When Running Out of Space Becomes Expensive For most of my investing life, hard drive makers lived in the bargain bin of technology. Demand surged, factories expanded, supply flooded the market, and margins collapsed. Investors learned not to get emotionally attached. Scarcity matters now in a way it never did before. Cloud consolidation means a handful of hyperscalers command enormous, predictable storage volumes, allowing them to pre-book multi-year capacity. Add AI workloads that generate massive cold data archives, and suddenly bulk storage demand is structurally sticky. Seagate now occupies the opposite end of the spectrum. Its manufacturing capacity is effectively committed through 2026, with hyperscale cloud operators already locking in supply. Instead of chasing unpredictable PC de

The Porphyry Pivot: How Aris Mining Is De-Risking Colombian Gold Through Industrial-Scale Underground Mining While analysts still tend to file Aris Mining under ‘high-risk Colombian juniors,’ the company is quietly doing something far less exciting—and far more valuable. It is building an underground gold operation designed to behave less like a prospecting venture and more like a factory. The market loves romance in mining. What it often struggles to price correctly is repeatability. That mismatch matters, because $Aris Mining(ARMN)$ no longer resembles the speculative profile it is routinely assigned. With a market capitalisation around US$4 billion, nearly US$770 million in trailing revenue, and operating margins above 40%, this is not a story

How Erie Indemnity quietly collects its cut while others wrestle the wreckage A Strange Beast in a Familiar Zoo Most insurance stocks behave exactly as you would expect. They obsess over weather maps, argue with regulators about rate filings, and spend earnings calls explaining why ‘this year was unusual’. Erie Indemnity does none of that, largely because it does not actually insure anyone. Erie sells tickets while others wrestle the storm That makes it a strange creature indeed. $Erie Indemnity(ERIE)$ lives inside the insurance ecosystem but outside its danger zone. It does not price risk, reserve for losses, or pay claims. Instead, it runs the machinery—distribution, systems, administration—for Erie Insurance Exchange and collects a management f

From Trainers to Operating Leverage: Why I Think On Holding Is Being Valued Like a Platform, Not a Brand $On Holding AG(ONON)$ rarely gets discussed in the same breath as software platforms, yet that is exactly how Wall Street is beginning to frame it. I find that intriguing, because on the surface this is still a Swiss sportswear company selling trainers, apparel and the occasional sharply priced jacket. Look closer, however, and the market’s enthusiasm starts to make sense—this isn’t fashion infatuation, it’s a numbers-driven bet on operating leverage. Not a shoe story—an operating system in premium apparel form The share price tells a familiar growth-stock story. After peaking above 64, the stock has pulled back into the mid-40s, leaving it dow

I am not revisiting $Lumen Technologies(LUMN)$ because I have suddenly developed a soft spot for telecom turnarounds. That story is well worn and usually ends with investors discovering that leverage and secular decline make poor travelling companions. What interests me in early 2026 is something far more specific and far less nostalgic. Lumen is quietly positioning itself as a control point in the physical flow of AI data, at precisely the moment when data movement is becoming as strategically sensitive as compute itself. This is not a comeback story. It is a bottleneck trade. Data flows everywhere—except where Lumen controls the route Control beats capacity The prevailing narrative around Lumen’s AI pivot focuses on selling fibre to hyperscalers

When Bitcoin Stops Behaving Like a Trade and Starts Acting Like Infrastructure This Is Not a Price Call, It’s a Plumbing Inspection Asking where Bitcoin trades next quarter is the wrong question—possibly even a boring one. The more relevant question is how it is now being used. BlackRock’s Bitcoin ETF has quietly crossed a threshold from speculative access point to something closer to financial infrastructure. An asset once explained with memes is now explained with risk committees, which may be the clearest sign of maturation yet. Bitcoin’s wiring is now being installed into mainstream portfolios IBIT is no longer behaving like a vehicle for excitement. It is behaving like a component. Components are not bought for thrills; they are installed, monitored, and only discussed when they fail.

