China blocks the Manus Meta acquisition deal

A summary of things I read this week

May 03, 2026

China blocked the Meta-Manus acquisition deal, citing a violation of Beijing’s investment rules. Why is it so messed up?

Benchmark Capital was the investor, and they have already distributed returns to the LPs.
Meta has already integrated Manus for its employees, and they are using it.
We already know the founders are stuck in China and can’t get out.

Financial Times@FT

Breaking news: China has blocked Meta’s $2bn acquisition of artificial intelligence platform Manus, after regulators reviewed whether the deal violated Beijing’s investment rules. ft.trib.al/JnwLniN

8:38 AM · Apr 27, 2026 · 1.19M Views

208 Replies · 676 Reposts · 2.92K Likes

China-US AI wars are getting messier. With Chinese models almost as close to US models serving inference at relatively lower cost and trained on Huawei’s chips instead of Nvidia GPUs, it’s going to get even messier. This seems more of a geopolitical move than an actual breach of law.

Sam Altman had a really interesting post on X about what the future of OS would look like. We are used to a standard typing interface, installing apps for various purposes, and using a browser to surf the web.

Now, a lot more things are moving to an audio interface (maybe eventually a brain interface), and apps are becoming on demand as creating apps is getting easier. What could be the future? Personal on-demand apps?

Sam Altman@sama

feels like a good time to seriously rethink how operating systems and user interfaces are designed (also the internet; there should be a protocol that is equally usable by people and agents)

3:46 PM · Apr 26, 2026 · 1.51M Views

1.85K Replies · 786 Reposts · 12.5K Likes

Is OpenAI working on its own phone to disrupt Apple? Well, we don’t know for sure, but it would be an interesting development.

OpenAI has also updated its partnership with Microsoft. They have been going back and forth in their love-hate relationship.

I think now they are not tied to each other, so OpenAI can use any other compute provider, and Microsoft can keep adding models from other frontier labs.

Sam Altman@sama

we have updated our partnership with microsoft. microsoft will remain our primary cloud partner, but we are now able to make our products and services available across all clouds. will continue to provide them with models and products until 2032, and a revenue share through

1:24 PM · Apr 27, 2026 · 2.08M Views

1.24K Replies · 843 Reposts · 14.2K Likes

OpenAI wrote a really fun article about how ChatGPT started mentioning goblins, gremlins, and other mythical creatures. This was due to the model incentivizing a nerdy personality. I love how our AI models can change behavior with a slight change in incentive. I know we all talk about achieving AGI, but models are what they are trained on (unless we really, really figure out how to add consciousness and human creativity to models)

I read an interesting study by Taylor Lorenz on how much of Substack is actually AI.

So, Technology posts are definitely no, as expected (if technology people are not using AI, then who will?).

Fiction is the lowest. I guess people still want to have their own original thoughts rather than AI-generated slop.

But I do believe, over time, we will see more and more of that graph shift in the red direction, so it would be interesting to keep tabs on who is writing in the green. Human originality does have something that AI cannot replicate (yet).

It’s a paid article, but you can still read this post to get some insights

If you have heard of the California vs. Texas narrative in tech, Jonathan Lacoste wrote an amazing article on X about how to think about those 2 states. He calls them complementary rather than competitive.

Deedy Das did an interesting study on what smart kids around the world do when they grow up. He studied ~18,000 International Olympiad medalists (IMO, IOI, and IPhO) over the last 25 years and traced their career paths.