Technology is fun, as are global events. Still figuring out some of the nuances of product and AI. Retweet/Like != Endorsement. Looking at product roles rn.
The world just built a machine that drinks clean water like a river.
And nobody is talking about what that actually means.
🚨 🚨 🚨 AI DATA CENTERS WILL CONSUME ENOUGH CLEAN WATER TO SUPPLY 1.3 BILLION PEOPLE BY 2030 — PER A UNITED NATIONS REPORT 🚨 🚨 🚨
A United Nations report just put a number on it. Not a vague warning. Not a projection buried in a footnote. A hard figure: AI data centers will consume enough freshwater to supply 1.3 billion people annually by 2030. That is not a typo. One sector. One cooling system. 1.3 BILLION people's worth of water.
THE WEAPON:
→ Global data center water consumption: 560 billion liters per year right now
→ Projected consumption by 2030: 1,200 billion liters (IEA)
→ US AI data centers alone: 17 billion gallons consumed in 2023
→ US projection by 2028: 38–73 billion gallons annually
→ Single large facility withdrawal: up to 5 million gallons per day — the daily need of a city of 50,000 people
→ Evaporative cooling rate: ~80% of withdrawn water evaporates and is NEVER returned
→ Per-query footprint: every 100-word AI prompt consumes roughly one 16.9 oz bottle of water
→ Indirect footprint: electricity generation adds another 60% on top of direct consumption
THE TARGET:
→ Phoenix region alone: data center water use projected to rise 870% — from 385 million to 3.7 billion gallons per year
→ Google's Council Bluffs facility: 3.9 million gallons withdrawn daily, on average
→ Hyperscale facilities expected to account for half of all future US AI water consumption
THE MATH:
→ 2021 US data center baseline: 163.7 billion gallons annually
→ Five years of AI acceleration later: 300%+ growth for key operators
→ By 2030: one sector's cooling needs rival or exceed the municipal water supply of entire countries
→ 1.3 billion people. That is the population of Africa. That is more than the entire Western Hemisphere.
Read that again.
💀 There are currently 2 billion people on Earth without reliable access to clean water
💀 AI infrastructure is being built fastest in water-stressed regions — Phoenix, the American Southwest, the Middle East
💀 Once evaporated, that water does not come back — it is not recycled, it is not returned to the aquifer, it is gone
⚠️ The world is already withdrawing freshwater faster than it is being replenished
⚠️ Agriculture uses 70–80% of global freshwater — AI is now the fastest-growing new competitor for what remains
⚠️ This is not a future problem. 560 billion liters are being consumed right now, today, this year.
They're showing you the AI boom.
They're NOT showing you what is being drained to power it — the aquifers that took thousands of years to fill, the rivers already running low, the municipal systems already competing with server farms for the same water table.
You don't build a machine that evaporates 80% of everything it drinks in a water-stressed world and call it progress. You don't scale that machine by 300% in five years and assume the water will keep showing up. You do that when you have decided, consciously or not, that the output is worth more than the resource that keeps 8 billion people alive.
I'll keep you updated. Turn on notifications. 🚨
🚨 SCIENTISTS EXPLAIN: Light doesn’t actually “slow down” in glass time does. And that’s exactly why rainbows exist.
For centuries we were taught that light slows down when it enters glass or water, causing refraction. But the deeper reality is more beautiful: light still travels at c between atoms. What changes is the time delay caused by constant absorption and re-emission by the material’s electrons.
This tiny delay is different for every wavelength → which is why white light splits into a rainbow.
Why this matters:
• In vacuum, light always travels at c
• In glass/water, the phase velocity and group velocity appear slower due to interactions with matter
• Different colors (wavelengths) experience different delays → dispersion
• This is what creates rainbows, prisms, and the beautiful colors we see in nature
The deeper implication is mind-bending:
Light doesn’t “slow down” like a car hitting traffic. It’s constantly being absorbed and re-emitted by atoms, and the accumulated time delay reshapes how the wave propagates. The universe uses time itself as a tool to bend light and paint rainbows across the sky.
What do you think is this one of the most elegant explanations in physics?
Follow for more frontier optics and quantum explanations.
🚨 S&P 500 JUST ENTERED A 94.1% TRAP
16 out of the last 17 midterm election years, the S&P 500 fell from May to October.
16 out of 17.