When AI Hits the Storage Wall For the past two years, AI investing has felt like a talent show judged exclusively on compute. Faster chips, bigger clusters, louder earnings calls. Storage, meanwhile, was expected to sit quietly in the background and do its job. SanDisk’s recent surge suggests that assumption has finally collapsed. When a stock climbs more than seventy percent in the first eleven trading days of the year, crossing $400 in the process, something more structural than enthusiasm is at work. AI’s speed meets storage’s choke point—SanDisk holds the gate My view is that SanDisk’s real edge is not AI excitement, but control over a narrow and suddenly irreplaceable choke point in the data-centre buildout. Intelligence without memory is not just useless; in AI training environments,

From Call Centres to Cognitive Infrastructure: Why IBEX Is Quietly Becoming an AI Cash-Flow Compounder Most AI investment narratives still orbit the same celestial bodies: chips, clouds, and code. I find $IBEX Ltd.(IBEX)$ interesting because it lives somewhere far less glamorous and far more accountable. It operates at the execution layer of customer experience, where AI is no longer judged on promise or demos but on whether it actually lowers costs, improves outcomes, and shows up in margins. In today’s environment, that distinction is everything. AI quietly replacing call centres, one customer interaction at a time IBEX is not selling AI as a concept. It is selling operational results. Enterprises have largely moved past the experimentation phas

From Science Project to Supply Chain Infrastructure: Why I See Kodiak AI Crossing the Commercial Rubicon I have grown increasingly sceptical of autonomous vehicle stories framed as moonshots. They tend to promise robo-everything, everywhere, all at once, and then quietly collide with reality. Kodiak AI feels different, not because its technology is flashier, but because its ambition is deliberately dull. I do not see $Kodiak Robotics(KDK)$ as an autonomous vehicle company chasing a sci-fi future. I see it as a nascent piece of freight infrastructure, inching its way into the most economically dense arteries of global trade. That distinction matters more than most investors realise. Autonomy succeeds when it becomes infrastructure, not spectacle Aut

Intel’s Quiet Repricing: Why Scarcity, Not Sentiment, Is Driving the Next Leg of the Turnaround Intel’s resurgence has been widely misunderstood. This is not a forgiveness rally, nor a nostalgic bet on past dominance. The market is not suddenly convinced Intel has become the most elegant chip designer in the room. What is happening instead is far more pragmatic. $Intel(INTC)$ is being repriced because it can ship at scale in a world where scale has quietly become scarce. When supply matters more than stories, scarcity builds a moat That distinction matters. Scarcity, unlike sentiment, is stubborn. It does not care about old narratives, only about who can deliver when demand arrives. Right now, demand is arriving in bulk. When ‘Sold Out’ Stops Bein

Duolingo’s share price collapse has been dramatic enough to reclassify it from ‘Magnificent Growth’ to something far less glamorous, closer to a forgotten compounder hiding in plain sight. Down roughly 67% from its record high and trading near the cheapest valuation it has seen since going public in 2021, the stock now sits in that uncomfortable no-man’s land where optimism has evaporated but fundamentals have not. The market’s judgement appears clear: $Duolingo, Inc.(DUOL)$ was a pandemic darling, and the party is over. I am not convinced that verdict holds up under scrutiny as we look towards 2026. This is not an argument about whether people will keep learning languages. It is an argument about whether the market is still valuing Duolingo as a

If Hong Kong’s stock market were a mosaic rather than a monolith, today would be the moment when you stop inspecting individual tiles and step back a few paces. Some pieces gleam with speculative polish, others look worn but dependable, and a few are still being pressed into place by capital that knows something is changing but hasn’t yet decided how fast it wants to move. Not one market — many ideas converging into something coherent This is not a rerun of an old tech rally, nor a simple relief bounce after a soft prior year. What I see instead is a market quietly reorganising itself around three patterns: a reshaping of the investible universe through high-technology IPOs, the continued gravitational pull of established technology and new-energy leaders, and a rotation into robotics and

