With that aside, let’s get into it.

Executive Summary

Water Oasis Limited (1161-HK) is a beauty service provider based out of Hong Kong whose stock is down materially (~45%) from the ’21 peak given an industry wide downturn driven by macroeconomic conditions and tourism flight. Despite the headwinds, the firm’s sales have continued to grow while earnings have risen from a beneficial mix-shift and footprint rationing and higher utilization. The stock trades at an NTM P/E of ~5.5x and holds ~900mm of cash in HKD on an $850mm market cap with no debt. My base case sees the stock re-rate to 8x (in line with comps) as revenue continues to grow and economics improve through modest M&A, with an upside of ~70%. My probability adjusted upside between base/bull/bear and a privatization scenario is ~50%.

Industry

Let’s start with the industry, it’s not pretty, and it frames the entire opportunity.

Post covid, China went into a property downturn given overinvestment and high interest rates to deal with inflation. That led to a weak consumption environment, which led to weak local discretionary spending. Total HK retail sales fell 7.3% in 2024. From management “At the end of March 2026 the Hong Kong market situation remained much the same as it was six months earlier [...] Many players in the beauty industry struggled in this environment, and some minor players exited the market.”

There was also spending that rotated to Shenzhen (the “northbound effect” as management calls it). Consumers have been travelling north to shop in lower-cost Shenzhen, which benefits from the weaker RMB vs HKD (which is pegged to the USD). Two of Water Oasis’ comps, Perfect Medical and EC Healthcare, specify that customers are going to Shenzhen and South Korea specifically for cheaper aesthetic treatments. This was the impact on industry incumbents:

To give you an idea of how serious this effect was/is, here are a few indicators that point to the magnitude of the impact:

Like I said, it’s not pretty. Real spending per visitor has gone down in real terms, and even overnight visitors (i.e. the group that spends money on things like beauty treatments) has also declined. The indicator on the right is the most telling, this basically shows the net inflows / outflows between Hong Kong and Shenzhen. Well, as it stands, the number of tourists that travel from Hong Kong to Shenzhen is 3x the number of tourists that travel the other way around. That’s a serious net outflow from Hong Kong.

Despite the industry turmoil, Water Oasis has continued to perform; group revenue has grown at a CAGR of 4% while services revenue has grown closer to 10%. This is partly due to its ability to take market share from struggling competitors:

EC Healthcare’s beauty segment and Water Oasis have both continued to grab share in this environment (note: industry sizing is partly based off my own estimates). This seems to line up with reviews they’ve received from online forums / google maps. From the table below, we can see the service quality that both EC Healthcare and Water Oasis provide stands out vs Perfect Medical.

Water Oasis’ management explains it as such: “our scale, long history and readiness to invest in the future all give customers great confidence in our brands and in our pre-payment business model.” That, coupled with celebrity marketing, brand-cluster flagships, and paying a staff premium, helps explain the share gains.

There’s also some good news on the macro front. Domestically, financial conditions seem like they’re turning around. I built a financial conditions index (see components below), to track the relationship between the FCI and services-PCE. Excluding the COVID and re-opening era in 2023-2025 where FCI diverged from services-PCE growth, the relationship seems pretty clear cut.

And the relationship still holds if we look specifically at HK personal care spending data, which is what Water Oasis specializes in.

Many of the FCI’s components are turning positive, and if the trend continues, we could see an uplift in domestic demand. The main drivers here are improving property prices (positive wealth effect), a weakening HKD, and an expansion in bank credit.

So, on one side, we see a structural headwind as tourism outflows continue moving in favor of Shenzhen. An environment in which Water Oasis has continued to grab market share. On the other side, we see improving financial conditions which seem to be driving improved services spending. So, although the situation isn’t ideal, I’m leaning towards continued organic growth for incumbents that are able to grab share. There is another reason I’m cautiously optimistic, and that’s inorganic growth, but more on that later.

