Hhhhhhhhhhkkkkkkkkkk

The market is on pins and needles ,Always;HK finance market and something else.

[Vivi @vivilinsv]

今天和一位长期居住在美国、但仍持中国国籍的投资人朋友吃饭，聊到最近中国税务部门对境外资产和收入的追缴。

他本人深受其扰：一年之内，已经补了三次税。

最近一次，税务人员直接打到他的手机。他起初还以为遇到了诈骗电话，随后试图解释：“该补的税，我不是都已经补完了吗？”

对方只反问了一句：“你真的有必要这样吗？”

紧接着，对方把他在香港和新加坡名下账户里的数字，一个个报了出来，甚至精确到了小数点后。

那一刻，他才真正意识到：在CRS和跨境金融账户信息交换之下，所谓的“境外账户”，对税务机关而言，就是如来神掌里的孙悟空，尽在掌握中。

而就在今天，中国财政部和税务总局又出台了针对离岸信托的个人所得税新规：中国税收居民将股票、房产等资产装入离岸信托时，资产增值部分可能需要按照财产转让所得纳税；信托及其控制的境外实体在存续期间产生的收益，即使没有实际分配给个人，也要按年申报纳税。

新规还要求补缴部分自2023年以来与离岸信托有关的历史税款，并提供90天的申报窗口。甚至已经取得外国国籍或海外永久居留权的人，如果主要经济利益仍然来源于中国境内，也可能继续被认定为中国税收居民。

这意味着，过去被许多高净值人士视为财富隔离、代际传承和税务安排“最后一道防线”的离岸信托，也开始被全面穿透。

真正让有产阶层心惊胆战的，或许已经不只是税率本身，而是规则可以追溯多远、身份如何被重新认定，以及政府对个人全球资产图谱究竟掌握到了什么程度。

过去人们问的是：如何把资产放到海外？

现在真正的问题变成了：资产到了海外，是否真的离开了中国的税务体系？

[段新星 Duan XinXing @xinxingduan]

看了下，中国具身智能类产品销量第一的，也是之前一位TEDxNanjing 的嘉宾, 刘维超的蔚蓝.
TEDxNanjing 2014 的讲者

[Michael Anti @mranti]

AI推翻了雅可比猜想，今年的菲尔兹奖可能是最后一届纯人类思考的数学奖了。在这个时刻，我再次推荐大家阅读Ted Chiang的科幻小说：未来科研完全由AI进行，人类需要特殊脑机接口才能读懂科研进展的科普性文章。

[goldenlabubuwatch @pandawatch88]

Bit of Hong Kong market copium for my friend @aceboy8844 from Laura at MS.

[Guohao Li 🐫 @guohao_li]

why can the china labs build glm-5.2, kimi k3, and many more to come? it is because of the openness. not just the open weights but the whole ecosystem. most of the work done in the china labs is carried by interns. i met brilliant undergrad and graduate interns who deeply understand the model training details, and they are 100x more open to share. that means the talent that knows how to train llms in china is 100x greater in number than the talent in the us, and it is growing

in contrast, the us ai ecosystem is too closed. frontier labs do not hire interns. i know brilliant phd students at stanford, berkeley, and so on. they struggle to get an internship and the compute to train a properly sized model. most of the secret recipes are locked away by a very small group of privileged researchers

it is not about china or the us. it is about open and closed science. the fact is that every average cs student can learn how to train an llm. they just need the opportunity. labs should be more open and hire more interns, like how deepmind and fair did in the pre-llm era

[Russ Salakhutdinov @rsalakhu]

I’ve been asked several times whether Zhilin Yang, the founder of @Kimi_Moonshot was my PhD student. The answer is yes and he is absolutely brilliant.

But I’ve been incredibly fortunate to work with so many outstanding PhD students over the years. So I thought I’d brag a little about them and their career paths (of course there are also many MSc and undergraduate students, sorry if I missed anyone):

Founders / Founding Team Members

Devendra Chaplot @dchaplot PhD, Founding Member Thinking Machines / Mistral
Zhilin Yang PhD, Founder & CEO, Moonshot AI
Jimmy Ba @jimmybajimmyba MSc/PhD, Co-founder xAI
Hubert Tsai PhD, Co-founder Spuree, Apple
Nitish Srivastava @nitishsr PhD, Co-founder Perceptual Machines; Co-founder Vayu Robotics
Charlie Tang PhD, Co-founder Perceptual Machines, DE Shaw.

