Chart #12 hits closest to home for me. The asymmetry of losses is simple mathematics, but it has major implications for how I invest. A 50% fall requires a 100% recovery just to get back to where you started. That also makes me slightly cautious about the message in charts #1 and #2. Markets have historically rewarded patience, but a 20-year positive index return does not mean every individual stock eventually recovers. Some companies permanently destroy capital or disappear altogether. For me, the strongest lesson across these charts is therefore not simply “buy and hold”. It is buy quality, diversify, avoid excessive leverage and give compounding enough time to work. Chart #11 reinforces this particularly well. Growth attracts attention, but sustainable ROIC and the ability to reinvest

My pick: A. Las Vegas Sands ($LVS). Both integrated resorts should benefit if Singapore continues strengthening its position as a premium tourism and entertainment hub, but MBS has the stronger global brand and luxury positioning. The US$8B expansion adds another growth leg through a new luxury hotel, arena and expanded MICE capacity. I particularly like the arena strategy because major concerts and events can drive spending across rooms, restaurants, retail and gaming rather than relying on casino growth alone. Genting Singapore may offer greater upside if RWS 2.0 executes well, but if I had to choose just one for 2031, I would favour LVS for the stronger competitive moat and ability to monetise high-value tourism.

MBS’s US$8B expansion looks less like a “fourth tower” and more like a long-term bet on Singapore becoming Asia’s premium tourism and entertainment hub. The 570-suite luxury hotel matters, but I think the 15,000-seat arena could be the bigger catalyst. Major concerts bring a multiplier effect: flights → hotels → F&B → retail → local spending. Add more MICE capacity and gaming, and LVS is building an ecosystem rather than simply adding rooms. For $LVS, the opportunity is backed by an already highly profitable MBS. Management believes the expanded property can exceed its return thresholds. But US$8B is enormous, with construction, financing and execution risks, while opening is only expected around 2031. For Singapore, I see the bigger winner as the tourism economy itself. MBS and RWS u

B: SNOW. It is the breakout I would most want to own after a pullback because the rally has fundamental support, not just momentum. Strong earnings, accelerating product revenue and growing AI adoption give SNOW a clearer path for earnings to catch up with expectations. BE would be my second choice, but after the breakout and S&P 500 catalyst, I would wait for the excitement to cool before entering. I would be cautious with HOOD/COIN/MSTR because much of their near-term upside depends on Bitcoin holding above $80k. MSTR adds another layer of leverage on top of that. I would avoid chasing TSLA. Cybercab is a major milestone, but the valuation already assumes enormous future scale while deployment remains tiny and regulatory risk is unresolved. My ranking: SNOW > BE > crypto baske

I would take Circle for the longer-term thesis, Coinbase for the cleaner cyclical trade, and Strategy only if I specifically wanted amplified Bitcoin exposure. Strategy gives the biggest torque when BTC rallies, but that cuts both ways. At 845,050 BTC, the thesis is increasingly Bitcoin plus financing mechanics rather than an independent operating business. Coinbase is different: it benefits from activity. If $80k brings trading volumes, institutional flows and broader crypto participation back, it does not need Bitcoin itself to double. Circle is the most interesting structurally. USDC already has scale, while the September 16 Arc launch adds another layer to the moat through institutional infrastructure. Visa, Mastercard and BlackRock involvement matters, but the 21-bank stablecoin conso

I would wait on the regulator before paying 350x for the scaling story. Cybercab carrying real passengers without a wheel or pedals is unquestionably a milestone, but 45 vehicles prove technical viability, not economics. Waymo currently has the stronger evidence: autonomous rides across 14 cities and expansion into three more markets at once. Tesla’s upside is potentially much larger if its camera-only system and owner-operated fleet can scale without Waymo’s infrastructure intensity. But that is precisely what the valuation already assumes. For me, the next Tesla KPI is not another demo or even miles driven. It is how quickly Cybercab can scale from dozens to thousands while maintaining safety and clearing NHTSA scrutiny. Until then, Waymo has the deployment lead while Tesla has the scala

I would trust the hold, but I am not ready to call peak rates yet. Waller has taken some pressure off, but he has not closed the door on another hike. Payrolls around the +56k consensus with wage growth easing to 3.0% would strengthen the case that the Fed can afford to wait. A much stronger jobs print, especially with hotter wages, could quickly revive the hawkish trade. More importantly, Waller himself has made August inflation the key test. So for now: September hold > hike, but peak rates still need confirmation from CPI. I would rather miss the first leg of a rally than price in the end of tightening too early.

