To The Moon

I would be more selective rather than outright trimming or chasing. The key observation is that if inflation really is re-accelerating while AI-related stocks continue to rally, it suggests investors believe AI earnings growth is large enough to offset higher discount rates. That is a powerful signal, but it also raises concentration risk. My current hierarchy would be: Still bullish: AI infrastructure, power, data centres, networking, optical communications, memory. Neutral: Mega-cap AI leaders that already trade at demanding valuations. Cautious: Highly speculative AI names with little earnings support. If inflation stays elevated and the Fed remains hawkish, the biggest risk is not necessarily AI demand collapsing. It is valuation compression. A company growing earnings 40% can still se

If I had to rank the AI chain today, I would separate it into three buckets: 1. Power & infrastructure (highest conviction, longest runway) The market spent 2023-2025 obsessing over GPUs. The next bottleneck is increasingly electricity, cooling, transformers, grid upgrades, and data centre infrastructure. Goldman Sachs estimates data centre power demand could rise over 200% by 2030.  Names and themes linked to power, cooling, nuclear, and digital infrastructure may have a more durable runway because AI cannot scale without physical energy and facilities. 2. Memory & storage (strongest momentum) HBM has become the critical bottleneck in AI servers. New memory-focused ETFs such as DRAM have attracted enormous inflows and delivered explosive gains.  The risk: memory is still

The divergence among major banks comes down to one question: is gold still primarily a safe-haven asset, or has it become a macro trade on interest rates and the US dollar? Bullish banks argue that central bank buying, fiscal deficits, geopolitical risks, and potential future rate cuts support gold over the long term. Bearish banks focus on sticky inflation, higher real yields, a stronger dollar, and reduced safe-haven demand if global growth remains resilient. As for ETF outflows, I would not blindly follow them. ETF flows are often momentum-driven and can overshoot in both directions. A 17% correction after a strong rally is painful, but not unusual for gold. That said, I would also avoid aggressively "catching the falling knife". If ETF outflows are accompanied by falling central bank d

Major banks are split because they're focusing on different drivers. Bears: Higher real yields, resilient USD, and ETF outflows. If rates stay high, gold faces a headwind. Bulls: Central-bank buying, rising government debt, geopolitical risks, and eventual rate cuts. They see the recent correction as temporary. For ETF outflows, I would not blindly follow them. ETF investors are often late to both tops and bottoms. More important is whether central banks continue accumulating. My stance: Short term: Neutral to cautious. Momentum remains weak. Long term: Moderately bullish. Strategy: Gradual accumulation rather than an all-in dip buy. The signal I'd watch is ETF outflows slowing while central-bank demand stays strong. If that happens, the current correction may look more like a reset than

The fact that AI stocks held up despite hotter inflation is important. It suggests the market is currently treating AI as a structural earnings story rather than a liquidity story. The bull case: Hyperscalers are still spending aggressively on AI infrastructure. Revenue is increasingly material, not just future promises. AI capex has become strategic. Companies fear underinvesting more than overinvesting. Productivity gains from AI could eventually offset some inflation pressures.  The bear case: Core PCE at 3.3% is moving in the wrong direction for the Fed.  Markets may still be underestimating the probability of rates staying higher for longer. If bond yields rise sharply again, long-duration AI names become vulnerable regardless of fundamentals. History shows that even the str

The AI chain is becoming so broad that stock picking and ETF investing are now two very different bets. Stock picking works best when you correctly identify the bottleneck before Wall Street fully prices it. That was NVIDIA in 2023, HBM memory in 2025, and arguably power, cooling and photonics today. ETF investing works better when you believe AI spending will keep expanding, but you are less certain which winner ultimately captures the profit pool. My current ranking: 1. Memory (Most bullish near-term) The market has realised AI is not just a GPU story. HBM supply remains constrained, pricing power is strong, and AI servers require enormous memory scaling. Recent launches like the Roundhill DRAM ETF show how aggressively capital is rotating into this theme.  2. Data centres + power T

The divergence among banks is not really about gold itself. It is about which macro force they think will dominate. Bullish banks such as [J.P. Morgan](https://www.jpmorgan.com/insights/global-research/commodities/gold-prices?utm_source=chatgpt.com), UBS and ANZ are focused on: Continued central bank buying Geopolitical fragmentation Fiscal debt concerns and de-dollarisation Potential Fed easing later in the cycle More cautious houses such as Macquarie and some Morgan Stanley analysts are focused on: Higher real yields Stronger USD ETF outflows Positioning excess after a massive multi-year rally  The key point is that gold's recent decline does not automatically invalidate the long-term bull case. Gold peaked near US$5,300-5,600 before correcting roughly 15-18%, which is painful but n

A first breakout above a major psychological level like the NASDAQ 100 at 30,000 is symbolically powerful, but historically these moments can represent either: 1. genuine mid-cycle acceleration, or 2. late-stage momentum concentration before volatility expands. Right now, the evidence still leans more bullish than bearish structurally. The important detail is what is driving the rally. This is not purely speculative software multiple expansion anymore. The move is increasingly backed by real AI infrastructure cash flows: exploding HBM demand, hyperscaler capex commitments, sovereign AI spending, power/networking buildouts, and earnings revisions still moving upward for firms like NVIDIA and Micron Technology. That makes this rally look more like the middle innings of a capital expenditure

The key debate now is no longer “Will HBM demand grow?” but “Can supply growth finally catch demand before hyperscaler capex peaks?” Right now, I still think demand is large enough for both Micron Technology and SK Hynix to outperform simultaneously through at least the next 12–18 months. The market is effectively treating HBM as strategic infrastructure rather than commodity memory. Capacity for both firms is reportedly heavily booked well into 2026.  But the nuance is important: SK Hynix likely compresses Micron’s premium multiple, not necessarily its earnings. Micron’s rerating came partly from the narrative that it was the “catch-up winner” in HBM. If SK Hynix massively expands HBM3E/HBM4 output while maintaining Nvidia relationships, Micron’s scarcity premium could narrow even if

The space trade is increasingly splitting into three very different risk profiles, despite the market currently treating them as one “SpaceX sympathy basket”. For me, Rocket Lab is still the strongest long-term institutional-quality setup. The difference is that RKLB is evolving from a speculative launch company into a vertically integrated defence and space systems contractor. The SDA milestones, hypersonic HASTE work, and multi-billion backlog visibility give it more durable revenue foundations than most peers. Neutron is still the key execution risk, but if it succeeds, RKLB’s valuation framework changes entirely.  AST SpaceMobile is the highest-upside but also the highest binary-risk name. The direct-to-cell thesis is massive if execution works, because it targets a potentially en