Yes, but the market is now demanding evidence that the spending produces cash returns. Alphabet’s underlying Q2 numbers were actually formidable: revenue +24%, Search +17%, Cloud +82% to US$24.8 billion, while Cloud operating margin reached roughly 35.6%. That suggests AI infrastructure is already generating meaningful revenue rather than being purely speculative spending.  The concern is the sheer scale. Alphabet lifted 2026 capex guidance to US$195–205 billion, while Q2 capex reached US$44.9 billion and free cash flow fell to negative US$5.9 billion.  I therefore see the 6.9% decline primarily as a valuation reset, not yet a thesis break. Alphabet has perhaps the strongest ability to absorb AI spending because Search, YouTube and Cloud provide enormous monetisation channels. Cl

I see this as more than a routine risk reset, but not yet a confirmed trend reversal. The 14.5% plunge reflects the market suddenly demanding evidence that Tesla’s AI valuation can eventually translate into cash flows. The numbers justify the concern. Q2 revenue reached a record $28.2B and deliveries rose strongly, but operating margin collapsed to 1.4%, versus the Street’s pre-report expectation of roughly 5.4%. Free cash flow was -$1.1B, largely because capex more than doubled sequentially. Tesla expects over $25B of capex in 2026, with spending continuing to rise as Robotaxi, Optimus, AI compute and manufacturing capacity expand.  The crucial distinction is that Tesla is not suffering from collapsing demand alone. It is deliberately sacrificing current profitability to finance busi

Intel’s Q2 looks like a credible turnaround milestone, not yet proof that the turnaround is complete. Revenue hit $16.13B, up 25% YoY, while data-centre revenue jumped 59%. More importantly, Q3 guidance also came in well above expectations.  The bullish case is strengthening: AI infrastructure is reviving demand for Intel CPUs, pricing power is returning, and adjusted gross margin reached 41.8%, with management guiding roughly 42% next quarter.  The catch is valuation and execution. At around $98, investors are already pricing in a substantial recovery. Intel still needs to demonstrate that 18A can ramp economically and external foundry customers can become meaningful, rather than the recovery being mainly stronger CPU pricing and AI server demand. I would view the next phase rou

This looks more like a valuation reset than a deterioration in the business. Alphabet's underlying fundamentals remain strong: Cloud revenue growth of 82% YoY suggests AI demand is accelerating, not slowing. Management says demand exceeds capacity, implying additional investment is driven by customer demand rather than speculative spending. Search, Cloud and AI together give Alphabet multiple earnings engines. The concern is that higher capex delays free cash flow expansion and compresses near-term returns, prompting investors to lower the valuation multiple even if long-term earnings potential improves. For long-term investors, that can create an opportunity, provided they believe management will convert today's AI spending into future cash flows. The key metrics to watch are Cloud operat

$SUPER MICRO COMPUTER INC(SMCI)$  The results are encouraging, but I would not conclude that the bear case has been fully disproven. The bullish arguments are compelling: Record Q4 orders suggest AI infrastructure demand remains robust. Improving gross margins indicate pricing pressure may be easing, rather than worsening. As a key NVIDIA GB-series server integrator, Supermicro is well positioned if hyperscaler AI spending continues. However, caution is still warranted: A near-20% one-day rally likely prices in much of the positive surprise. Investors will want several quarters of consistent execution, not just one strong update. The company still needs to rebuild credibility after its accounting and governance issues. Overall, this looks mor

Markets are increasingly rewarding credible long-term returns on AI spending, not AI spending by itself. Tesla's after-hours drop suggests investors now want stronger evidence that higher capex will translate into sustainable cash flows. Valuations should be re-anchored around four questions: Return on invested capital (ROIC): Will each additional dollar of capex eventually earn attractive returns, or merely support growth? Free cash flow timing: Companies with heavy investment phases deserve lower near-term valuation multiples until cash generation recovers. Execution risk: Proven operators such as NVIDIA have earned premium multiples because they consistently convert investment into profits. Others must still prove they can. Competitive moat: Spending to widen a durable lead deserves a h