Why I Prefer Control to Excitement Markets have a habit of confusing activity with progress. Every cycle produces its share of dazzling narratives, but very few companies actually control outcomes rather than react to them. My conviction in $Taiwan Semiconductor Manufacturing(TSM)$ and $Palo Alto Networks(PANW)$ comes from that distinction. These are not bets on which technology story wins the news cycle; they are bets on who gets paid regardless of which story wins. One sits at the physical choke point of advanced computing, the other at the operational choke point of enterprise security. Neither is flashy in the way the market currently rewards, which is precisely the point. Control points dictate outcom

XRP’s latest surge has reignited the usual crypto shouting match, which tells me one thing straight away: something practical is happening beneath the noise. When an asset that half the market loves to hate starts rallying without a catchy narrative or a celebrity mascot, it is rarely an accident. XRP’s move feels less like a speculative sugar rush and more like a response to pressure building in the real financial system, particularly where that system is weakest. XRP moves quietly where traditional finance struggles to keep pace At around $2.34, XRP is up sharply on the day, carrying a market capitalisation close to $142 billion and daily trading volume pushing $8 billion. This is not a thin, excitable market. It is deep, liquid and being actively repositioned. The interesting question i

From Search Queries to Self-Learning Systems Baidu’s recent 9% surge has been treated as a feel-good rally in a battered China tech name. I think that misses the point. The market continues to value $BIDU-SW(09888)$ as a search engine with advertising ambition, yet the company is quietly building something far more ambitious: a self-reinforcing machine that learns from the real world, improves itself, and then sells that intelligence back to everyone else. Baidu transforms data into a self-learning AI machine The most underappreciated piece of this puzzle is Apollo. Few investors realise that even if robotaxis themselves never turn a profit, the data they generate may already be paying dividends by improving Baidu’s AI models and infrastructure.

From Moonshots to Margins: Why Meta’s Next Rally Is About Capital Discipline, Not the Metaverse For years, owning $Meta Platforms, Inc.(META)$ has required a strong stomach and an even stronger imagination. I have watched the company oscillate between operational brilliance and strategic excess, often within the same earnings call. The reported plan to cut the metaverse division’s 2026 budget by as much as 30% feels like a genuine inflection point—not because it flatters near-term earnings, but because it signals something more important: a renewed commitment to capital discipline over corporate theatre. The market’s reaction to Meta’s most recent earnings was swift and unforgiving, leaving behind a visible earnings gap and reviving talk of whethe

CES used to be where we pretended to care about smart refrigerators. Now it’s where Nvidia and AMD duke it out to see who gets to power the robot that might replace your job—or at least the one you didn’t want anyway. This year, I’m less interested in marginal GPU gains and more focused on whether these chip titans can convincingly present themselves as robotics platforms rather than pure AI silicon vendors. If robotics is the next stock engine, it will look very different from the last one. AI is leaving the cloud and learning to lift real weight When Compute Leaves the Cloud The first structural shift investors often underestimate is how robotics changes the AI compute demand profile. Data-centre spending remains lucrative but cyclical, tied to hyperscaler budgets. Robotics flips that mo

Applied Digital’s share price does not do understatement. With a five-year beta north of seven, this is a stock that treats volatility less as a feature and more as a calling card. Yet focusing solely on price swings, or even the eye-catching contract backlog, misses what I see as the real value multiplier here. The core opportunity is not simply how many megawatts $APPLIED DIGITAL CORP(APLD)$ can promise, but how effectively it converts contractual certainty into immediate build-ready capital. In an industry throttled by development bottlenecks, that is a rare and underappreciated advantage. Where contracts turn into capital before the concrete even sets Most commentary frames Applied Digital as a leveraged bet on AI data centre demand. In my vie