Business Overview

The business runs an extensive portfolio of beauty, aesthetic-medical, and related wellness services, plus a smaller skincare-product retail/distribution arm. Operations sit in Hong Kong, Macau, and Mainland China. In June 2021, the Company acquired Millistrong Group, which was a material transaction that reshaped the company, adding beauty-service brands (spas, a medical beauty outlet) and product brands to the portfolio. Here’s a useful graphic that gives you an overview of services they offer:

The company reports exactly two operating segments (from the segment notes): (i) Product - sales of skincare products; (ii) Service - provision of treatment services in beauty salons, spas, and medical beauty centers. The business has swung steadily toward services; services were ~90% of FY2025 revenue and an even larger share of profit. Services carry higher margins.

Importantly, service revenue is almost entirely prepaid. Customers buy treatment packages up front; the cash received is booked as a contract liability and recognized as revenue over time as treatments are delivered. This creates a float (contract liabilities) of ~HK$752m that turns over roughly once a year, alongside ~HK$917m of cash. It also means they have a deeply negative cash conversion cycle (which is great!):

Although Water Oasis doesn’t report it themselves, we know from comps that the model is structurally repeat-oriented (customers buy multi-session packages up front); EC Healthcare reports a 66-69% repurchase rate.

Economics

Margins have partly been improving from the sales mix. As per the below chart, product margins collapsed because Water Oasis was re-selling skincare products, and demand steadily fell as customers started ordering online where products were cheaper. There were fixed costs allocated to product sales (ex. rent) and so the margin tanked. In any event, it’s a small contributor to the business now, not overly relevant. Service margins on the other hand have continued to steadily rise.

The key drivers of margin expansion have been:

• Footprint optimization – management re-negotiated rents in a soft leasing market, have increased footprint efficiency (more operations per location) and closed 14 underperforming stores from 2023 – 2024.

• Depreciation – capex went from 10% of revenue in FY22 to 1.5% in FY25.

• Operating Leverage – service revenue grew against a declining rental base and lower depreciation amount.

• The offset is that staff costs have been stable as a % of revenue; Water Oasis pays heavily on a commission basis, which scale with sales.

Digging into rent, management closed ~14 underperforming stores in 2023 & 2024. The Company also increased the number of operations per location. For example, they’ve added Oasis Hair Spas and InMedic centers inside existing beauty center locations. We can see the co-location intensity (total operations / physical locations) has been marching upwards.

The improved footprint, coupled with the increasing number of total operations per location, mean their rent as a % of sales has gone down considerably (note H1 2026 is annualized rent per location).

Management wants to keep focusing on co-location “thus avoiding any increases in rental costs”; we could continue to see higher efficiency on existing leases, leading to margin improvement.

On the downside, the Company is due for renewed maintenance capex as per its depreciation / replacement cycle. Management is guiding to higher 2026 capex as new renovations and equipment purchases are conducted. Depreciation should pick back up in tandem on the P&L.

Given the rental efficiency and D&A are moving in opposite directions, I’m modeling no margin improvement in my base case, but we could see some further out (2028, 2029).

On the revenue side, growth in recent years has come almost entirely from same-store sales (SSS) rather than footprint. The beauty center count has held at 55–57 across Hong Kong and Macau ever since the Millistrong acquisition stepped it up in 2021. This was flat by design, in line with management’s co-location strategy. Management doesn’t disclose an SSS figure, but one can be inferred from the period-end center counts in the half-year filings. On the latest read, implied SSS accelerated to ~4.6% in H1 FY2026, and this is arguably a floor, since the two “new” net centers are co-located units occupying no additional premises. If we treat those as existing capacity, the entire +8% revenue growth is comp. Also an important side note, beauty centers aren’t the same thing as physical locations (which is why the numbers differ vs my above chart), but they’re effectively the category driving service revenue, which means they’re the best KPI to drive SSS.