I lean towards buying the pullback gradually rather than waiting for expectations to collapse. The $200m Q4 revenue miss matters because AVGO is priced for near-perfect execution, but the underlying AI story actually strengthened. AI semiconductor revenue hit $16.7bn, +221% YoY, with Q4 expected to accelerate to $21.7bn. More importantly, Broadcom now sees ~$115bn of AI semiconductor revenue in FY2027 and says demand exceeds that outlook. That makes the sell-off look more like a valuation reset than a thesis break. The warning from Credo is still important: at elevated multiples, beats are expected and margins/guidance determine the reaction. I would not chase AVGO immediately after the drop, but I also would not wait for a perfect entry. Scale in if weakness continues. The bigger question

I think weak data and sticky yields can coexist, especially when the market is worried about inflation rather than simply growth. ADP’s 38,000 gain confirms hiring is losing momentum, but the labour market still looks more “slow hire, slow fire” than recessionary. Meanwhile, services input prices have climbed to a three-year high, keeping the Fed’s inflation problem alive. That explains why the 10-year barely responded, holding around 4.8%. Friday’s payrolls are therefore crucial. Consensus is roughly +56,000 with unemployment at 4.1%. A clear downside miss plus softer wages could finally pull yields lower. But weak payrolls with sticky wage inflation may reinforce the uncomfortable regime we are already seeing: slower growth without cheaper money. For now, I would not aggressively positio

Three very different winners, but one common theme: technology is being repriced on its ability to create durable earnings. $DELL shows that AI demand is moving beyond chips into the infrastructure needed to deploy compute at scale. A huge backlog is bullish, but execution and memory supply will determine how quickly that demand becomes revenue. $PANW represents the second-order AI trade. More AI means more data, endpoints and attack surfaces, making cybersecurity increasingly essential. The catch is valuation: necessity does not automatically justify any price if growth slows. $MRNA is perhaps the most interesting. Successful oncology development could transform Moderna from a post-COVID story into a broader mRNA platform, but biotech remains binary and commercialisation still has to fol

I would wait for the print rather than front-run it. Broadcom's AI story is clearly real: Q2 AI semiconductor revenue hit $10.8bn, +143% YoY, and management guided Q3 to $16bn. But the market already knows that. The real hurdle is whether those huge orders translate into durable margins and higher FY27 guidance. The heavier AI mix itself is expected to compress gross margin towards 74%, while Broadcom explicitly warns that custom accelerators and AI systems carry lower gross margins. The Google-Marvell threat also looks more medium-term than immediate, with Marvell saying the Google deal becomes much more significant only in FY29. So my trigger is not simply "$16bn AI revenue achieved". I want $16bn+, resilient margins and, most importantly, an upgrade or stronger evidence behind the $100b

I would rotate modestly toward cash-flow certainty, not abandon growth. A 10-year yield near 4.8% is more than noise: it raises the discount rate on long-duration AI earnings and makes debt-funded capex increasingly expensive. With oil adding another inflation impulse and September hike odds now around two-thirds, richly valued tech has less room for disappointment. But payrolls are the potential circuit-breaker. A genuinely weak jobs print could pull yields back quickly and revive duration-sensitive tech. So rather than chase the bond selloff, I would favour profitable, cash-generative companies and keep some dry powder. If payrolls surprise stronger, 5% on the 10-year becomes a much more uncomfortable possibility. If they disappoint sharply, today's tech weakness could become the entry p

I see this as more of a valuation reset than the AI bubble bursting, although I would not rush to buy everything that has fallen. The key distinction is between AI demand and AI-stock valuations. AI infrastructure spending, cloud demand and data-centre investment remain substantial, but many stocks had already priced in years of exceptional growth. Once expectations become that high, even strong earnings can trigger sharp corrections. I would favour gradual accumulation of profitable AI leaders with strong cash flow and durable demand, rather than trying to catch the biggest losers simply because they are 40–50% cheaper. The biggest risks now are tighter Fed policy, compressed valuations and any evidence that AI capex is producing weaker returns than expected. My outlook: AI remains a long

