Stanitut @stanitut

China is now dominating the global auto market: China's vehicle exports jumped +68.7% YoY in May, to ~930,000. This is almost +1,100% above levels seen in May 2019. This comes as new EV exports surged +112.6% YoY, to 424,000, accounting for ~46% of the total. As a result, China exported a record 8.6 million vehicles in the 12 months ending May. By comparison, Japan exported just 4.2 million, or -51% fewer, during the same period. To put this into perspective, China exported just 1.0 million vehicles in 2019, while Japan exported 4.8 million, or +380% more. China is now the undisputed leader of the global car market.

we. are. so. early. $spcx

keyboard_arrow_left Previous keyboard_arrow_right Next

>60% of tech is down 20% or more from respective 52-week highs.. here's the latest drawdown heat map @finviz_com

The latest hyperscaler forecast chart from Goldman

It is time for Operation Paperclip 2: repatriate the jewels of mittlestand to America. wsj.com/economy/china-…

SemiAnalysis puts cumulative AI capex (IT plus datacenter) at roughly $11.1 trillion from 2024 to 2029, with annual capex exceeding $2 trillion by 2028, and expects outstanding AI-related debt to reach about $7 trillion by 2029, second only to the US mortgage market

DB: AI capex crowding out non-AI investment

How quickly are the members of the DRAM cartel building in the next two years? Samsung is adding 25% of its 2026 capacity, SK Hynix is adding 39%, and Micron is adding 66%. CXMT is adding 89%. Turns out that the smaller you are, the easier it is to build. This means that Micron gains three points of market share at the expense of Samsung (2%) and SK Hynix (1%). But it is still nowhere near enough to catch up in absolute terms, with Samsung still maintaining 30% share and SK Hynix 25%, while Micron only climbs to 20%. Also, the CXMT number is misleading because it is wafer starts and thus does not account for the fact that they are a node behind and cannot produce as many bits per wafer. Their bit share is closer to 12%.

keyboard_arrow_left Previous keyboard_arrow_right Next

@StadtMuenchen Der Nahverkehr in München ist wirklich eine Katastrophe - keine Bahn kommt pünktlich und die Fahrzeiten sind im Durchschnitt deutlich länger. Wenn die gruenen vom Auto weg wollen muss man eine funktionierende an Alternative anbieten.

McKinsey have issued their verdict on software industry : Some may not survive. Ever since the advent of AI, billion dollar question has been what happens to the software $IGV names that ruled the roost. Mckinsey have released their framework. Here's the winner/loser framework for agentic AI era, straight from the report. Bookmark this. The core shift: Agents replace human users. Fewer users = fewer seats = per-seat monetization breaks. - 63% of software leaders expect their business model to fundamentally change within 3-5 years. - Competitive advantage moves from features to proprietary data. Features get copied. Data doesn't. That gives you a 3-question screen for every software name you hold: > Does it own the system of record + the data? If agents must go through it to act, it keeps pricing power. If an agent can hit an API and bypass the app, the UX layer gets commoditized to zero. > Has it already moved to consumption pricing? McKinsey names $CRM, Zendesk, Intercom, and LexisNexis ($RELX) as already monetizing AI this way, often at higher per-customer revenue than their old seat-based SKUs. Consumption-priced software companies doubled 2015-2024. > Does it have domain expertise agents can't train around? The "agents as experts" model (legal, healthcare, financial services) is where McKinsey sees premium pricing surviving. My screen on that framework: Positioned well: - $MSFT (data + consumption via Copilot/Azure) - $CRM (Agentforce is exactly the archetype McKinsey describes, execution TBD) - $NOW (owns the workflow layer agents have to route through, and it's already selling agent orchestration as the product, not a feature) - The sleeper passes: $TRI , $VRSK , $VEEV , all domain data agents can't train around Positioned badly: the $5-20B horizontal per-seat layer. Project tools, e-signature, communication point solutions. Seat pricing + no proprietary data + workflows agents absorb first. That's where the "some may not survive" line lands. > The margin stakes: AI-centric firms already report 20-40% lower opex and +12-14pp EBITDA margins. The gap between the two lists compounds every quarter. Which names are positioned badly for the agent era: $ZM , $MNDY or something else? And which one is the value trap? Repost this. Most portfolios are still scared of $IGV meltdown. Your followers holding software names will thank you.