For context on the chart: FY2020 collapsed under mandatory closures, border shutdowns and the tail of the 2019 protests; FY2021 rebounded on relaxed restrictions and government consumption vouchers; FY2022 was hit by the Omicron fifth wave and its 104 days of forced closures; and the deeply negative FY2022 SSS bar is further distorted by the acquisition itself, which added 15 centers to the denominator mid-crisis (stripping Millistrong, legacy revenue fell roughly 11%). FY2023 then delivered the sharp reopening recovery as the mainland border reopened. From ~2024, the “northbound consumption” headwind (Hong Kong residents spending in Shenzhen) coincides with comps cooling to low single digits. The takeaway: through the worst consumer backdrop Hong Kong has faced in a decade, comps never went negative post-reopening, and they’re re-accelerating into FY2026 on an unchanged store base.

The Float

Now if you just stop and look at that enormous cash balance, you might wonder why this thing could be priced so attractively. It’s a growing business with cash worth more than the market cap, can’t management just distribute it to shareholders? Well, yes, but also not quite.

The reason being that the float isn’t free money; it needs to be matched against its related contract liabilities. It’s effectively deferred revenue. That said, the costs to service leaves Water Oasis with about a 19% margin on that revenue. This means that with HK$752mm of contract liabilities, around HK$144mm of that will be left over as cash for the firm. with HK$917mm in existing cash, that means the “owner accessible cash” is really ~HK$309mm. That said, the management could theoretically take its entire cash balance and use it, as service costs can be funded through incremental additions to the float (i.e. more customer prepayments). If management did treat the full cash balance as available cash, it would effectively be taking on leverage, since contract liabilities need to be matched, but the cash is theoretically at management’s disposal. In any event, the real capital structure looks more like this:

So, what can it do with this cash? Well, it has 3 main options really; organic growth, shareholder initiatives (Buybacks + Dividends), and M&A. The company cut its dividend payout ratio in FY 2024, and later cited that it wanted to prioritize M&A in FY 2025. Its payout ratio went from >80% in FY2023 to ~50% in FY2025. Although dividends will continue to be paid out, I don’t envision them increasing the payout ratio significantly from here, as they explicitly cut it to leave room for M&A. The same goes for buybacks, as Water Oasis hasn’t historically done many. That leaves organic growth and M&A.

M&A

We don’t have too many precedents to work with, but, the good news is most of the M&A in the beauty services space have worked out. Here’s a summary:

The most successful deals were direct comps to Water Oasis, targets they could realistically acquire. These were done at disciplined prices. If we contrast these to similar businesses, ones that depend on doctors, we see the deals have not performed nearly as well. The aesthetic clinics don’t carry the same degree of key man risk as medical clinics, and so integration becomes easier.

Water Oasis conducted one of the better deals; they bought Millistrong from a desperate seller, and they acquired it at ~0.30x EV/Sales. The skincare distribution part of the business died a few years later, so a write-off took place. If the brand’s margin was anything close to Water Oasis’, then the deal paid for itself in ~3 years regardless. That said however, since there’s not a lot of precedent surrounding their execution ability, we should apply a discount related to the turnout of potential M&A.

Since the industry has taken a serious hit, let’s assume they remain price disciplined and can purchase distressed players at 6x LTM P/E. On ~300mm of accessible cash (float adjusted), they can generate an incremental 50mm in net income. Water Oasis’ adjusted net margin is ~14% (H1 FY2026), and it’s quite likely they’ll acquire firms with margins below that, but for the sake of conservatism I’m not assuming any margin improvement in my base case. I’m modeling 50% available cash deployed by year-end FY 2027 with no margin improvement from targets.

Catalysts

I see 3 possible catalysts; a take-private (MBO), M&A execution, and a recovering industry.

Privatization

It seems like the family might be considering a take private, the structure would suggest it:

Aging controllers (~70% family, three founders 70–75, the largest holder their similarly-aged sibling), a next-gen CEO already installed (Tam Siu Kei, 48), a debt-free, cash-rich business trading cheap, zero buybacks, and a suppressed payout that keeps the share price low.

