🔥 China Blocks Meta's $2B AI Deal, Placebo Sleep Hacks Your Bra…

🚨 The Musk vs. Altman Showdown Begins

Elon Musk's $134 billion lawsuit against Sam Altman officially kicked off in Oakland this week, with jury selection underway. Musk's central claim: OpenAI abandoned its original nonprofit mission and morphed into a for-profit entity bent on maximizing returns for Microsoft. Reporters are on the ground tracking developments, and while Altman and co-founder Greg Brockman were spotted on-site, Musk has yet to make an appearance.

Meanwhile, in a separate but related twist, OpenAI restructured its partnership with Microsoft, gaining the flexibility to serve its products across any cloud provider — not just Azure. This means OpenAI could potentially leverage Google TPUs, Amazon Tranium chips, or other infrastructure. AWS CEO Andy Jassy weighed in, calling the update "very interesting" and signaling excitement about vending OpenAI tech through AWS.

"OpenAI can now serve all of its products to customers across any cloud provider."

Translation: OpenAI just gained serious optionality in the AI infrastructure game.

🇨🇳 China Pulls the Plug on Meta's $2 Billion Manis Acquisition

In a stunning late-stage regulatory intervention, Beijing has ordered Meta to unwind its $2 billion acquisition of AI platform Manis, citing violations of China's investment rules. The decision is extraordinary — it involves two non-U.S. companies and comes after the deal had already closed and Manis software was integrated into Meta's ad manager.

According to the Financial Times, a source familiar with the matter said Beijing demanded:

Full return of funds
Re-registration of company ownership
Immediate halt to Meta's use of the Manis algorithm

If Meta fails to comply, the Chinese government could impose penalties, limit Meta's China-related business operations, or even pursue criminal charges against individuals involved.

One insider described the move as "pretty harsh," suggesting it may serve as a warning shot for future deals rather than a realistic unwinding of this one. Still, the signal is clear: any AI company founded in China remains subject to Beijing's regulatory reach — regardless of where it relocates.

"Manis did everything right. They even moved their entire business to Singapore to comply with U.S. outbound investment restrictions. Their only mistake was that they originally founded the company in China."

This raises serious questions about the future of AI entrepreneurship in China. Why would founders start companies there if success means losing control to the state? The irony: this may hurt China's AI ecosystem more than it hurts Meta.

☀️ Meta's Moonshot: Beaming Solar Power from Space

Meta announced two cutting-edge energy partnerships aimed at powering its data centers with next-gen solutions:

Space Solar with Overview Energy: Meta plans to beam up to 1 gigawatt of solar power from orbit to Earth for around-the-clock energy production. This involves using mirrors in space to redirect sunlight — a remarkably simple solution to the intermittency problem plaguing ground-based solar.
Ultra-Long Duration Storage: Meta is also deploying 1 gigawatt of battery storage with Noon Energy, featuring batteries capable of up to 100 hours of capacity — enough to weather several days of cloudy weather.

The logic: if you can deliver 24/7 sunlight, the economics of solar for data centers improve dramatically. Current data centers rely heavily on natural gas turbines because they need constant, reliable power. Solar intermittency has been a dealbreaker — until now.

"If you have this ability to bring basically 24/7 sun, can you bring a lot more solar projects online because the economics just make more sense?"

This is still early-stage tech, but if it works, expect a flood of buyers. Energy constraints are one of the biggest bottlenecks in AI infrastructure today.

🧠 The Golden Retriever Mindset: Science Says Delusion Works

A viral study made waves over the weekend with a startling finding: placebo sleep affects cognitive functioning. In other words, believing you slept well — even if you didn't — can measurably improve your performance.

Researchers assigned 164 participants to one of four conditions. Some were told they spent 28% of their sleep in REM (above average), while others were told they only hit 16.2% (below average). Here's the kicker: assigned sleep quality predicted cognitive test scores — not actual self-reported sleep quality.

The takeaway?

"Mindset can influence cognitive states in both positive and negative directions, suggesting a means of controlling one's health and cognition."

Translation: Golden Retriever mode isn't just a vibe — it's a legitimate cognitive strategy. Sleep badly? Convince yourself you're well-rested. Stressful day? Reframe it as fuel. Failed? Call it useful data.

As one observer put it: "Literally just having a delusional golden retriever mindset measurably changes outcomes and physiology."