I’d watch Hormuz transit before aggressively chasing energy stocks. The US-Iran exchange clearly restores the geopolitical premium, with Brent back above $90, but the key question is whether this translates into a sustained physical supply disruption. There are already warning signs: visible commodity-vessel traffic through Hormuz fell to around five ships a day over the weekend, while a tanker was reportedly struck by a projectile. Yet Gulf oil exports have recovered substantially from their March lows, suggesting flows have not collapsed. So I wouldn’t chase XLE purely on headlines. I’d consider energy as a partial hedge, then add only if tanker traffic deteriorates, insurance/freight costs surge or export infrastructure is hit. If Hormuz flows keep recovering, Brent’s war premium could

I would choose tech, but add gradually rather than chase. Warsh’s message materially changes the near-term regime. He explicitly said 2% PCE is a “firm, fixed target”, financial conditions are not broadly restrictive, labour markets are consistent with full employment, and inflation progress has been modest. Markets now price roughly a 60% probability of a September hike, while Barclays has shifted to expecting September and December hikes.  My ranking would be Tech > Gold > BTC for the next several months. Tech faces valuation compression from higher yields, but AI capex and earnings growth provide a fundamental earnings anchor. Gold remains attractive structurally, but after its enormous run, a stronger dollar and rising real yields could force further consolidation. Bitcoin i

I think the July AI selloff was part forced liquidation, part overdue repricing, but the liquidation probably amplified what would otherwise have been a healthier correction. AI fundamentals did not suddenly collapse. Demand for compute, cloud infrastructure and enterprise AI remained strong. What changed was the market’s willingness to pay increasingly high multiples while hyperscaler capex kept rising faster than near-term monetisation. Forced selling then turned a valuation reset into a sharper decline as crowded positions were unwound. The subsequent broad rebound across Nvidia, software and cybersecurity supports this view. I would not interpret the recovery as a return to “buy anything AI”, though. From here, earnings growth, margins and evidence of returns on AI spending should inc

I’m watching CRWD most closely. Unlike some of the more extended names, CrowdStrike’s momentum indicators are still recovering, which could leave more room if buyers continue to follow through after the earnings surge. The key test is whether it can consolidate above the post-earnings breakout rather than quickly filling the gap. NVDA is my second watch. Its earnings confirmed that AI infrastructure demand remains powerful, but margin pressure from higher memory costs gives the market something tangible to debate. For me, CRWD offers the more interesting risk/reward after this rally: strong fundamental momentum without looking quite as technically stretched as CRM, NOW or FTNT. I would watch for consolidation rather than chase another vertical move.

Q1: Nvidia’s guidance makes me more constructive on AI hardware into September. Q3 revenue guidance of $108bn and Data Centre growth of 117% YoY show the capex cycle remains powerful. But I would wait for Jackson Hole before adding aggressively because a hawkish Warsh could compress valuations even if earnings remain strong. Q2: I think META/SNAP is the start of a broader regulatory theme, not a one-off. Meta’s settlement comes amid thousands of lawsuits involving Meta, Snap, TikTok and Google, while parts of Meta’s settlement specifically encourage competitors to adopt similar protections. Smaller platforms may feel the compliance burden more heavily. Q3: AI capex drives the earnings, but Fed policy determines the multiple investors are willing to pay for those earnings. So I remain bull

My pick is Nvidia. Thursday finally broke its four-quarter post-earnings losing streak, with NVDA +8.74%, backed by revenue more than doubling YoY and a supply-constrained FY28 outlook. That looks more durable than simply catching a sector re-rating. Software is the more interesting tactical trade. Salesforce +22.58%, CrowdStrike +20.50% and Okta +28.63% showed that AI may expand enterprise software demand rather than destroy SaaS. Salesforce’s AI-related ARR reached $3.9bn, while CrowdStrike is seeing AI expand both cyber threats and security spending. But after 20–29% one-day gaps, I would not chase immediately. So: NVDA for conviction, software on a pullback, and Intel/Broadcom only as secondary catch-up trades. The key question now is whether software can hold Thursday’s gains once the

Marvell beat, raised guidance, and still broke. Q2 revenue rose 37% YoY to $2.74bn, while Q3 guidance of $3.15bn topped consensus. Yet MRVL fell 1.49% in regular trading and nearly 8% after hours. The issue was expectations: after a 184% YTD rally, investors wanted more, particularly from the Google custom-chip deal. Management indicated its bigger contribution comes in FY29 rather than FY28. My pick is Broadcom. It offers the strongest combination of custom AI silicon, networking and optical exposure without relying on one part of the supply chain. Marvell still has an excellent growth story, but valuation and expectations make execution risk high. For higher-risk upside, I prefer upstream optics such as Lumentum or Coherent. AI clusters need increasingly more optical connectivity regardl