However, under Rule 2.10 of the HK Takeovers Code, a privatization scheme needs (i) approval by ≥75% of the disinterested shares voted, and (ii) votes against ≤10% of all disinterested shares. “Disinterested shares” explicitly excludes the offeror and its concert parties, so the family’s ~70% doesn’t count. This means that over 75% of the 30% disinterested shares (common shareholders) must vote “yes” while 10% of that same 30% float could kill the deal if they vote “no”. So, to ensure 3% of the total ownership of the company don’t vote against a deal, the family would need to offer a hefty premium.

If the family continued to acquire shares, the price paid would be nominally lower for them in the event of an MBO. The float composition would also skew in their favor as the most willing sellers would be exiting their positions.

That’s exactly what they’re doing:

M&A Execution

M&A has the potential to be a significant growth engine. If the company can execute and prove its ability to maintain margins as it does M&A, it can acquire aesthetic clinics at depressed valuations given the downturn and hopefully drive accretive acquisitions. The base business has proven its ability to continue grabbing market share organically, so I see this as an additional growth vector that could lead to a re-rating.

Industry Recovery

I’ve laid out my argument for why I think there’s a chance the industry could be turning, but I wouldn’t attribute a high probability to this just yet. I’m tracking the macro-economic indicators I showcased before as they would be clear leading indicators of a turnaround. I’ll keep updates posted on the Substack.

Upside

Whichever way you look at this, the stock looks cheap (note these multiples are without M&A contribution and with a float adjusted cash flow, which inflates the EV mutliples).

4 scenarios; base – bull – bear – MBO

• Base - 50%: reasonably well executed M&A with minor net income contribution + continued organic growth (5%) with no meaningful tailwind from financial conditions. Margins remain flat, and the multiple re-rates to 8x LTM earnings in line with comps; Perfect Medical and Beauty Farm trade at ~10x LTM P/E.

• Bull - 10%: optimistic M&A execution + higher organic growth from industry recovery (7%); margins grow on leverage and multiple re-rates to 12x.

• Bear - 20%: financial conditions don’t recover, no M&A modeled, northbound effect continues to erode industry, revenue contracts 5% per year, P/E multiple stays depressed at 5.5x and net income erodes on operating leverage.

• MBO – 20%: I think there’s a fair chance they take the company private, in which case we could realistically see an upside anywhere from 30-50%. Let’s take 40% as the midpoint.

Overall, a solid risk/reward. There are real risks, I don’t want to make it seem like this can’t lose money as they are operating in an industry with real headwinds, but I see the odds skewed in their favor for being able to capitalize on these headwinds rather than falling victim to them.

Executive Summary: The US has too much debt to realistically pay back. Because it prints its own currency, it won’t default, it will inflate the debt away (debasement). The cleanest way to bet on this is to own gold.

The real problem isn’t the size of the debt but whether the government can afford the interest on it. Tax revenue can’t keep pace, and no politician will cut spending or raise taxes to close the gap, so inflating the debt away becomes the only viable path.

To inflate without letting interest rates spike, the administration is engineering “financial repression” through several levers: pushing rates lower via a captured Fed, forcing banks and stablecoins to buy government debt, and putting the infrastructure in place to mask the true inflation rate. Gold is the one asset the government can neither print nor easily confiscate.

Gold tracks the dollar far more than it tracks interest rates, and both a weakening dollar and heavy central-bank buying point higher. Near-term, sentiment is poor and the Fed is still hawkish, so the trade needs a catalyst; clearer Fed action and a confirmed end to the Iran war. The two biggest risks are sticky inflation and that war reigniting.

This report will be split into 2 parts; Part 1 will focus solely on the macro and Part 2 I’ll outline ways to express the trade, primarily through gold miners.

Revisiting the Debasement Trade

Here’s a quick recap for those who aren’t familiar with the debasement trade that took off in 2025. The debasement trade is centered around the idea of monetary debasement through the risk of a sovereign debt crisis.

Typically, when debt/GDP of economies approach a certain level, debt servicing becomes unsustainable and the country goes into a sovereign debt crisis. However, if your country can print its own currency, it avoids a sovereign debt crisis and instead faces currency debasement.