The market's reaction makes sense. An investment by NVIDIA is not merely a financial transaction. It is often viewed as validation that a company has strategic importance within the AI ecosystem. However, whether this becomes a lasting re-rating depends on Nebius proving it can translate infrastructure investment into sustained earnings growth. Here is how I see it: Bullish case NVIDIA's 9.3% stake is a strong signal. NVIDIA has visibility into AI infrastructure demand through its GPU customers. Its investment suggests confidence that Nebius could become an important AI cloud provider. AI cloud demand remains robust. As enterprises increasingly train and deploy large AI models, demand for GPU cloud capacity continues to exceed supply in many segments. Positive read-across. The gains in Cor

This rally was not driven by a single news headline. Instead, it reflected a combination of improving pricing expectations, renewed institutional buying, and optimism that the memory downturn may have been shorter than feared. The main catalyst is pricing. If Morgan Stanley's forecast of a roughly 25% increase in memory prices proves accurate, earnings leverage for memory manufacturers can be substantial because many production costs are relatively fixed. Rising average selling prices can therefore translate into a disproportionately large improvement in margins. SK Hynix's earnings on 29 July will be particularly important because it is the world's leading supplier of high-bandwidth memory (HBM) used in AI accelerators. Investors will focus on: HBM demand and pricing. DRAM and NAND pricin

Tesla's Q2 report is likely to be judged much less on whether it beats EPS or revenue and much more on the quality of those earnings. The key areas I would watch are: Automotive gross margin: This is probably the most important metric. Deliveries have already surprised to the upside, so investors now want to know whether Tesla had to sacrifice profitability through discounts and incentives. A stable or improving margin would be a strong positive.  Free cash flow and capital expenditure: Tesla is investing aggressively in AI infrastructure, Robotaxi, Optimus and manufacturing. Higher spending is acceptable if management clearly demonstrates progress, but weak cash generation without convincing milestones could disappoint.  Energy storage: Energy continues to become a larger contri

AMD's Helios AI rack is strategically important, but it is unlikely to meaningfully erode NVIDIA's dominance in the near term. The launch is more significant as a long-term positioning move than as an immediate competitive threat. Why Helios matters AMD is no longer selling just GPUs. It is offering an integrated AI platform combining GPUs, CPUs, networking and software, similar to NVIDIA's rack-scale approach. Large cloud providers increasingly prefer complete, validated systems because they simplify deployment and improve performance. AMD can leverage its strengths across EPYC CPUs, Instinct GPUs and Pensando networking to provide an end-to-end AI infrastructure solution. Why NVIDIA's moat remains formidable CUDA ecosystem: NVIDIA still has the most mature AI software stack, libraries an

The rally appears to be driven by a fundamental improvement, but a 19.6% one-day gain likely also reflects sentiment and momentum. Why this is more than just hype: US$2.8 billion of new multi-year AI cloud contracts is tangible business, not merely an AI announcement.  IREN increased its 2026 AI Cloud annualised run-rate revenue target from US$3.7 billion to over US$4 billion, with about 85% of that target already under contract, giving investors much better revenue visibility.  The customer list includes major names such as Microsoft, NVIDIA, Perplexity, Together AI, Figure AI and Fluidstack, suggesting demand comes from credible counterparties rather than speculative startups.  Around 45% of GPU capital expenditure for recent deployments is covered by customer prepayments,

My favourite dividend ETF is VWRA, even though it is an accumulating ETF rather than a high-yield one. I prefer focusing on long-term total returns and letting dividends be reinvested automatically. If I were building an income portfolio, I would target an overall yield of around 3.5% to 4.5%, balancing sustainable income with dividend growth. My biggest challenge is avoiding yield traps, where a high dividend ends up being cut because the underlying business is weakening. For a SGD 1 million income portfolio, I'd allocate 40% to global dividend ETFs, 30% to Singapore REITs, 20% to Singapore blue-chip dividend stocks (such as the local banks), and 10% to cash or short-term bonds. The goal would be diversification, resilient cash flow, and room for long-term capital appreciation rather than

$Netflix(NFLX)$  Based on the scenario described, I would lean towards "potentially mispriced, but only if the business can reignite growth" rather than calling Netflix an obvious bargain. The bullish case A share price down around 50% has reset expectations significantly. If the market has become overly pessimistic, even modest improvements in revenue growth or margins could support a recovery. Netflix remains one of the strongest global streaming brands, with a large content library, international reach, and significant pricing power compared with many competitors. If the acquisition of Ben Affleck's AI film startup meaningfully improves production efficiency or content creation over time, it could help address one of Netflix's biggest chal

