Could the first goal of the 2026 World Cup be scored by the Atlas humanoid robot?

With smooth movement, precision, balance, and power, Atlas shows impressive whole-body coordination even in simple soccer drills.
---
(Tremble, humans…humanoid robots may soon dominate the pitch)

🇺🇸 Trump just handed Pam Bondi a seat on his AI advisory council.

The former AG with no AI background will now coordinate between the U.S. government and the top names in tech: Nvidia, Meta, and Oracle.

She also picked up a new advisory role linking tech policy to data centers and AI energy infrastructure.

PCAST advises Trump on AI, quantum computing, and the technologies that determine whether the U.S. stays ahead of China.

Is political trust the new prerequisite for shaping America's AI future?

Source: White House

🇺🇸 The U.S. military just hit another drug-running vessel in the Eastern Pacific.

Joint Task Force Southern Spear struck a boat run by designated terror groups along known trafficking routes.

One narco-terrorist killed, two survivors recovered.

🇺🇸 The UH-1Y and AH-1Z might be the most “America” helicopter combo ever created:

One flies in loaded with Marines.

The other arrives carrying Hellfires, rockets, and enough firepower to erase a city block.

🇺🇸 What 3 flights of iteration actually looks like.

Starship Flights 10, 11, and 12 side-by-side.

This is what "move fast and fix things" produces when the stakes are orbit-class reusability.

🚨MACRO TOMORROW:

🇺🇸 PCE Price Index
🇺🇸 US GDP
🇺🇸 Initial Jobless Claims

It all drops at 08:30 AM ET.

Prepare for volatility!

Quant feels impossible until one day you suddenly start seeing markets as systems instead of just charts

🚨 BREAKING: Starcloud just turned Starlink’s laser network into the backbone for orbital AI data centers.

A company called Starcloud has ordered 50+ Starlink Mini Laser terminals to equip 25+ future satellites.
Not ground stations.
Not fiber cables.

Direct laser-linked computing nodes in orbit plugged straight into SpaceX’s space-based optical mesh.

This is the sci-fi future arriving now:
Orbital cloud computing
AI servers floating in space
Powered by 24/7 sunlight
Connected globally at light speed via Starlink lasers

The insane part:
Starcloud says its satellites will eventually handle full AI inference and training workloads directly in orbit.
Data won’t always need to come back to Earth to be processed.

The advantages are massive:
• Unlimited solar energy (no grid limits)
• Zero land or water constraints
• Passive radiative cooling in vacuum
• Instant global relay with zero terrestrial bottlenecks
• Near real-time Earth observation analysis
Their first major spacecraft (Starcloud-3) is designed for 200 kilowatts in orbit a full-on space-based data center node, not just a satellite.

And here’s the bigger picture:
SpaceX has filed plans for up to ONE MILLION orbital data centers of its own.

Read that again.

We may be watching the birth of the first true space-based computing infrastructure layer for civilization.
The internet already left the ground.

Now AI might be next.
What happens when the cloud literally moves into space?

Follow for more frontier physics and future technology.

🇪🇺🇺🇸 Europe wants to shut U.S. Big Tech out of its cloud and satellite markets to make room for European alternatives.

The problem: European alternatives barely exist.

The EU faces a €1 trillion investment gap versus the U.S. on cloud infrastructure.

Amazon, Microsoft, and Google hold 63% of the global cloud market combined. Their closest European rivals are in the low single digits.

Brussels is pushing sovereignty. It just hasn't figured out what to be sovereign with yet.

Source: Reuters

Boring Company’s Prufrock is turning tunnel construction into an automated production system.

It digs while lining the walls simultaneously.

Moves debris through giant underground conveyor systems.

Can reverse itself for maintenance.

And even exits the ground under its own power.

Most tunnel projects still move like somebody lost the paperwork in 1997.

Meanwhile, Boring Company is aiming for a mile per week.

That pace completely changes what cities can realistically build.

boringcompany.com

A 3D printed drone built in someone’s bedroom just raised $1B and reportedly outflies DJI on every metric.

No factory. No team. One printer, one AI chip, and a guy at a desk.

Range over 100km. Battery that outlasts anything DJI ships. Optical tracking that locks on and doesn’t let go.

Night vision baked in. A custom AI chip running onboard, making flight decisions in real time, no pilot needed.

