Alibaba is clearly choosing growth over near-term profits. A 75% jump in capital expenditure, largely directed toward AI infrastructure, looks painful today, but the 45% growth in cloud revenue suggests the investment is beginning to generate real demand. The bigger issue is whether this spending can eventually create operating leverage. A roughly 75% decline in reported net profit shows that Alibaba’s margin structure is still under serious pressure. I would not treat Alibaba as simply a “cheap AI stock.” It is a bet on whether AI and cloud can become the next profit engine. If AI monetization accelerates, today’s margin compression could prove temporary. If growth slows, however, investors may discover that the margin floor is lower than expected. For me, Alibaba is a long-term platform

I think the biggest mistake is treating Samsung and SK hynix as the same AI trade. Samsung’s foundry problem is not technology—it is execution credibility. TSMC’s moat comes from years of stable yields, massive capacity, advanced packaging and repeat orders. Samsung can narrow the gap, but it needs flagship customers to return generation after generation. Its recent 2nm engagements are encouraging, but the real proof will be sustained volume production. SK hynix is already further along the value chain. HBM demand is translating into profits, cash flow and now aggressive shareholder returns. Its KRW40 trillion buyback and cancellation—about 3.3% of shares—shows management believes the market is undervaluing its future cash generation. So my view is simple: Samsung is the turnaround bet; S

Moderna’s 2026 transformation is bigger than a one-day stock rally. The 177% surge followed by a 23.6% pullback shows how aggressively markets are repricing its mRNA platform. The real catalyst is mRNA-4157’s Phase 3 success, potentially opening a multibillion-dollar personalized cancer-vaccine market. Yet investors should separate platform potential from valuation reality. Moderna still burns billions annually, while COVID revenue continues to decline. At $133, expectations for rapid oncology commercialization are already high. The bull case is compelling: successful cancer-vaccine approval could create a new growth engine beyond respiratory vaccines. The bear case is equally clear: manufacturing complexity, competition and delayed profitability could expose the stock to another sharp co

If I had to choose, I’d rather own the “toll collector” in the memory cycle than the companies forced to absorb higher costs. The key point is that rising memory prices are no longer an isolated semiconductor story. They are spreading downstream—from smartphones to GPUs and AI infrastructure. Xiaomi’s adjusted profit fell 42.6% year over year as higher memory costs squeezed margins, while Intel’s Arc Pro B70 prices have reportedly risen sharply in some markets. That tells me pricing power currently sits upstream. But I would not blindly chase memory stocks after their huge run. The better strategy is to own the suppliers with strong pricing power, healthy balance sheets and long-term AI demand, while avoiding companies whose margins are being compressed. In short: follow the money upstrea

Alphabet Still Wins Among these four calls, I agree most with the Alphabet upgrade. The key is that Google’s AI investment is no longer just a massive spending story. Google Cloud is accelerating, margins are improving, the backlog is huge, and its in-house TPU technology could eventually strengthen both competitiveness and profitability. That gives GOOG a compelling AI re-rating opportunity. I’m more cautious on CRWD and PLTR. Both remain excellent businesses with strong growth, but their valuations already assume years of near-perfect execution. At such multiples, even a small slowdown can trigger a sharp correction. PYPL is the wildcard. The potential M&A deal could create upside, but it is fundamentally an event-driven trade rather than a pure turnaround story. My ranking: GOOG fi

A. Yes — still bullish on DBS / OCBC / UOB I’d choose A. To me, DBS hiring more young talent is more than a recruitment story—it shows the bank is preparing for where future growth will come from. Lower interest rates may pressure net interest margins, but wealth management, AI, data, technology and fee-based businesses can increasingly offset that pressure. DBS’s strong wealth-fee growth and rising AUM are already evidence of this transition. Singapore also continues to strengthen its position as a regional wealth and financial hub, attracting capital, global institutions and high-value talent. That doesn’t mean DBS is cheap or risk-free. Valuation still matters, especially after a strong run. But for long-term investors, I remain bullish on Singapore banks, particularly DBS, OCBC and UO

