Missed this when I was off earlier this week, but on Monday the EU Commission spox confirmed what I'd reported last Friday:
The Commission will also come up with a solidarity mechanism to counter Chinese retaliation against any of its own trade measures
China’s humanoid robot boom has a dirty little Secret
Here’s how the game allegedly works:
A local government SOE places a massive order—sometimes 500-600 humanoid robots worth ~$25M+.
The company books revenue, gets cash upfront, and now has a great slide for investors: “Government customer secured.”
The robots don’t end up working in factories.
Instead, many are shipped to “data collection centers,” supposedly to generate training data for embodied AI.
Some people compare these facilities to the giant parking lots where unsold EVs piled up before China’s EV subsidy boom cooled off.
The government entity leases the robots to an operator.
Then the data collected is sold back to the robot company’s own AI/data platform.
Money goes in a circle.
On paper:
• Robot sales
• Data services
In reality:
Cash out through robot orders.
Cash back through data purchases.
Revenue gets booked.
Order backlog grows.
Valuation narrative gets stronger.
Meanwhile, China already has close to 100 humanoid data collection centers built or under construction.
This is a reminder that in every investment boom, headline order numbers don’t necessarily tell you where the real demand is.
I didn’t really consider the possibility that an Anthropic customer would sue to enjoin BIS’s export control action.
The legal arguments in the complaint are very strong, as you’d expect. If Anthropic was making them, they’d probably prevail. But my somewhat uninformed guess is that this other company probably doesn’t have standing. That said, they make a pretty decent standing argument in the complaint, so who knows.
Most likely outcome is that the situation is resolved soon by negotiations between Anthropic and USG and this case is moot. But you never know. Worth keeping an eye on.
China: *accidentally awakens giant sea monster that eats Shanghai and Shenzhen*
Jostein Hauge: "Despite being destroyed by a sea monster, Chinese cities are amazing places to live"
Paul Triolo: "This demonstrates China's superiority over the West in sea monster technology"
Arnaud Bertrand: "Western narratives are getting the sea monster story all wrong. Here's what they're not telling you"
Glenn Luk: "Here's the story of how decades of successful Chinese industrial policy resulted in the sea monster breakthrough"
David Fishman: "If you actually went to China instead of just tweeting about it, you could witness the sea monster for yourself"
Carl Zha: "This sea monster was obviously summoned by the Zionists. Subscribe to my YouTube for more info"
Hu Xijin: "There is no sea monster. It's a fabrication based on --" (internet access mysteriously cuts off)
Michael Pettis: "This sea monster is just another unsustainable way to prioritize investment over consumption"
Brad Setser: "Here's how the sea monster added to China's current account surplus"
Dan Wang: "In China, the problem is too many sea monsters. In America, the problem is too few"
Rui Ma: "Here's how the sea monster makes China a great investment story"
Lingling Wei: "Sea monsters are bad"
Michael Ron Bowling: "We must match China in sea monster weaponry capabilities"
Chen Weihua: "Shut up. YOU are the sea monster. Bitch!!"
Mistral employees use their own models only "one day a week to give feedback".
The rest of the time they use Claude Code or Codex.
Everyone is hyping the Palantir FDE playbook right now. But what Mistral is running is the opposite of Palantir. It's the oldest move in the book.
For 30 years the pattern has been the same. Your product doesn't sell, so you start selling your time. Consulting is what failure looks like when it still has a cool logo.
Mistral's actual playbook:
1. Release impressive open source models
2. Get famous for it
3. Hire top-tier AI/SWE talents
4. Sell their time through enterprise consulting deals, like Capgemini or Accenture
The team is genuinely talented. That's what makes it sad.
They stopped competing on the model and the product, which is the only place the future gets decided.
You don't reach the frontier building a RAG chatbot for the French unemployment agency (France Travail).
I'm starting to realize that the problem here is that Anthropic is a company of nerds who communicate via whitepapers and think-pieces, faced with an admin that doesn't like to read.
Reporting in Handelsblatt:
"Ausschlaggebend für den Kurswechsel des Kanzlers war unter anderem ein Forschungspapier der Ökonomen Tordoir und Setser. Merz habe sich die Studie daraufhin vorlegen lassen und sei zu dem Schluss gekommen, dass Europa neue Handelsinstrumente brauche"
U.S. government representatives are holding secret talks with Berlin on drug prices, according to three people familiar with the talks.
The Trump administration is sending a message: American patients are paying too much, and Europe better chip in.
🔗 https://t.co/4T5Ogalkgi
Key point the European (micro)managerial class ignores: You can't decide top-down that our continent becomes a peer to the US in AI, space or any other critical technology. You can only create the framework to enable it bottom-up by the builders. Whom we prefer to tax & regulate.
How does something like this not torpedo the AI intelligence explosion bull case? USG establishing precedent that access to anything as smart as Fable 5 (which is not RSI and nowhere near AGI) will be banned.
Even if Ant could make the model accessible to US nationals, how will any customer ensure compliance seeing as not all employees of US enterprises are US nationals…
So we have a situation where the labs need to spend increasing amounts of capex to build more powerful models, but are restricted from monetising them.
I do not think the intelligence levels of opus 4.8 and gpt 5.5 are enough to justify anywhere close to the amount of AI capex being spent, let alone projected to be spent…
People who live in areas with brutal winters always overdo summer. Try to fit in as many activities as possible. You can feel the tension with them. Winter is always on their mind, even in the middle of July
@internpierre@michaeljburry I don’t think the French Riviera is the right proxy. Alo is mostly aspirational atleisure while Lululemon is more “inclusive” and performance oriented.
(Lululemon still needs to replace their creative director 🤦♂️)