The whole thing started as parts on a printer bed and open source code.

DJI spent 20 years and billions building a moat.

One bedroom and a spool of filament just jumped it.

So here’s the question:

if a kid with a printer can out engineer a $30B giant from his desk what exactly is the moat anymore?

A traffic camera in Bangkok now reads the speed of every car in frame at once.

Not one. Every one.

8 lanes. A red sedan at 8 km/h. A yellow hatch at 12. A black Mercedes at 53. Each box redraws 30 times a second.

The whole thing is 1 Python file. The model is YOLOv8n 6MB, the smallest weight in the family. It runs on a laptop.

It sorts traffic into 4 classes. Car. Motorcycle. Bus. Truck.

Then it does the part a human can’t. It turns pixels into meters. 3.75 m per lane. 18 px per meter. 70 meters of road. From that, speed. No radar. No officer.

The camera feed was public the whole time.

That’s the part cities want. One feed replaces a fleet of patrol cars. A junction that polices itself.

No gun. No badge. Just a script that never blinks.

docs.ultralytics.com/models/yolov8

Perplexity just open-sourced their internal security scanner. It's called Bumblebee.

It scans your machine for compromised packages, malicious browser extensions, infected editor extensions (VS Code, Cursor, Windsurf), and MCP server configs.

Read-only. Doesn't execute anything. Just tells you what's exposed.

Here's why the timing matters.

On May 11, a hacker group injected malicious code into 160+ software packages used by millions of developers. The code activated automatically the moment you installed the package. No click required. No warning. If the package was in your project, you were already compromised.

Bumblebee was built to catch exactly this. It reads the metadata on your machine, compares it against a catalog of known compromises, and flags exact matches. No package managers run. No scripts execute. Nothing on your system changes.

But the part most people will overlook is that this is the first open-source tool that treats MCP server configurations as an attack surface.

The last six months have been an MCP gold rush. People connecting servers from GitHub repos to Claude, ChatGPT, Cursor, Windsurf. Most of those installs happened on trust. A README looked clean, the tool seemed useful, the config file got added.

MCP configs can carry environment values and credentials in their env blocks. A compromised config doesn't need to run malicious code. It just needs to point somewhere it shouldn't.

Bumblebee scans those configs the same way it scans npm lockfiles and browser extensions. Same read-only approach. Same exposure catalog matching.

Free. Single binary. Zero dependencies. Apache 2.0 license.

Worth running once even if you never run it again.

github.com/perplexityai/bumblebee

Hyundai is teaching Atlas robots how to move like athletes by training them on football footage.

It's target: 30,000 humanoid robots per year in the U.S by 2028.

Natural language trading bridge for MetaTrader 5

github.com/ariadng/metatrader-mcp-server

Glencore is considering a secondary listing on the ASX.

Australia's largest pension fund just said it would be positive.

The $240 billion Rio merger failed but Glencore's ambitions haven't shrunk.

AustralianSuper portfolio manager Luke Smith at the AFR Mining Summit: "If they were to list here we think it would be positive for Glencore and for the Australian stock exchange."

The fund has already discussed the topic directly with Glencore.
A secondary ASX listing would give Australian institutional investors direct access to the world's largest commodity trading and mining house.

Glencore and Rio Tinto explored a $240 billion merger earlier this year.
Rio walked away couldn't see sufficient cost advantages.
Glencore hasn't moved on.

An ASX listing keeps it in the Australian investor conversation, builds local relationships, and positions it for the next M&A attempt.

Why Australia specifically?
Smith's answer is blunt: "The Australian share market is the best and most informed mining share market in the world."

ASX-listed miners like BHP and Rio trade with deeper, more sophisticated institutional coverage.

A secondary listing is a valuation arbitrage play as much as a capital markets move.

$109,990 Tesla Model S Plaid vs $640,000 Ferrari Luce EV

The comparison captures how far Tesla has pushed electric performance, delivering supercar pace and autonomy at a price that legacy makers like Ferrari, even with their first EV, are nowhere near matching.

Proud to unveil the Luce Migliore

A Ferrari electric vehicle designed by me and ChatGPT in 8 seconds at a cost of $0.10

Ferrari to introduce a new logo.

The side by side is actually sad why is such a prestigious company ruining their reputation