That is a 94.1% hit rate.
Some of the worst drops:
1974: -32%
2002: -30%
1962: -21%
1966: -21%
2022: -19%
May → October. Over and over again.
Now look at 2026:
Rate hikes are back on the table.
Inflation just hit its fastest pace in 3 years.
The 10Y yield is above 4.60%.
Mortgage rates are back above 6.5%.
War with Iran is escalating.
And the S&P 500 just hit a new all-time high.
The market gives you strength at the worst possible time.
Midterm year. Peak uncertainty.
Worst statistical window of the cycle.
And history says May to October is when this trap usually closes.
Reminder: I’ve called all the market tops and bottoms for the last 15 years, including the Bitcoin bottom at $16,000 and the top at $126,000.
The next call will be even more important.
When I exit the markets completely, I’ll post it here publicly like I always do.
Turn notifications on. If you’re not following yet, you’ll understand why that was a mistake later.
🩸 WARNING: $BTC JUST BROKE THE MOST IMPORTANT LINE IN ITS HISTORY.
14 years of support. Gone.
That trendline survived Mt. Gox. 2018. COVID. FTX.
It just failed.
Someone just turned a cheap USB stick into a private offline AI assistant.
It’s called Portable-AI-USB.
An open-source setup that runs an AI chatbot entirely from a flash drive.
No cloud.
No login.
No internet after setup.
No data leaving your USB.
You plug it into a computer and run your own local AI assistant.
What it gives you:
• Runs from USB
• Works offline after setup
• Keeps data local
• Works on Windows, Mac, and Linux
• Uses Ollama + AnythingLLM
• Supports multiple AI models
• Lets you bring your own GGUF model
• Needs 16GB+ USB space
• 32GB recommended for multiple models
The wild part:
You can carry your AI setup in your pocket.
Your chats.
Your models.
Your files.
Your workspace.
All on one drive.
Most people use AI like a rented website.
ChatGPT tab.
Cloud account.
Internet required.
Data sent somewhere else.
Portable-AI-USB flips that.
Download once.
Install on USB.
Pick your model.
Run offline anywhere.
This is not trying to beat ChatGPT.
It solves a different problem:
Private AI you can physically carry.
Perfect for travel, research, offline work, sensitive notes, or using AI on machines you do not want to install anything on.
Welcome to the NVIDIA RTX Spark channel.
A new superchip for the age of personal AI.
Don't worry, your favorite NVIDIA local AI content continues on right here, just with a new headliner.
Let's get started...
🔥 NVIDIA ANNOUNCES RTX SPARK: ARM-BASED PROCESSOR WITH RTX 5070-LEVEL GPU, GREAT BATTERY LIFE, AND HIGH AI PERFORMANCE
NVIDIA unveiled its new ARM-based RTX Spark processor:
• GPU performance equivalent to RTX 5070
• Runs modern games at 1440p with 100 FPS
• Strong performance even on battery (no drop when unplugged)
• Long battery life
• Targets both laptops and desktops
• Shown running 007 First Light and Forza Horizon 6
• High AI processing power
• Expected Fall 2026
Big step forward for portable high-performance computing.
None of this is satire.
→ A company spent $500,000,000 on Claude in one month because nobody set usage limits
→ Uber ran leaderboards ranking engineers by how much AI they used, not what they shipped
→ Uber burned their entire 2026 budget by April. Their COO said he can’t connect any of it to consumer features
→ A CTO told Axios employees were using enterprise AI to check the weather
→ Microsoft canceled most Claude Code licenses because the token bill spiraled
→ Companies are now laying people off to pay the AI bill. Not because AI replaced the work. Because the bill replaced the headcount.
Grok foundation model V9-Medium (1.5T) has finished training. Evals look good. A lot of Cursor data was added in supplementary training and there is more to come.
Fine-tuning is underway and reinforcement learning begins in a few days. 2 to 3 weeks to public release.
This will be a major improvement over the 0.5T v8-small that currently serves all Grok production traffic, especially for difficult coding tasks.
Keir Starmer says his “strong leadership” has cut a family of 4 Alton Towers ticket from £200 to £198.36 under the ‘Great British Summer Savings Scheme’.
He said: “It currently costs £200 for a family of 4 to visit Alton Towers but with my great British summer savings scheme that is reduced to £198.36.” (Satire)