keyboard_arrow_left Previous keyboard_arrow_right Next

This thread is awesome but very technical for a generalist like me. Here's a cheat sheet from GPT as you go through as I needed one... 🤣

keyboard_arrow_left Previous keyboard_arrow_right Next

Big Boss @0xBADB01E

3 days ago

I agree with this post whole heartedly but I’d push it even further. The interconnect IS the binding constraint for AI even more so than memory. If we want faster inference & training with better economics we are best served by designing our interconnect first and then working

48 83 1K 342K 1K

Tech is the focus of YTD alpha around the world:

Nomura just dropped a blockbuster forecast for the global memory market. It sees total memory revenue exploding from $181 billion in 2024 to $2.625 trillion by 2030. That is roughly 14.5x growth in six years. The real driver is data centers. Nomura expects them to account for more than half of all memory demand soon. AI-specific needs are ramping hard, but even standard cloud, training, and inference workloads are pushing massive growth. This goes beyond a normal chip cycle. Memory is becoming the lifeblood of the entire AI infrastructure buildout. Every GPU cluster, hyperscale facility, and edge deployment needs more of it. As models get bigger and real-time inference spreads, the demand for high-bandwidth, power-efficient DRAM and NAND looks insatiable. For the supply chain, this is a multi-year secular tailwind. Samsung, SK Hynix, Micron, equipment makers, and materials suppliers all stand to gain. But it also spotlights coming bottlenecks in energy, advanced packaging, and next-gen architectures like HBM. The shift is clear. In 2024 data centers took about one-third of memory demand. By 2030 that share tops 50 percent, with AI driving a big portion. The industry is moving from consumer/PC-led cycles to enterprise/AI-led ones. If you are tracking the AI boom, memory infrastructure deserves close attention. The data center memory supercycle is just getting started. Which memory names or themes are you watching most closely right now?

RedHotChillyPeppers @joejo2038

2 days ago

노무라(Nomura)가 추정한 글로벌 메모리 시장(Total Memory Revenue) 전망 2024년 1,810억 달러에서 2030년에는 2조6,250억 달러 까지 성장하는 것으로 전망 이는 약 14.5배 규모 향후 메모리 시장의 절반 이상은 데이터센터가 소비하는 메모리가 될 것 [도PB의 생존투자]

3 34 217 72K 62

Goldman Sachs dropped the most precise map of where $7.6 trillion is going over the next five years and it tells you exactly which companies are standing in the middle of an unavoidable flood of capital (Save this). The numbers are worth understanding precisely before talking about who benefits. Goldman's baseline projects $765 billion in AI capital expenditure in 2026 alone, growing to $1.6 trillion annually by 2031. Over the full 2026 to 2031 period, cumulative spend breaks down to $5.1 trillion in compute, $2.1 trillion in data centers and $358 billion in power. Nvidia is assumed to command 75% of all compute spend throughout the period, using the Rubin VR200 chip at $80,500 per GPU as the baseline. The data center specification charts reveal how dramatically physical requirements are escalating. A standard cloud data center runs 5–15 kW per rack while a transitional Blackwell era AI data center runs 130–200 kW per rack. The AI factory of the future, running Rubin and Feynman silicon operates at 500+ kW per rack, at greater than 1 gigawatt scale, with liquid cooling only. Traditional hyperscale data centers cost roughly $10 million per megawatt to build while the next generation AI data centers are being discussed at $15 to $20 million per megawatt. Goldman identifies silicon useful life as the single biggest swing factor in the entire model. At a 3-year replacement cycle, cumulative compute depreciation hits $3.99 trillion and at 7 years, it drops to $2.23 trillion, a $1.76 trillion difference on one assumption alone. Power is only $358 billion of the total, but Goldman is explicit, it is the only component that can prevent the other 95% of the stack from deploying. Now here are the companies standing directly in the path of each layer of this capital. Nvidia is still the most concentrated bet on the compute layer. At 75% of $5.1 trillion in compute spend over six years, that is $3.8 trillion in cumulative revenue flowing through one company's products. The 75% gross margin on data center GPUs is the reason every hyperscaler is trying to build custom silicon to escape it while simultaneously continuing to buy Nvidia because nothing else performs at the same level. Vertiv is the direct infrastructure play on the data center upgrade cycle. Every rack going from 40 kW to 500+ kW needs liquid cooling systems, power distribution, and thermal management infrastructure that simply did not exist at prior density levels. Vertiv just deepened its liquid cooling capabilities through a strategic acquisition and was named a key partner on Hut 8's large AI-focused Texas data center campus. The liquid cooling market is growing from $5.5 billion today to $15.75 billion by 2030, and Vertiv is the dominant provider in that market. Vistra is the power thesis in its most direct form. The $358 billion power segment is the critical path for the entire $7.6 trillion, and Vistra has spent the last 18 months locking up that critical path through long-term nuclear power purchase agreements. Vistra secured a 20 year agreement with Meta for over 2,600 MW of nuclear energy, plus a separate deal with AWS from its Comanche nuclear facility. Goldman Sachs and Jefferies both upgraded the stock after the Meta deal was announced. The architecture of this trade is simple. Goldman's model is not a prediction of whether AI spending happens but rather a model of the minimum physical capital required to deploy infrastructure that has already been contracted, already been announced, and is already under construction. The compute layer requires the chips, data center layer requires cooling and power infrastructure and the power layer requires nuclear at scale on multi-decade contracts. All three layers are being funded simultaneously, and all three have identifiable public companies sitting directly in the path of the capital. Come join Milk Road Pro and get our full $7.6 trillion infrastructure breakdown which names across compute, data centers, and power we're currently positioned in and our full thesis on the AI trade. Link below!

