip technology that crams nearly 100 billion transistors onto a fingernail-sized chip. Tests showed 50% better performance and 70% more energy efficiency than its 2nm predecessor, though production is still years away. BBC
- Politician Accidentally Pastes "Claude Responded:" Into Legislation - Florida Rep. Anna Paulina Luna's draft amendment included the telltale phrase "Claude responded:" suggesting someone copy-pasted an AI conversation and forgot to clean it up. Her evolving explanations made the story even better. Engadget
- Notion Embeds Cursor's Coding Agent in Weeks - Notion integrated Cursor's coding agent using the Cursor SDK, letting users tag Cursor in docs or threads to plan, build, test, and open pull requests autonomously. The whole integration took just weeks, not months. Cursor
- World Cup Teams Drown in 150 Million Data Points Per Match - FIFA is tracking 150 million data points per World Cup match, with sensors in the ball logging 500 movements per second. Teams are racing to use AI to crunch that data faster than their opponents. Wired
- UK Police Built a Prediction Machine They Couldn't Trust - A WIRED investigation revealed that Avon and Somerset Police built at least 23 predictive models scoring hundreds of thousands of people using mental health records, housing data, and police intelligence, with questionable accuracy. Wired
- Scouts Trade Campfires for AI Badges - Explorer Scouts in the UK are launching badges in content creation, digital communication, and online safety after surveying nearly 3,000 teenagers. It's the Scout movement's first major overhaul in almost 25 years. The Guardian
Today's Takeaway:
Here's the thing nobody's talking about: AI infrastructure was built for a world of quick questions and instant answers, but the industry quietly moved on without updating the plumbing. Sail Research's $80 million bet, reported exclusively by Fortune, targets exactly this mismatch. When an AI agent spends hours autonomously screening job candidates or crawling codebases, it burns through tokens at 50 to 500 times the rate of a simple chatbot exchange. Goldman Sachs forecasts a 24-fold increase in token consumption by 2030. Translation: your AI bill is about to look like your teenager's phone bill in 2008.
Now layer Amazon's $13 billion India investment on top, as TechCrunch reported Thursday. Between Amazon, Microsoft, and Google, roughly $50 billion is flowing into Indian data centers alone. IBM's sub-nanometer chip tech, covered by the BBC, promises to make that hardware dramatically more efficient. These stories aren't separate headlines; they're three layers of the same problem. AI agents need cheaper compute, cheaper compute needs better chips, and better chips need massive infrastructure to house them. The companies solving this stack will define the next era of AI. The ones ignoring it will wonder why their agent bills keep tripling.
🏆 5-Minute AI Challenge
Snap Your Space, Design Your Dream Room
The challenge: Take a photo of any room in your home and use an AI image tool to reimagine it with a completely different style - whether that's cozy Scandinavian, bold maximalist, or sleek modern. No design experience required.
Step by step:
- Grab your phone and snap a clear photo of any room - your living room, bedroom, or even a home office corner works great.
- Open your favorite AI image or design tool (Adobe Firefly, Microsoft Designer, or any AI image editor that accepts photo uploads).
- Upload your photo and type a simple prompt like: 'Redesign this room in a cozy Scandinavian style with warm lighting and minimal furniture.'
- Generate 2-3 variations by tweaking your style description (try 'modern industrial' or 'bohemian with lots of plants').
- Screenshot your favorite result and save it as inspiration - or share it with a friend or contractor to communicate your vision.
Why this matters: As AI infrastructure rapidly evolves beyond simple chatbots toward software that takes real-world action autonomously - as investors are now betting $80 million on - tools that let you upload an image and get immediate, useful visual output are a glimpse of that more capable, multimodal AI future arriving right now.
The Bottom Line
The Pattern: Thursday's stories all point to the same awkward truth: AI has outgrown its original wardrobe. The infrastructure, the chips, the policies, even the Scout badges are all scrambling to catch up with what AI actually does now versus what it was designed to do.
Why It Matters: If you're running a business that uses AI, costs are about to shift dramatically. The companies paying attention to infrastructure, not just the flashy model releases, will be the ones who can actually afford to run agents at scale. Everyone else gets a very unpleasant invoice.
Your Move: Before your next AI project, ask one question: is this a quick-answer task or a long-running one? If it's the latter, your current infrastructure might be costing you five to ten times more than it should. That's worth a Thursday afternoon investigation.
What We're Working On
✨ Founding Cohort Special - 60% Off! - Use code MAF20 to join for just $20/month (regularly $50). Get weekly group sessions & workshops, self-paced courses for all levels, access to tools & templates, challenges with peer feedback, and 24/7 support community. → Join Now
✨ Free 30-Minute AI Consultation - Discover how My AI Fluency can help your business unlock the potential of AI. We'll discuss your goals, explore practical AI opportunities for your industry, and outline clear next steps. → Schedule Free Call
✨ How AI-Fluent Are You? - Test your AI fluency with our interactive quiz. See how you stack up and discover what to learn next. → Take the Quiz
💬 Community | 📞 Book a Consultation | 🌐 Website
Fluently yours, The My AI Fluency Team