Sabitlenmiş Tweet
Mars
108 posts
Mars
@MarsLT
A man looking at the world through a keyhole and forever trying to widen it.
Katılım Kasım 2010
267 Takip Edilen28 Takipçiler
@EquityBrian Chinese consumers are extremely price sensitive and the competition is fierce. I'm long BABA but wouldn't read too much into this, it's a trade off between market share/traffic vs. cost. BABA can make it profitable today but then they will lose it all back to Meituan.
English
Surprised more people aren’t talking about what management said on Quick Commerce during the $BABA call.
Management went from targeting positive unit economics by FY29 to FY27. That is a pretty meaningful two-year pull forward…
Still have to prove it, obviously, but Quick Commerce losses have been the easy bear talking point for a while.
Brian Coughlin@EquityBrian
Some thoughts on $BABA and why this quarter felt like a turning point… coughlincap.com/p/alibaba-baba…
English
@cz_binance @TheFP CZ, can’t think of too many business person that went to prison and came back with reputation intact and even stronger! Wear it like badge of honour 👊
English
Good interview with The Free Press @TheFP.
Sharp questions, often started with assumptions that are different from mine.
Title: Former Binance CEO Speaks: From Crypto's Richest Man to Prison
I'd add "And Back" to the title if I could.
youtube.com/watch?v=wsqq-d…
YouTube
English
@TJ_Research 而且现在的token都是以文字、code为基础，未来如果是physical ai为主，ar/vr，就好比现在的netflix 4k streaming在早期的互联网时代都是天方夜谭
中文
完全同意这个比喻，更可怕的可能是在TOKEN那么贵的早期时候，软件工程师的人力就已经遭到了破坏，难以想象TOKEN便宜到白菜价时可以破坏的市场。这是一股非常强的通缩力量
Baby Mango@StefanShi651815
@TJ_Research 现在的人 token还处于非常早期的阶段，给我的感觉就是我们处于15-20年前中国的移动互联网早期时代，那时候流量一个月500k,1.2m都要花钱买，用流量都要卡着用。但是看看现在，一切应用都建立在几乎无限的流量上面。再也没有人用app顾及流量不够用，消费者也愿意为这个流量买单。token就是过去的流量
中文
@bgurley @rohanpaul_ai Because competition is fierce and capital is still abundant.
That said, some of these demand are a result of this and non-productive.
This is all a big bet on achieving some sort of AGI and fundamentally improve productivity to a different level.
English
@rohanpaul_ai Most companies that are supply constrained raise price. This would balance everything. Yet that isn’t happening. Why?
English
"Anthropic is making great money. OpenAI is making great money. If they could have twice as much compute, the revenues would go up 4 times as much. These guys are so compute constrained, and the demand is so incredibly great."
~ Jensen Huang on CNBC
English
@DavidLe76335983 @BrianTycangco This is likely saying that taking lottery tickets away from low income people takes away their chances of becoming millionaires. Credit cards are not a good will service to them, it’s more like lottery tickets, a tax.
English
They will just cancel credit cards for low credit score public leaving millions without credit cards. These are the segments that use credit card to buy food and necessities as their paycheck can only cover last month credit card payment. They will now unable to survive in the system
English
Credit card companies will have no viable business model if they comply by Jan 20 on interest rate cap. You simply cannot justify the default risk at 10% interest. $V $MA $JPM $BAC
*Walter Bloomberg@DeItaone
TRUMP: CREDIT CARD COMPANIES WILL BE IN VIOLATION OF LAW IF THEY DON’T COMPLY BY JAN 20 ON INTEREST RATE CAP
English
@BrianTycangco Top 5 of this list is pretty much my top china individual stock holdings minus sungrow lol
English
Goldman Sachs $GS has $BABA forward PE at 17X with a 28% EPS CAGR.
English
@zeeshanp_ Does this also proves Nvidia's moat in the space? Upgrading from NVDA to NVDA is already hard enough, let alone transitioning to a competitor product/ecosystem?
English
People underestimate the difficulty in bringing up the next generation of AI hardware. Blackwell was announced in 2024, but only near the last quarter of 2025 some frontier labs began using Blackwell for large-scale training.
There’s an ungodly number of problems that come with new hardware on the software side. For example, with the Grace Blackwell (i.e. GB200) chips, there were major incompatibility issues with deep learning frameworks like PyTorch and the Grace CPU in the early days. It was nearly impossible to even use Blackwell in 2024 for training without major software modifications. On top of that, a whole new set of kernels and algorithms must be written for maximum hardware utilization and low-precision FP4 training. Fundamental breakthroughs in deep learning science have to be made to guarantee no training instabilities on a fleet of hundreds of thousands of chips. One loss spike or a few silent hardware failures can cost a flagship training run. And these are just the software issues. There’s a whole set of different problems in the world of atoms to build these massive supercomputers.
We’re only in the early innings for the AI infrastructure buildout. There will be many more infrastructure cycles in the future that will present new challenges. There will be a constant need for more compute, and soon we’ll have AI systems capable of automating the entire process of designing and bringing up new hardware end-to-end.
Elon Musk@elonmusk
@SawyerMerritt It will take another 9 months or so before the hardware is operational at scale and the software works well
English
@MyLordBebo It's kinda obvious that the guy wearing uniform is waiting for the escalator to be clear of people to re-start it, it is the safety protocol. It has nothing to do with the "wheel-chair" in this case. Don't spread this fake narrative plz.
English
@vayzenburger @pmarca Question is, if we double the GDP growth, would that double the real growth? As long as that’s true, or at least strongly correlated, then the argument stands.
English
💯 The difference between 4% growth and 2% growth is doubling the economy every 18 years vs every 36 years. For the material welfare and living standards of the citizenry, literally nothing matters more than that.
David Sacks@DavidSacks
GDP growth is currently an outstanding 3.9%, and AI is 40% of that. It’s easy for politicians to posture and grandstand by beating up on tech. But the real question is whether they want 4% growth rates or 2% growth rates.
English
@APompliano Fed cuts 25BP - "Big mistake, should've done 50";
GDP revise up - "Recession worries are ridiculous";
See the problem here?
English
Q2 GDP was 3.8% and people are worried about a recession.
Ridiculous.
English
@charliebilello Not entirely accurate. 74% of the time the price drops lower, but only a fraction of the people waited would pull the trigger at the lower price🤷
English
Waiting for a Pullback to invest idle cash at lower prices?
You're not alone.
Historically, 74% of the time you would've gotten that lower price at some point in the future, but 26% of the time the market kept on running and never looked back.
Video: youtube.com/watch?v=J_Df0W…
YouTube
English
