Hi fam!
A few years ago, most AI conversations were about what might happen someday.
Today, we're watching it happen in real time.
In this issue, we're looking at the latest developments in AI, a simple idea that can unlock better decisions, and a fascinating example of what happens when people stop optimizing a solution and start rethinking the problem itself.
The future of AI feels Like Magic, and it’s here!
Over the past few days, we've seen new questions around safety, fresh investments in AI adoption, and growing evidence that the future of AI may be just as much about infrastructure and security as it is about smarter models.
Google DeepMind has published a new roadmap outlining how it plans to monitor and control increasingly capable AI agents if they begin acting in unexpected ways.
Why this matters: AI companies are spending less time talking about how to make AI smarter and more time thinking about how to keep it under control.
China has announced a new package of measures designed to accelerate AI adoption across everyday products and services, including smart devices and humanoid robots.
Why this matters: The AI race is no longer just about building better models. It's increasingly about getting AI into the hands of millions of people as quickly as possible.
Researchers have demonstrated a new attack method called Agentjacking, which targets AI coding agents such as Claude Code, Cursor, and Codex through fake error messages and manipulated instructions.
Why this matters: As AI gains access to codebases and business systems, security is becoming just as important as model capabilities.
New financial figures highlight just how expensive it is to compete at the forefront of AI. Revenue continues to grow rapidly, but infrastructure costs are growing even faster.
Why this matters: AI may look like software, but behind every model sits an enormous amount of computing infrastructure that someone has to build, power, and maintain.
More than 75 major data center projects worth roughly $130 billion have reportedly been delayed or blocked across the U.S. this year as local communities push back over concerns about electricity consumption, water use, noise, and environmental impact.
Why this matters: We often talk about AI models, but the real bottleneck may be infrastructure. As AI demand grows, the challenge is no longer just building smarter systems—it's finding sustainable ways to power and cool them.
The 10-Minute Version Prompt
We all know some things simply can't be done in 10 minutes.
You probably won't write a book, launch a business, or master a new skill before your coffee gets cold.
But you'd be surprised how often asking for the 10-minute version can help you get started.
Try this prompt:
"How would I do this if I only had 10 minutes?"
For example:
"How would I improve my website if I only had 10 minutes?"
Or:
"How would I learn more about AI if I only had 10 minutes today?"
Or even:
"How would I prepare for a job interview if I only had 10 minutes?"
The answers are often surprisingly useful.
Why?
Because AI is forced to focus on the highest-impact actions instead of generating a long list of tasks you'll probably never get around to doing.
Instead of twenty recommendations, you might get three.
Instead of a detailed strategy, you get a practical starting point.
And sometimes that's exactly what you need.
The goal isn't to finish the job in 10 minutes.
The goal is to figure out what matters most if that's all the time you have.
Because while you may not have time to do everything, you almost always have time to do something.
And more often than we'd like to admit, that first 10-minute step is the one that actually gets the ball rolling. 🚀
Why Are AI Servers Going Underwater?
What can an underwater AI data center teach us about productivity?
Quite a lot, actually.
In this week's 99% More Efficient, we're exploring one of the most unusual AI projects we've seen recently: a data center sitting beneath the ocean floor. Instead of building bigger cooling systems, engineers asked a different question—what if the ocean could do the cooling for them?
It's a fascinating example of a principle that applies far beyond technology. The biggest gains in efficiency often don't come from doing the same thing better. They come from rethinking the problem entirely.
From The Archive: We Love PowerPoint, But F It
Back in 2024, we reviewed a wave of AI presentation tools that promised to become "PowerPoint killers."
Looking back from 2026, the prediction was both right and wrong.
AI presentation tools have improved dramatically. Platforms like Gamma, Canva AI, Napkin AI, and Plus AI can now generate slides, diagrams, visuals, and even entire presentations in minutes. Creating a first draft has never been easier.
But PowerPoint didn't disappear.
Instead, something more interesting happened. Rather than replacing PowerPoint, AI became part of the presentation workflow. Many people now use AI tools to brainstorm ideas, generate content, create visuals, or build an initial deck before refining everything in the software they already know.
The biggest shift wasn't the death of PowerPoint.
It was the rise of AI as a presentation partner.
Text generation | Image Generation | LMAI recommends |
|---|---|---|
Like Magic AI NFT 🏞️
Our master plan is to publish an NFT image in each newsletter and hand it out to our subscribers. The earlier you subscribe, the smaller the series are. It's a future collectible, a piece of digital art that captures the essence of this moment in time.
Thank you for being a valued subscriber. Together, let's embrace the magic of AI and creativity!
LMAI209-19062026
Was this email forwarded to you? Sign up here 👇