Here we can see both types of countries, and how their respective crises were resolved.

If you’re predominantly an equity investor and reading this, think of currency as equity and sovereign debt as, well, debt. If you default on your debt, you need to restructure and potentially liquidate your assets. To avoid this, you issue equity to save your company. Naturally, the value of your equity is going to fall as it gets diluted. The mechanics are similar.

Based on research from Moody’s and the CBO, this is the path for the US Debt/GDP:

This doesn’t seem all that bad compared to the situations mentioned above. We could have more than a decade before Debt/GDP starts to be flashing red, and even more time if Bessent’s “grow out of it” plan works out (more on this later).

But, the Debt/GDP only tells part of the story, what causes countries to default isn’t the level of debt, but the serviceability of said debt. Countries pay off debt through 2 ways, directly through tax collections, and by issuing new debt (called “monetizing” the deficit).

On the first method, tax collections, the US government collects tax revenue which typically amounts to ~18% of GDP. If you’re an equity investor, you can think of a country’s interest coverage ratio as its interest payments / total tax revenue.

Historically, when a country gets to 30%, it enters the crisis zone.

But once again, these countries couldn’t print their own currencies. And countries that haven’t typically reached the 30% interest / tax revenue figure. This is because 30% would lead to a default, and to prevent that, the country will debase its currency before that happens.

Here we can see the US could get to the crisis zone by as early as mid-2030. Moody’s forecast gets it there by 2040, which would imply the US has no choice but to debase its currency.

Well, what about the reasonable option of paying off the debt? No politician wants to do this, period. It would be political suicide. Scott Bessent has repeatedly advocated the only way to solve the US’ debt problem is to grow their way out of it.

“This election cycle is the last chance for the U.S. to grow our way out of this mountain of debt without becoming a sort of European-style socialist democracy.” – Scott Bessent

“What’s important is that the economy grows faster than the debt… If we change the growth trajectory of the country, of the economy, then we will stabilize our finances and grow our way out of this.” – Scott Bessent

The stated plan is to grow out in real terms, but with r > g (rates larger than growth) and no primary surplus on the table, the only achievable version is nominal. (i.e. inflating the denominator). Below are some examples of economies that debased their currencies:

The commonality here is that real interest rates were negative half the time through 1945 – 1980. All these countries were holding their interest / tax revenue ratios *down* while they were doing this, since interest payments are fixed on nominal bond amounts, while tax collections are a % on nominal GDP. One stays largely unchanged in an inflationary environment (interest) and one rises (tax collections).

But this only works if rates don’t rise in tandem. If yields on new bond issues rise, then interest costs rise with them, and the ratio continues to climb. Enter financial repression.

4 Levers to Manufacture Financial Repression

How do you engineer a decline in real rates while inflating your economy and not letting yields rise? You need to be creative. You need a few levers:

1. Lower rates

The nomination of Warsh as the fed chair was aimed at keeping rates lower. Trump spent 2 years attacking Powell for insufficient cuts and threatening to fire him, then publicly claimed on NBC News that he would not have nominated Warsh unless he thought Warsh would lower rates quickly. The conclusion is Warsh was brought in to cut rates, opposing arguments to this view are weak in my opinion.

That said, it’s important to note the fed chair doesn’t have absolute power, and as it stands the FOMC members are skewed hawkish. The direction for rates is still down, but over the next few years.

In the near term, whether Warsh will be able to ultimately cut is dependent on what inflation looks like, and it does look like services could be sticky (more on this below). However, with news that the Iran conflict might finally be turning, we might see headline come down as oil prices fall. I’ll detail my thoughts on the Iranian conflict in the risk section, as this is the biggest wildcard in my view.

Beyond inflation, what matters is the current administration’s pressure on the Fed, and Warsh’s architecture for cutting rates. We have evidence for Fed captivity, and it’s been taking hold:

• The blueprint (Miran’s writings). Miran published a plan to weaken Fed independence; more presidential oversight, shorter governor terms, killing the revolving-door insulation. He swore the opposite at his confirmation, then pushed for aggressive cuts the moment he was seated.