Professors
Paul Liang @pliang279 PhD, MIT
Ben Eysenbach @ben_eysenbach PhD, Princeton University
Ruosong Wang @RuosongW PhD, Peking University
Bhuwan Dhingra @bhuwandhingra PhD, Duke University
Roger Grosse @RogerGrosse Postdoc, University of Toronto
Alexander Schwing Postdoc, UIUC

Research Scientists
Shuyan Zhou @shuyanzh36, Postdoc, Meta Superintelligence Lab
Tiffany Min @SoYeonTiffMin PhD, Microsoft AI
Murtaza Dalal @mihdalal PhD, Tesla AI
Minji Yoon @MinjiYoon90 , PhD, Microsoft AI
Shrimai Prabhumoye PhD, NVIDIA AI, Mistral
Haitian Sun @sun_haitian PhD, Google DeepMind
Emilio Parisotto PhD, Google DeepMind
Lisa Lee PhD @rl_agent, Google DeepMind
Manzil Zaheer @ManzilZaheer PhD, Google DeepMind
Jamie Kiros PhD, Google Brain, OpenAI
Yuri Burda PhD, OpenAI, Anthropic
Cody Severinski PhD, Amazon

[Rui Ma @ruima]

Wow cofounder of @tryramp @eglyman speaks Mandarin!
This seems to be some debate show / contest thing for college students from 2011 that aired on one of the CCTV stations

[Fenng @Fenng]

香港中文大學、香港城市大學和香港大學三家高等教育院校的圖書館與出版社聯合發起，香港首個開放取用（Open access）圖書計劃。

推荐一下

https://t.co/BuJUgctyPq

[Bloomberg @business]

Wall Street banks were largely absent from a $5.8 billion fundraising spree by Chinese AI stars in Hong Kong this week https://t.co/YnhDs4aoNI

[Du Jun @DujunX]

关于李博杰提到的投资事宜，我补充几点事实：

1、2024 年 2 月 26 日，ABCDE Capital 向李博杰创办的 Metagent 支付首笔 50 万美元投资款。双方当时明确约定：先支付 50 万美元，再根据约定安排后续 100 万美元投资。这符合基金正常的投资运作方式。

2、 早期项目投资并不是一次性、无条件支付全部资金。后续资金安排通常需要结合项目进展、团队稳定性、信息披露、财务情况以及投资协议约定的相关条件综合判断。作为基金管理人，ABCDE 需要对 LP 负责，也必须确保每一笔后续资金投入都有充分依据。

3、2024 年 6 月，距离首笔投资到账约 4 个月，团队向投资人展示了一个极其潦草（抱歉，有点主观）的Demo，结合项目公开信息（项目主要社媒账户@metagent_xyz 整个生命周期一共发布了 14 条推文，项目进度以及质量可以大家自行研究），团队沟通情况及其他投资机构的反馈，我们认为项目整体进展未达到预期，对团队后续执行情况也产生了进一步关注。

4、2024 年 7 月开始，李博杰基本不再回应投资人群内的问题，由 Metagent 合伙人庄思源负责沟通。随后，李博杰注销了 Telegram 账号并退出相关群聊。

5、2024 年 12 月 5 日，李博杰通过邮件告知其已离开 Metagent，并表示公司资金已经不足。但自 2024 年 2 月首笔投资到账至 2024 年 12 月期间，以李博杰为核心的 Metagent 团队始终未向 ABCDE 提供完整、详细的财务报表。此后，我们继续要求 Metagent 提供完整、明确的财务报表及经营信息，但李博杰及庄思源始终未能提供充分、清晰的回应。

写到哪儿算哪儿…

[goldenlabubuwatch @pandawatch88]

Awkward moment last month at Vivatech in Paris right after Claude Mythos got banned for foreigners.