📊 AI's Job Impact: Not What You'd Expect

If AI were poised to devastate employment, you'd expect the first signs to appear in places like India and the Philippines — home to millions of call center workers and low-cost white-collar outsourcing.

But the data tells a different story. In 2025, employment in the Philippines rose 4% — despite widespread belief that Filipino call centers are among the most AI-exposed jobs on the planet.

Meanwhile, U.S. call center employment has been in decline since 2016 — well before ChatGPT. The peak was around 900,000 workers; today it's closer to 650,000. The culprit? Outsourcing, not automation.

"It's hard to think of a more AI-exposed job than Filipino call centers. But oddly, in 2025, employment was up 4%."

The lesson: AI diffusion takes time, and second-order effects are unpredictable. Technologies don't instantly replace workers — they reshape workflows, shift geographies, and create new bottlenecks.

🤖 Token Maxing: Meet the Man Who Outsourced His Life to AI

A Semaphor profile spotlighted a founder who made a fortune selling multiple companies to Apple and now runs a voice recognition startup called Olive. His latest project? An AI assistant that pretends to be him.

The AI:

Listens to every conversation
Reads every message
Emails and schedules meetings on his behalf
Suggests parenting advice based on conversations with his three kids

When he wakes up, he consults an agenda crafted by his AI and spends the day following its directions. The AI even sets up in-person meetings with people he's never met.

"I didn't ask it to help me. I asked it to be me."

This is part of an emerging class of "token maxers" — power users plunging tens of thousands of dollars into orchestrating AI systems with unlimited tokens, massive compute, and automated checking. The goal: give the system access to every conceivable piece of relevant data and let it run.

One observer predicted: "I think this will be normal in five years."

🌍 Poland's Economic Breakout

In a stunning reversal, Poland is on track to overtake the United Kingdom in income levels. Once a communist third-world country, Poland is now growing faster than Britain, luring record numbers of UK immigrants with its restored economy and robust job market.

A British businessman running a construction firm in Poland recently advertised a marketing role. 35 of the applicants were British — all eager to relocate permanently.

"Not long ago, people weren't interested in moving here. This time, we were inundated with Brits eager to work in Poland."

The Telegraph called Poland a rising "European superpower," highlighting its economic momentum and cultural confidence.

🔮 The Inference Era: Why Model Weights May Matter Less

Anthropic researcher Noam Brown floated a provocative idea: as inference becomes more important, model weights become relatively less critical. In other words, securing weights still matters — but securing inference capacity becomes the strategic advantage.

This has huge implications for AI safety and geopolitics. The old worry: someone smuggles model weights out on a hard drive, and suddenly a rival nation has AGI. The new reality: even with the same weights, the side with more inference capacity wins.

"If I'm able to put 10 agents securing my bank account against your one agent trying to break into it, I will have 10 times the amount of solutions. I should win that battle almost all the time."

This shifts the focus from theft of models to control of compute. It also complicates the geopolitical calculus: if AGI comes soon, America wins. If it takes longer, China's massive infrastructure buildout becomes decisive.

🎨 Generative UI: A Glimpse of the Future

A viral demo this week showed ChatGPT rendering a video game-style weather map on the fly — complete with accurate neighborhood-level data. The aesthetic felt like something out of Lord of the Rings or Game of Thrones, but the functionality was real.

This is part of a broader trend toward generative, on-the-fly UI — interfaces that adapt dynamically to your query rather than relying on pre-built templates. As AI models improve, expect less "chrome" (static UI elements like top and bottom bars) and more contextual, ephemeral interfaces that feel magical.

"When you look at something and get something that perfectly sums up exactly what you're looking for on the fly — that's what feels magical."

Ben Thompson recently wrote a bullish piece on Meta's augmented reality headsets (Ray-Ban and Orion), arguing that generative UI is the killer app for AR — not pre-built apps.

📌 Quick Hits

🥤 Diet Coke Instagram is a real subculture now. Highlights include restock videos, Sigma Chad memes, and debates over 16 oz vs. 12 oz cans.
🎬 Yes Man (2008) starring Jim Carrey is the ultimate golden retriever mindset movie. Highly recommended for anyone who needs a vibe reset.
🔔 Looking for a gong? Gongsunlimited.com has you covered. Go big or go home.

That's all for this week. Stay sharp, stay skeptical, and remember: if you convince yourself you're well-rested, your brain might just believe you.