I would wait for the earnings report rather than buy immediately before it. Here's why: Reasons to wait High event risk. Tesla's earnings often trigger moves of 8% to 15% (or more). Buying beforehand means accepting that binary risk. Several key uncertainties remain: vehicle gross margins, demand trends, guidance, Robotaxi/autonomous driving progress, and management's outlook. Any one of these could dominate the market's reaction. Competition continues to intensify. Mercedes-Benz's CLA 250 Plus is another entrant targeting the premium EV segment. While it is unlikely to derail Tesla on its own, it reinforces the trend of increasing competition. Reasons to stay constructive long term Tesla's valuation is increasingly driven by its autonomy, AI, energy storage, and robotics ambitions rather

The contradiction exists because different parts of the memory market are being priced very differently. The bearish case (cycle peak): Memory has always been one of the most cyclical semiconductor industries. Strong profits eventually encourage capacity expansion, leading to oversupply and falling prices. A 30% decline in Micron despite solid earnings suggests investors are looking beyond today's results and worrying about 2027 to 2028. SanDisk's roughly 580% gain this year means expectations are extremely high. Even excellent execution may not justify such valuations indefinitely. HBM (High Bandwidth Memory) is a hot segment, but it is still only one part of the broader memory market. The bullish case (AI supercycle): AI demand is fundamentally different from previous PC or smartphone cy

Apple's strength makes sense in the current market. When investors rotate away from high-beta AI infrastructure names, they often seek companies with resilient earnings, strong free cash flow, and massive balance sheets. Apple fits that profile. That said, I would be cautious about adding aggressively at fresh all-time highs. Bullish case: Apple generates enormous cash flow and has one of the strongest balance sheets in the market. Its ecosystem provides recurring revenue through services, reducing earnings volatility. If Apple successfully strengthens its in-house AI silicon through acquisitions or internal development, it could improve its long-term AI competitiveness. In a risk-off environment, investors often favour companies with predictable earnings. Reasons for caution: Much of the

Based on the scenario described, I think it is still too early to confidently call a bottom. Several factors suggest the correction may not be over. Reasons to be cautious: Valuation reset: Memory stocks had rallied dramatically on AI and HBM optimism. When expectations become stretched, even good long-term fundamentals cannot prevent sharp multiple compression. Negative sentiment: Downgrades and profit-taking often feed on themselves, especially in a highly cyclical sector like memory. Forced selling by leveraged investors can continue beyond what fundamentals justify. Macro uncertainty: If geopolitical tensions continue to pressure global risk appetite, cyclical semiconductor stocks are usually among the first to be sold. Memory cycles remain volatile: History shows DRAM and NAND stocks

Apple's rally looks less like a new growth story and more like a rotation into perceived quality. The market backdrop supports that interpretation. Cooling CPI and PPI reduced concerns about inflation, while strong bank earnings improved overall risk sentiment. At the same time, investors were exiting the most crowded AI hardware trades after the sharp sell-off in memory stocks. Capital needed a home, and Apple, with its enormous free cash flow, resilient Services business and balance sheet, became a natural destination.  That said, Apple is not without challenges: The stock is trading near record highs after a strong rebound. Hardware growth is expected to moderate after recent strength. Rising DRAM and NAND costs could pressure hardware margins, although Apple has historically been

The pressure is now squarely on TSMC. ASML has effectively told the market that customers are still ordering advanced chipmaking equipment aggressively, with Q2 revenue and profit beating expectations and 2026 guidance raised to €43 to €45 billion. Even more importantly, management said AI customers continue to accelerate capacity expansion.  The question is no longer whether AI demand exists. It is whether TSMC can show that wafer demand is translating into sustained, profitable production. There are four numbers I would watch: 2026 revenue guidance. Any increase would reinforce the AI investment thesis. Capital expenditure. If TSMC raises capex again, it suggests customers such as Nvidia, AMD, Apple and Broadcom continue to reserve future capacity. Gross margin. Investors want to se