AI can justify today's valuations, but only if revenue growth translates into sustained earnings growth. The market is already pricing in massive adoption, so good execution may no longer be enough. Companies need exceptional execution. As for tightening, this looks more like a precautionary inflation response than an aggressive hiking cycle. Unless inflation accelerates materially, central banks are unlikely to tighten indefinitely. For the bull market, the key risk is not rates themselves but earnings. Bull markets usually end when profits weaken, liquidity dries up, or recession risks surge. So far, earnings remain relatively healthy despite higher rates. My view: this is more likely a late-cycle repricing than the beginning of the end. Expect higher volatility, narrower leadership, an
Fresh highs are bullish, but parabolic moves are where risk and reward start to diverge. The memory story is fundamentally stronger than it was in previous cycles. AI training clusters and inference workloads are driving demand for high-bandwidth memory, benefiting companies such as Micron Technology and Sandisk. Unlike past DRAM booms driven mainly by PCs and smartphones, AI data centres are creating a new source of demand. That said, markets rarely move in a straight line. A stock making new highs after a 10% single-day surge often attracts momentum traders, making the trade increasingly crowded. When expectations become extreme, even good results can trigger profit-taking. If you're already long, holding or trimming into strength is easier to justify than chasing. If you're underweight,
A hawkish Fed changes the timing of returns more than the long-term value of quality businesses. Higher rates compress valuations, especially for long-duration growth stocks, but they do not necessarily damage the underlying earnings power of companies like Meta Platforms and Microsoft. If inflation is genuinely re-accelerating and the market begins pricing out cuts, value sectors such as financials, industrials, and energy could continue to outperform in the near term. However, betting heavily on a rapid Fed pivot has historically been risky when inflation remains above target. For long-term investors, a balanced approach often makes more sense than a wholesale rotation. Trimming positions that have become oversized and rebalancing into cheaper areas is reasonable. Abandoning quality grow
A first sell rating matters more as a sentiment signal than a valuation discovery. At current levels, the debate is no longer whether SpaceX is a great company, but whether the market has already priced in years of success from Starship, Starlink, defence contracts, and future businesses. History shows that strong narrative stocks can remain detached from traditional valuation metrics far longer than bears expect. The first sell call rarely marks the exact top. However, once expectations become extreme, execution misses tend to be punished much more severely. If I already held a large gain, I would be more inclined to gradually de-risk than aggressively add. Taking partial profits preserves upside exposure while reducing the risk of a sharp sentiment reversal. If I had no position, I would
I lean toward MANGOS over Magnificent 7 because it better reflects the AI stack: models, compute, cloud, distribution, and infrastructure. If forced to choose between compute and infrastructure, I'd pick compute for this decade. AI demand is exploding faster than chip supply, and every major model still needs massive compute. If I could own only one for 10+ years: 🥇 NVIDIA - best combination of dominance, profitability, and execution. It is the "picks and shovels" provider to the entire AI industry. 🥈 SpaceX - highest upside. If Starlink and Starship achieve their ambitions, today's valuation could look cheap. 🥉 Meta Platforms - underrated due to unmatched user distribution and AI monetisation potential. My ranking: 1. NVIDIA (highest conviction) 2. SpaceX (highest ceiling) 3. Meta 4. Goo
If I had to choose between holding the leader and rotating into weaker names, I would generally prefer holding the leader. A 2.4% decline in NVIDIA versus much larger drops in AMD, Marvell, Intel, and leveraged semiconductor ETFs suggests relative strength. When risk appetite fades, capital often concentrates in the highest-quality companies with the strongest balance sheets, margins, and competitive positions. The more important question is time horizon: If you're a short-term trader, this kind of sector rotation and volatility argues for tighter risk management and potentially reducing exposure. If you're a long-term investor, a 10-20% swing in semiconductor stocks is not unusual. The key thesis is whether AI infrastructure spending remains intact. What would concern me more than a singl
The answer depends on whether you believe this is a temporary rotation or the start of a longer leadership change. My base case would be that this looks more like a rotation than the end of the AI theme. AI infrastructure demand has not disappeared simply because semiconductor stocks corrected. Historically, the strongest secular growth themes often experience multiple 20-30% drawdowns while remaining intact. That said, when a trade becomes crowded, reducing concentration risk is sensible. If AI hardware has grown into an outsized portion of a portfolio, trimming some exposure and reallocating toward quality financials, industrials, or healthcare names can improve diversification without abandoning the theme. For new capital, I would be more inclined to buy quality AI leaders on weakness t
A first sell rating rarely ends a momentum story by itself. What matters is whether the narrative keeps attracting new buyers faster than early investors take profits. For SpaceX, the bull case is straightforward: dominance in launch, rapid growth in Starlink, and the possibility that Starship unlocks entirely new markets. Bears argue that at current prices, investors are paying today for many years of future success, leaving little room for execution mistakes. The fact that shares held above $200 despite a public sell call suggests sentiment remains extremely strong. However, sharp intraday reversals often indicate volatility is increasing and conviction is becoming more divided. If I were already sitting on large gains, I would be more inclined to trim gradually and lock in some profits
I lean toward this being a positive signal, with an important caveat. A bond deal that attracts US$85 billion of orders for a US$25 billion issuance suggests credit investors view NVIDIA as a very high-quality borrower. The ability to borrow at only a modest spread over Treasuries gives Nvidia a powerful advantage. It can fund data centres, networking, software, and AI infrastructure without heavily diluting shareholders. That strengthens the moat because: Lower cost of capital than most competitors. Greater flexibility to invest through cycles. Ability to scale faster if AI demand remains strong. The caveat is what the money is funding. If AI demand keeps growing, cheap debt today could look brilliant in hindsight. If industry capacity expands faster than demand, today's investment could
I would keep some powder dry. A market that can swing from "AI bubble bursting" to "new bull leg" within a week is telling you one thing: sentiment is driving price more than fundamentals right now. The bullish case: U.S.-Iran tensions easing reduces a major macro risk. Liquidity concerns around the SpaceX listing are fading. AI spending remains enormous, supporting names like NVIDIA, AMD and Broadcom. Many stocks corrected sharply before rebounding, resetting some excess optimism. The cautious case: Valuations are still demanding. AI infrastructure spending must eventually justify expectations. Geopolitical risks can re-emerge quickly. Leveraged products like SOXL amplify both gains and losses. If I had fresh capital today, I would probably deploy it gradually rather than go all-in. Somet