I was watching the video this morning. French interviewer asks Alibaba Joe Tsai (paraphrasing, full text of the question/answer in the screenshot):

-"we are disappointed with the US. But how can we trust the Chinese won't do the same to us?"

and Joe answers

-"You can't. But right now you have all the eggs in one basket, and we are offering a second basket so you can spread them"

Silence follows, then half of the people in the audience start clapping, the other half look at the ceiling like "man we are so fkd".

The battle of Europe. The saying goes "in order to convince, appeal to interest not to reason". How do US and China stack?

-The US side represented by its labs and a16z and palantir etc are taking the christian evangelist route "our models have our values, and these are the values that do best for humanity. Others dont. So best for everyone if we run the world on our models." They spice this up with a bit of rapport sales technique calling potential paying customers "our allies" and a bit of FUD "backdoors, spying".

-the China side sales technique is basically giving a "90% as good" product away, and hope to make money from those that decide to still hire them as chefs even though you have the recipe already. To this now the "spread the eggs" argument is added. No talk of "values" in the model, no talk of "backdoors" in others. Missing a bit of charm. "Very cheap, very good, you try you try".

In the meantime, also check the third pic for the state of datacenters in Europe, from a recent MS report.

[The Economist @TheEconomist]

IVF is failing most women. But new research holds out hope for the future https://t.co/ZIB88FZaB5

[The Economist @TheEconomist]

America must acknowledge that it will share its third century with China, writes Rush Doshi in a guest essay. Register for free to learn why he argues its strategy must be built on two pillars https://t.co/dfiibnPMIy

[Jonathan Cheng @JChengWSJ]

“In the early 2000s its firms invested in Latin America to secure natural resources; in the 2010s they built lots of infrastructure to help export excess steel. Now China is seeking places where it can dominate in high-tech fields.”
https://t.co/OJQNWxCxdr

[Kevin S. Xu @kevinsxu]

The Meituan AI team has a fascinating back story. It didn't come out of nowhere.

In early 2023, soon after ChatGPT launched, one of the cofounders of Meituan, Wang Huiwen, launched his own lab to build "China's OpenAI". He threw in $50 mil of his own money, attracted investments from other Meituan co-founders including CEO Wang Xing, and built a decent team. You can say he started the wave of Chinese AI labs looking to compete with OpenAI, Anthropic, that now includes DeepSeek, Moonshot, Z, etc. The lab was called "Light Year Beyond".

It didn't last long. By June 2023, Wang was suffering from mental health issues, the pressure was too much, and Meituan bailed him out and acquired it.

The acquired team is now known as GN06, focused on AI native applications. It is not the foundation model team that is building Longcat, but the two work together, and will likely do more to feed AI capabilities into Meituan's vast user base and use cases as the model layer matures.

The focus on this Longcat 2.0 will likely all be on whether and on what domestic processors it was trained on. But the team's history is also quite fascinating, certainly did not come from no where.

[Mike Bird @Birdyword]

Sort of amazed that it's only twice

[Sophia Xing @xingsthatmatter]

I’m a founder in SF and recently spent time in China.

A few observations from going back and forth between the two worlds.

Not about geopolitics. Just daily life, consumer products, work culture, and what “modern infrastructure” feels like on the ground.

A thread 🧵

[Mike Forsythe 傅才德 @PekingMike]

BREAKING: From me. Chinese billionaire and Bannon Associate Guo Wengui sentenced to 30 years for fraud. I have been writing about Guo for 9 years now. But this may be the last story about him I ever write.

[nico @nicochristie]

Tianhang is finally on twitter

He was formerly a founding team member and head of RL at Qwen, and leads model training here at FRL / shortcut

He will be posting more on twitter now

Probably the most cracked person i've ever met. Excited for you to see his work soon!

[Joe Weisenthal @TheStalwart]

NEW ODD LOTS:

@tracyalloway and I talk to Baidu CFO Henry He

We discuss:

- Going up against Waymo in driverless cars
- The company's full-stack strategy, from chips to datacenters to models to applications:
- Former Baidu engineer Dario Amodei

https://t.co/cBXMZbOAoe