I’m leaning toward A — normal profit-taking, with the AI hardware trend still intact. Tuesday’s semiconductor selloff looks more like a valuation reset than a fundamental breakdown. The Philadelphia Semiconductor Index fell about 5%, while memory and optical names such as SNDK, MU and CRDO were hit much harder than Nvidia. The key issue is macro: the 30-year Treasury yield recently reached its highest level since 2007, while Brent crude moved above $90. That combination naturally pressures high-multiple growth stocks. But AI infrastructure demand has not suddenly disappeared. Memory, storage, networking and GPU demand remain tied to massive data-center investment. So I wouldn’t call this an AI-cycle reversal yet. Instead, I’d watch whether SNDK and MU stabilize and reclaim key moving aver

My Pick: AMLX — But This Is a Catalyst Trade If I had to pick one for the next 30 days, I’d choose Amylyx Pharmaceuticals (AMLX) — not because it is the safest name, but because it has the clearest near-term binary catalyst. The stock has already exploded higher, so chasing momentum here is risky. But the real story is still ahead: Amylyx expects Phase 3 LUCIDITY results for avexitide in late August or early September. A positive readout could materially change the company’s valuation and potentially support a 2027 commercial launch. That makes AMLX fundamentally different from simply buying a beaten-down stock like Adobe or Intuit. My choice: 🔥 Momentum, with a catalyst-driven setup. The key is position sizing — this is biotech, so one clinical result can create either a breakout or a br

The 30-year Treasury yield at 5.31% is becoming an increasingly attractive entry point, but I wouldn’t rush to lock in long-duration bonds yet. The key issue is that this selloff isn’t purely about Fed policy. Persistent inflation risks, higher oil prices, massive fiscal deficits, weaker foreign Treasury demand and growing corporate debt supply are all pushing the long end higher. That makes this a classic “wait for confirmation” moment. If yields eventually stabilize around 5.5%–5.7%, long-duration bonds could offer compelling returns. But if inflation expectations continue rising, buying too early could mean sitting through another painful price decline. For now, I’d favor short-duration Treasuries and cash, while gradually preparing to extend duration if yields spike further. The best

My take: 1) SOXS, 2) HIBS, 3) TECS. The clustering of inverse ETFs is a warning that investors are increasingly hedging duration and high-beta exposure, not necessarily calling for a full market crash. For the next 30 days, I expect the 10-year yield to stay around 4.7%, with 5% possible if inflation and Treasury supply worsen. The 30-year has already hit a 19-year high, showing how serious the bond-market pressure has become. The bigger threat to AI stocks is rising yields. AI debt issuance matters, but it is ultimately another channel through which higher financing costs can pressure valuations. Morgan Stanley expects global AI-related debt issuance to approach $570 billion this year. My view: this is a valuation reset, not necessarily the end of the AI cycle.

The Q2 13F season reveals a clear message: institutional money is not abandoning AI—it is becoming more selective. Berkshire’s 83% increase in Alphabet, taking the position to roughly $38 billion, is perhaps the strongest vote of confidence in Google’s AI ecosystem.  Meanwhile, Tepper is rotating away from memory names such as Micron while adding Amazon, Meta, Alphabet and TSMC, suggesting investors may be taking profits after the semiconductor rally. The SpaceX story is equally important. Its IPO has brought massive institutional exposure into the public market, although some reported “new” positions are simply legacy private holdings becoming reportable. My takeaway: the next phase of the AI trade may shift from chips toward platforms, infrastructure, power and space. But 13Fs are

I’d pick Ivan_Gan’s view as the most actionable. Bitcoin and gold offer clear technical levels, but macro policy is still the bigger driver across asset classes. If Fed hike expectations continue to fade, liquidity-sensitive assets like QQQ and SPY could remain supported even if markets stay range-bound. That said, gold’s breakout deserves attention. A short squeeze may explain the speed of the move, but sustained strength would suggest deeper institutional demand rather than just positioning. For Bitcoin, $67K is the key confirmation level, while $57.8K remains the line bulls cannot afford to lose. Personally, I’d rather wait for the breakout than chase the middle of the range. @WallStreet_Tiger [你懂的]

