$6135.T In a March interview, Makino's sales executive said: “We’ve heard about Starlink-related components, so that might be a positive development for our company as well.” I checked, and he is right - SpaceX likely uses Makino's machines to make Starlink satellites. SpaceX manufactures Starlink satellites at the dedicated facility in Redmond, Washington. GPT 5.5 helped me unearth local county permit records, which show new foundations for relocated machinery and related fire-alarm alterations in early 2025 under the project name "SPACEX SE03 - MAKINO EXPANSION AND HITL". In addition, SpaceX is looking for experienced machinists for the Redmond facility to operate multi-axis CNC machines, including 5-axis mills and 6-axis lathes. This convinced me that Elon is also a fan of Makino's gigantic "mother machines".

One of the main reasons I invested in Santec $6777.T was its highly lucrative medical business – MOVU. I generally understood where the optics are going, but couldn’t imagine the speed and scale of transition. On the other hand, Santec’s success in the eye diagnostics market in partnership with Alcon $ALC was evident, but not everyone paid close attention. Two years ago, I thought investors failed to price in the upcoming launch of ARGONAUT – the new multi-modal eye imaging system that would supplement the ARGOS biometer. The irony is that the ARGONAUT is still awaiting FDA approval, while Santec’s share price took off without it. We might finally see the ARGONAUT later this year, though. On the recent earnings call, Alcon’s management indicated that Unity DX whole-eye analyzer (Alcon’s branded version of the ARGONAUT) is about to launch. Considering Santec's rich valuation, I don’t think this will move the share price much when it happens, but it could still be a nice surprise.

Sold Sphere ($347700.KQ) at a small profit to double down on HVM ($295310.KQ). HVM is rapidly expanding output of advanced metal alloys to support Starship production. Orders from SpaceX are rising faster than sales, boosting near-term visibility. Now all we need is daily Starship launches by August 2027, as Musk promised. Easy-peasy 👍

$4368.T FUSO Chemical reported strong Q1 numbers across the board: revenue up 22%, operating profit up 32%, and net income up 115% (incl. government subsidies). This time, both the semiconductor materials and life sciences businesses contributed to earnings growth. Higher sales volumes of ultra-high-purity colloidal silica, price revisions announced earlier this year, and weak JPY were the key drivers. The life science business (food acids) finally began to recover with operating profit up 47%. The unchanged guidance implies a surprisingly weak Q2, which I think is too conservative. To meet overwhelming compute demand, chip foundries worldwide are ramping up wafer output and consuming increasing volumes of CMP slurries as they bring more production lines online. If anything, I expect an acceleration in FUSO’s silica sales over the next few quarters. Add to that 5-10% price hikes + moderating depreciation charges from the recently launched silica reactors, and you get 25%+ annual OP growth in the next three years. I think the company's valuation today leaves about 30-40% upside.

I think Okano Valve Manufacturing (6492.JP) is an exceptional investment opportunity. It is the cleanest and cheapest way to invest in the nuclear renaissance I have found. I’m grateful to @JimJimmeny11 for the idea. Okano is a highly specialized manufacturer of high-temperature, high-pressure valves for nuclear and thermal power plants, with almost 50% of revenue from maintenance services. It holds a near-monopoly position in the Japanese market, protected by decades of operating experience and impassable qualification barriers. Okano is entering a multi-year earnings upcycle as Japanese utilities recommission reactors and replace aging valve infrastructure, driving revenue growth off a depressed base with significant operating leverage. I expect the recent disruptions to energy supply will break remaining political resistance to nuclear restarts and compress the launch timeline. Despite strong demand tailwinds, the firm’s valuation doesn’t price much growth. On my FY2026 numbers, P/E is 13, and EV/EBITDA is 8. Investors evidently do not consider international expansion or next-generation reactors as valuable options. Like Togami Electric Manufacturing (6643.JP), Okano’s risk/reward ratio is heavily skewed today, so I made a large investment last week under ¥10,000 without overthinking it.

At least Okano Valve ($6492.T) didn’t drop after Q3 results. The 9M numbers look good: 90% progress on revenue guide and 134% on OP (FY shift impacts YoY comp). There will be a seasonal slowdown in Q4, but I don’t see OP going negative. Last quarter’s OP margin hit 27%, driven by maintenance work at Onagawa No. 2, Kashiwazaki-Kariwa No. 7, and Fukushima Daiichi. New valve sales included Kashiwazaki-Kariwa, Tokai No. 2, and Fukushima Daiichi. Overall, the nuclear thesis remains on track. Okano remains undervalued, so I’m holding.

We built high-throughput materials labs in Menlo Park to create a loop between experiments and models. The labs generate fresh data, the models learn from it, and then help us decide what to try next. Using only 1,300 H200s, plus months of our experimental data, we mid-trained and RL’d an open-source model to surpass GPT-6 Astra on our analysis benchmark. We call it Neon. This is real footage from our lab. We’re focusing first on hard problems in materials science, including superconductors, magnets, and semiconductor materials. Read our blog posts below.

Very bullish read-through for Santec $6777.T from its major client, Fabrinet $FN. Bottom line: optical testing tools are taking center stage. On the earnings call yesterday, Fabrinet’s CEO noted that as NPO scales to 6.4, 12.8 TB, and beyond, manufacturing complexity and yield become increasingly important. He also outlined plans to double production capacity within 3 to 4 years. So, increased capex combined with higher test intensity = durable demand for Santec’s equipment.

JP companies like Enplas that lead in their niches (TYO 6961) are stupid cheap sometimes. 55% operating margin in this segment. Almost a net-net too! Wild. Competitors use glass which is much more expensive with no real benefits in datacenter use cases.

$255A.T GLTechno had arguably the best quarter among my holdings, yet the stock fell 4% after the results came out last Friday and finished down 11%. The semiconductor business is firing on all cylinders: sales are accelerating, margins are expanding, and orders are booming. The order backlog is up 130% YoY and 51% QoQ, with a book-to-bill ratio at 1.7. Applied Materials $AMAT is placing massive orders for GLTechno’s quartz glass components to service installed equipment and build new tools. In turn, GLTechno plans to double its flame-processing capacity in Japan starting in January 2027. As a bonus, the analytical instruments business is also performing well, fueled by robust sales of water-testing consumables. At 12 P/E, I think GLTechno has the highest risk-adjusted upside across my AI-related positions and buy candidates. It deserves a higher portfolio weight.

Bartolomé Bermejo, Descent of Christ into Limbo (detail), c. 1475

Short update on my SpaceX bags in Korea – Sphere Corp ($347700.KQ) and HVM ($295310.KQ). Sphere went on an investor roadshow last week and confirmed a few things. First, it almost certainly has a 10-year supply deal with SpaceX, but can’t name the client due to an NDA. Second, HVM is in fact Sphere’s qualified supplier of specialty alloys. Third, space orders are real and accelerating. Finally, Sphere is considering three new distribution centers near Los Angeles, Texas, and Florida, consistent with SpaceX’s operating footprint. On the call yesterday, SpaceX CFO Bret Johnsen said: “We continue to make progress building the infrastructure required to support thousands of Starship launches per year, including accelerating Raptor and launch vehicle production, the build-out of our Gigabays and making significant progress towards activating multiple launch pads at Starbase and in Cape Canaveral at Pad 39A and Pad 37.” In short, Elon needs an absurd amount of metal, and the Koreans are working triple shifts to deliver it.

Santec Holdings Corporation is pleased to announce the acquisition of HM Solutions (HMS), expanding Santec’s capabilities in fiber optic cable assembly production test and measurement.