Heroic Lifesaver

$Monolithic Power(MPWR.US)

Post 3 of Episode 25

Context: What interests me about Monolithic Power Systems is that the AI power bottleneck doesn’t end when electricity reaches the data centre. As GPU clusters become increasingly power-hungry, efficiently delivering and converting that electricity inside the server rack becomes just as important. MPWR is expanding beyond traditional chip-level power management into higher-voltage solutions for next-generation AI infrastructure.

My trade: I see MPWR as an interesting way to participate in the AI buildout through power efficiency rather than computing performance. Its Enterprise Data revenue grew an impressive 164% year-on-year in Q2 2026, and its move into 800V power architectures could expand its opportunity further. The recent manufacturing partnership with GlobalFoundries also strengthens its ability to support growing demand.

Takeaway: AI doesn’t just need more electricity. It needs to use that electricity more efficiently. As computing density increases, every watt saved in power conversion becomes another watt available for productive computing. Sometimes the most valuable technology isn’t the one doing the calculations, but the one making those calculations possible.

@Captain's Treasure

$APAC Realty(CLN.SG)

Post 2 of Episode 25

Context: What initially attracted me to APAC Realty was its asset-light business model and exposure to Singapore’s property transaction market through ERA. It offered a different source of earnings from my technology-heavy portfolio. However, the business remains sensitive to transaction volumes, property market sentiment and the broader economic environment.

My trade: I decided to close my position, not necessarily because of APAC Realty’s fundamentals, but because I was reassessing where my capital could work hardest. With other opportunities offering potentially more attractive growth prospects and risk/reward setups, I felt that freeing up the capital made more sense than continuing to hold simply for diversification or dividend income. Staying in cash while waiting for better opportunities was also a perfectly acceptable alternative.

Takeaway: One lesson I’ve learnt is that an investment can still be viable before you decide to sell it. Sometimes the opportunity cost of holding a decent business becomes greater than the benefit of staying invested. Knowing when to move on is just as important as knowing when to buy.

@Captain's Treasure

$AEM SGD(AWX.SG)

Post 1 of Episode 25

Context: One development that has caught my attention is AEM’s entry into memory testing. While much of the AI semiconductor conversation revolves around GPUs and HBM, the broader memory industry is also facing increasingly sophisticated testing requirements. AEM expects to ship its first memory final-test handler in Q4 2026, potentially opening another meaningful growth opportunity in 2027.

My trade: What interests me is how this could gradually change AEM’s business mix. Its AI/HPC customer has already become its largest revenue contributor, but expanding into memory testing could introduce another growth driver and reduce reliance on individual customer programmes over time. I’m particularly watching whether these initial shipments translate into sustained production orders.

Takeaway: A company’s transformation becomes more convincing when its technology starts finding applications beyond its original customer base. For AEM, the bigger opportunity may not simply be winning more business from existing customers, but proving its testing expertise can scale across several semiconductor markets.

@Captain's Treasure

$GE Vernova(GEV.US)

Post 5 of Episode 24

Context: What has become increasingly interesting to me about GE Vernova is that its opportunity may extend well beyond the current gas turbine boom. The recent US regulatory approval for its 300 MW BWRX-300 small modular reactor marks an important milestone. With construction already underway for the same reactor design in Canada, nuclear power is gradually becoming a more credible additional growth opportunity.

My trade: I originally viewed GEV primarily through the lens of power generation and grid infrastructure. That thesis remains intact, but the progress in small modular reactors adds another dimension. If standardised nuclear designs can eventually be deployed economically at scale, GEV could participate in both today’s urgent power shortages and tomorrow’s longer-term energy infrastructure expansion.

Takeaway: What makes GEV particularly interesting is that it doesn’t need to bet on one energy technology winning. Whether the solution is gas, grid expansion or eventually nuclear, the underlying demand for reliable electricity remains. Sometimes the strongest investment thesis is having several ways to benefit from the same structural trend.

@Captain's Treasure

$AEM SGD(AWX.SG)

Post 4 of Episode24

Context: What has become particularly interesting to me about AEM is that its growth story may extend well beyond selling more semiconductor testing equipment. In Q2 2026, device-specific configurables and collaterals reached S$72.3 million, growing 12.6% sequentially. These are the components and upgrades customers need as their chips evolve, creating opportunities for repeat sales from AEM’s installed equipment base.

My trade: I increasingly see AEM as a business with a potential compounding effect. Every new testing platform deployed can establish a longer-term customer relationship, while successive chip generations create demand for additional configurations and upgrades. With its AMPS backlog exceeding S$400 million and FY2026 revenue guidance raised to S$630–680 million, that installed-base opportunity is becoming particularly interesting.

Takeaway: Sometimes the real value of selling equipment isn’t just the initial sale. It’s the repeat business that follows. If AEM executes its production ramp successfully, today’s equipment deployments could become tomorrow’s recurring revenue opportunities — potentially making its earnings more resilient over time.

@Captain's Treasure

$Sandisk(SNDK.US)

Post 3 of Episode 24

Context: What is becoming increasingly interesting to me about SanDisk is that the story is no longer simply “NAND prices are going up.” Fiscal 2026 Datacenter revenue grew 437%, with exabytes sold nearly doubling and revenue per gigabyte rising sharply. At the same time, SanDisk has now signed 10 New Business Model agreements, designed to deepen relationships with strategic customers and potentially make demand and returns more durable across the cycle.

My trade: That changes how I think about SNDK. Historically, NAND has been a brutally cyclical commodity business — great pricing eventually encourages too much supply. If SanDisk can combine disciplined capacity, higher-value enterprise SSDs and longer-term customer arrangements, there is a possibility that the next cycle looks structurally better than the ones that came before. The numbers are already reflecting the shift: Q4 revenue reached $8.97 billion, up 51% sequentially, with Datacenter revenue more than doubling quarter-on-quarter.

Takeaway: The most interesting part of the SanDisk story may no longer be simply that NAND is scarce. It is whether SanDisk can turn a historically cyclical business into a more disciplined, higher-value and increasingly predictable one. If that happens, the market may eventually value SNDK very differently from the memory company it used to be.

@Captain's Treasure