U.S. stocks remain in a strong but increasingly selective bull market. The S&P 500 broke above 7,800, marking its third straight weekly gain, but sector rotation is accelerating as investors move beyond mega-cap tech into energy and industrials. The biggest warning sign is the consumer. July retail sales fell 0.6%, while sentiment weakened, raising concerns about economic momentum. At the same time, higher oil prices and geopolitical tensions could revive inflation risks. The key event this week is Powell’s Jackson Hole speech. A dovish tone could reignite the AI rally, while a hawkish message could trigger profit-taking. My view: the bull market isn’t over—it’s broadening. The next winners may come from sectors beyond technology. @W

If I had to choose one part of the AI infrastructure stack for the next six months, I’d pick memory and storage. The market often treats AI as a GPU story, but that misses what happens behind the scenes. Every new AI cluster requires huge amounts of HBM, DRAM and enterprise SSDs, while increasingly data-intensive models are creating even more storage demand. That’s why $SNDK and $MU stand out to me. Their upside is not simply tied to AI enthusiasm, but to a real hardware bottleneck: memory capacity and pricing. If hyperscalers continue spending aggressively on AI infrastructure, memory could remain one of the biggest beneficiaries. The key risk is obvious: if AI CapEx slows or new supply arrives too quickly, pricing and margins could reverse. But for the next six months, I’d rather own th

If I had to rank the next 30 days, my picks are 1) MU, 2) SK Hynix, 3) SNDK. MU is my favorite because it offers the cleanest exposure to AI-driven HBM and DRAM demand, while tight supply and strong pricing could continue supporting margins and earnings. SK Hynix comes second because HBM remains its biggest weapon, with hyperscaler AI spending still expanding. The main risk is valuation and Korea-market volatility. SNDK has the strongest momentum, but after its massive YTD rally, chasing the stock becomes increasingly risky. I would rather wait for a pullback than blindly follow the breakout. For the next leg, I believe pure-play AI memory demand is the strongest theme. AI servers need dramatically more memory, while meaningful new capacity takes years to build. The shortage may eventuall

The Real AI Risk Isn’t Spending — It’s Monetization I’d pick A: AI revenue takes too long to materialize. The $3 trillion commitment shows that AI demand is being locked in, but spending does not automatically create returns. Hyperscalers are committing huge amounts to chips, data centers, power and leases before AI revenue fully catches up. Hardware suppliers may benefit first, but eventually investors will ask whether AI revenue can cover depreciation, interest, rent and electricity. If monetization disappoints, CapEx will eventually slow, creating a second wave of pressure across semiconductors, memory and infrastructure stocks. In my view, the biggest AI bubble risk isn’t overspending itself—it’s spending faster than profits can catch up.

If I had to choose one AI semiconductor stock for the next six months, my pick would be $Broadcom (AVGO). Nvidia remains the GPU king, but I think the next opportunity is shifting toward the infrastructure behind AI. Broadcom is benefiting from two powerful trends: custom AI accelerators and high-speed networking. Hyperscalers such as Google, Meta, Microsoft and Amazon are spending heavily on AI, and many are developing custom chips to reduce their reliance on Nvidia. Broadcom is positioned directly in that transition, while its networking business benefits as AI clusters become larger and more complex. What makes AVGO especially attractive is the combination of high margins, recurring demand and multiple AI growth engines. The biggest risk is an eventual slowdown in hyperscaler capex. Bu

If I had to pick one stock from the list, I’d choose Charles Schwab (SCHW). What makes Schwab interesting is that this is more than a traditional bank trade. It benefits from rising client assets, stronger trading activity and growing banking revenue. Q2 revenue reached a record $7.1B, up 21% YoY, while adjusted EPS jumped 42%. I also like BNY as a second choice because its custody and asset-servicing business gives it a different growth engine from traditional lenders. That said, I wouldn’t chase every bank at all-time highs. Earnings momentum is real, but valuations are also moving higher. A pullback could offer a much better risk/reward entry. My vote: A — U.S. Banks & Brokers. My pick: SCHW. 30-day dark horse: BNY. The bank rally still has fuel, but from here, stock selection matt