keyboard_arrow_left Previous keyboard_arrow_right Next

Recent AI compute pricing deals (chart via claude, underlying numbers from RBC table - attached) What do you notice... is that price moving up or down?

keyboard_arrow_left Previous keyboard_arrow_right Next

China’s gross investments total $5.9T annually – ~30% of its GDP. The US invests $5.1T. The EU invests $3.1T. In net terms, China attracts 3x–5x more productive capital each year than the US and Europe combined. Where productive investment is happening: mck.co/competitiveness

$JPM on Meta Compute

The AI infrastructure buildout is entering a new phase: US tech companies are committing to spend a record $850 billion on data center leases over the next several years. This marks a +$570 billion YoY increase, or +204%, and +$200 billion QoQ increase, or +31%. Meta, $META, added the most in Q1 2026, committing +$79 billion in new leases, a +76% QoQ increase, bringing its total to ~$183 billion. At the same time, Microsoft, $MSFT, added +$41 billion, a +26% QoQ increase, bringing its total to ~$197 billion. Oracle leads with the largest total commitments at ~$250 billion, having already secured many of the key sites needed to fulfill its contract with OpenAI. Tech companies are doubling down on AI.

Morgan Stanley: Robotics > Humanoid robots represent the largest embodied AI opportunity, with a projected global total addressable market (TAM) of US$7.5 trillion by 2050, accumulating an estimated global stock of 1 billion robots. > While the industry is in its infancy, Morgan Stanley forecasts that small-scale commercialization and pilot deployments will begin this year (2026). > The global humanoid component market is projected to reach a TAM of US$780 billion by 2040. Key hardware components are expected to see massive demand multipliers by 2050 compared to 2025 levels, led by Edge Compute (1,904x), Bearings (370x), Motors (170x), and Reducers (157x). > Shipment Forecasts: China’s humanoid shipments are expected to grow at an 85% CAGR between 2025 and 2030, rising from 12k units to 262k units. > Cost Advantage: China possesses a significant Bill of Materials (BoM) cost advantage over non-China supply chains. Survey data reveals that potential adopters view a price range of US$14k–$28k (Rmb100k–199k) per unit as the sweet spot for broad adoption (with products like the Unitree G1 already selling at US$16k). > Government Support: Government guidance funds for robotics exceed Rmb187 billion (~US$27 billion). Local governments are heavily subsidizing the ecosystem and actively creating early orders via "data collection centers". > In 2025, China's humanoid shipment mix was heavily dominated by R&D/Education (42%), followed by Data Collection (19%), Interaction (19%), and Entertainment (16%).

keyboard_arrow_left Previous keyboard_arrow_right Next

P Equity Research 📰 @pequityresearch

5 days ago

x.com/i/article/2072…

10 9 85 115K 154

GS: Countries are at different points on the Demographic Wave for peak consumers (ages 35-55) with India pre peak, US early peak, China mid peak, and Japan post peak