Chinese startup MindOn in Shenzhen dropped a demo of their “robot brain”.
Watch it autonomously water plants, open curtains, tidy rooms and carry packages & clean.
MindOn’s world model turns the Unitree G1 into an independent chore bot.

🚨 JUST IN: The astronauts are currently SUITING UP for the launch of Artemis II to the Moon, from Florida's Kennedy Space Center

HISTORIC!

MIT students developed a magnetically controlled, shape shifting robot that can turn into anything.

The Artemis II astronauts, now suited up for launch, are headed to the launch pad.

In a few hours, four astronauts will leave Earth's orbit for the first time since 1972.

Artemis II isn't just a space mission. It's a $93 billion procurement pipeline.

The companies building this ecosystem are already public:
$RKLB - Builds satellite buses + operates launch infrastructure
$LUNR - Contracted to deliver payloads to the lunar surface
$ASTS - Testing direct-to-phone satellite comms that Artemis-era networks need
$PL - Earth observation data that feeds mission planning

The Artemis program exists because governments decided the Moon is infrastructure, not exploration.

That changes the entire investment thesis.

NOW - Artemis II Moon crew head the the launch pad on a road filled with "crowds" and "lined with people."

I POINTED A NASA SATELLITE AT 6 COUNTRIES AND STARTED TRADING OFF FIRE COORDINATES.

$500 -> $7,778. 16 trades. 11 wins, 5 losses.

All on prediction markets. All on free data from space.

NASA publishes thermal anomalies from orbit. Every 6 hours. The VIIRS satellite scans the entire planet and logs every fire, explosion, missile strike. Coordinates, power in megawatts, timestamp.

This is for forestry rangers. They track fires in Montana. I track fires in Gaza.

6 regions. Simultaneously. Middle East, Ukraine/Russia, Taiwan Strait, Korean Peninsula, Israel/Gaza, Persian Gulf.

One API call per region - fire count over 24 hours in any rectangle on Earth.

The logic: 14-day rolling average. If today's fires are 3x the baseline - anomaly.

Ran 270 days blind. The bot didn't know dates, didn't read news. Just numbers.

Gaza, August 2023: 6 fires per day. Normal. October 7: 69. An 11x anomaly. The bot screamed hours before most people grasped the scale.

Kursk, summer 2024: zero fires for weeks. A quiet agricultural region. August 2: 16 fires. August 3: 21 fires, 94 megawatts. Heavy equipment. The official offensive began August 6. The bot saw it 4 days early.

Syria, November 2024: 76 fires. Then 81. Then 61. Three weeks before Assad fell.

One satellite - not enough. Added 9 more sources. OpenSky - military aircraft. USGS - seismic activity. VIX, gold, crypto, forex, fear & greed.

10 inputs. Claude analyzes everything and outputs: contract, price, edge, position size, BUY or SKIP.

Dynamic Kelly: 11x anomaly + 6 confirming sources = 50% of bankroll. A random 3x blip with one source = 3%. A false alarm cost $41. A real signal returned $470.

5 losses - one mistake. The bot kept buying Kursk after the region was on fire permanently. The anomaly became the new normal. The edge evaporated.

Anomaly detection works on the first spike. Not on the continuation.

News tells you what happened. Coordinates tell you what's about to.

Someone built an interactive map of the world's entire electricity infrastructure.

Every dot is a piece of grid infrastructure. Every line carries gas, power, or data.

Free. Public. OpenGridWorks.com

How is this not everywhere?

🚨Artemis II crew is now boarding Orion and preparing system checks

Just under 4 hours till launch 🚀

Market Cheering 👏

Go Artemis II

Me and my homies at the gas station when Trump said the war with Iran is over

2 years ago Claude was a research project.

Now:

> $1B in a single quarter
> 19M users.
> $350B valuation.
> $41.5B in funding.
> And now it's trading your 5 min BTC on Polymarket

Chinese AI labs got caught stealing its capabilities.

"What a time to be alive" is an understatement.

Shenzhen just switched on China’s first 10,000-card AI cluster built with Huawei Ascend chips, marking a serious jump in China’s effort to build its own large-scale AI infrastructure.

This new phase delivers 11,000 petaflops, and with Shenzhen’s earlier 3,000-petaflop phase, the site now reaches 14,000 petaflops, which shows demand is no longer theoretical.

A 92% booking rate says local AI labs, robotics firms, and universities already need far more compute than the market can easily supply.

The technical catch is chip efficiency, because reports say Ascend 910C runs at about 60% of an Nvidia H100, so China is using scale, system design, and software compatibility work to close a hardware gap rather than pretending it does not exist.

The new steam age.

This is actually becoming true.

It's possible to do so much more on your own now.

$XYZ Block's Business lead, Owen Jennings:

The link between company headcount and output broke in early December, after AI agents quickly became capable of working inside large, complex codebases.

1 engineer + 1 designer with AI, now move 10–100× faster.

Chamath exposed the biggest accounting illusion in Silicon Valley history.

The two most powerful AI companies on earth are reporting revenue in completely different ways and almost nobody in the media has bothered to check the math.

OpenAI says it has $25 billion in annual revenue while Anthropic says $19 billion.

Every headline screams that Anthropic is closing the gap at breathtaking speed but those two numbers are not measuring the same thing.

OpenAI counts only the money it actually keeps, when Microsoft resells its AI through Azure, OpenAI deducts Microsoft's cut before reporting a single dollar as revenue.

Anthropic does the opposite, when Amazon Web Services resells Claude, Anthropic books the full gross amount including the portion it immediately hands back to Amazon.

Bank of America analysts estimated Anthropic could owe hyperscale cloud partners as much as $6.6 billion in revenue sharing in 2026 alone.

That money flows in and flows straight back out the door but anthropic still counts it as revenue.

Strip that out, and the competitive gap between these two companies looks nothing like the headlines.

Here is what the actual businesses look like underneath.

OpenAI is a consumer company, roughly 75% of its revenue comes from people paying monthly for ChatGPT.

Anthropic is almost entirely enterprise API, embedded inside tools like GitHub Copilot and Cursor.

These are structurally different businesses on different growth curves with different margin profiles.

Both businesses are genuinely historic.

Anthropic grew from $1 billion to $14 billion in annual revenue in roughly twelve months while OpenAI crossed $20 billion in ARR and is adding roughly $1 billion in new API revenue every single month.

But the next time a headline tells you one is overtaking the other, remember what Chamath just said.

The scoreboard is broken and when the IPO filings land, the whole race gets run again with real numbers.