• Personnel capture. Personnel capture. The seven governors are permanent FOMC votes, so the board is the prize. Trump already holds four seats: Waller, Bowman, Miran, and his Chair, Warsh. This is a working majority; Cook’s removal would make it five, leaving only Powell (who kept his seat after stepping down as Chair) and Jefferson outside his column. Trump said it himself: “majority very shortly.” No need to pressure votes when you can replace them.

• Removing a Board Governor (Cook). First attempt to fire a governor in 112 years, on a manufactured pretext that surfaced after Trump’s rate complaints. If it sticks, and Cook is fired, the Fed will be largely captive, and the administration will be able to steer the rate path with more ease.

As for Warsh’s architecture, he’s laid it out at the latest fed meeting. He’s assembling a “task force” to overhaul the Fed’s definition of inflation, change the sources of data used to compute inflation, and drive down the balance sheet while keeping rates lower (more on this below).

2. Manufacture captive buyers

If you regulate institutions into holding government debt, demand will hold down the rate for you. During WWII, we saw capped deposit rates so savers couldn’t escape to higher yields, high reserve requirements, and rules steering banks, pensions, and insurers into sovereign bonds. Today, we’re seeing a modern equivalent. Warsh wants to shrink the Fed’s balance sheet while pushing that supply of treasuries onto the private sector.

They’re doing this through a few channels:

• Channel 1: Free up bank balance sheets. The “Basel III Endgame” regulations are changing the way risk-weighted assets are calculated; a planned ~16–19% increase in big-bank capital requirements has flipped into a roughly ~5% cut. Less required capital per dollar of assets means more room to lend. This is inflationary; expanded credit capacity allows banks to create more loans. Provided the demand is present, these new loans will fall directly into the hands of consumers / businesses and create inflation.

• Channel 2: Unbind Treasury holdings specifically. From Channel 1, treasuries are basically applied a risk weighting of 0, banks aren’t really limited to hold treasuries based ofon how much capital (equity) they have. The bottleneck to holding treasuries is the leverage cap (eSLR), the ratio that measures capital against total bank assets. The eSLR for the biggest banks’ depository units is dropping from a flat 6% to ~3.5–4.5%. This removes the eSLR as the binding constraint, restoring banks’ incentive to hold and intermediate more treasuries.

• Channel 3: Create a brand-new buyer. The GENIUS Act requires stablecoins to be fully backed by cash and short-term T-bills, and the market is projected to grow from ~$300B today to $2–3T by 2028–2030. Every new stablecoin dollar is effectively a forced T-bill purchase; an estimated ~$1T of fresh demand for government debt.

As a recap, channel 1 allows banks to hold more assets per $ of capital; this mainly acts as a lending channel and is inflationary. Channel 2 allows banks to take on more leverage and reduces the asset bottleneck, allowing banks to buy more treasuries. Channel 3 redirects existing dollars into forced T-bill demand from stablecoins purchases. Channels 2 and 3 drive yields lower. The net effect of all 3 channels is inflation + lower yields, exactly the set-up financial repression requires.

3. Masked inflation

I can’t claim I know with certainty where true inflation sits, but I don’t need to. The defensible claim is about the policy reaction function. Here is what we do know:

• BLS has been cutting CPI data collection for budget/staffing reasons

• The share of estimated prices has exploded: roughly 19% of CPI prices are now imputed, up from 5.1% at the end of 2022

• A concrete, documented case of imputation biasing inflation down: for the October 2025 housing component, BLS used a carry-forward imputation method that reduced the rent and OER indexes and their 12-month changes

• Institutional independence is under visible pressure: after large downward revisions in the July 2025 jobs report, Trump dismissed the BLS director and accused her of manipulating data for political gain without evidence, and the Labor Department’s Inspector General launched a review of CPI, PPI, and employment data.

Importantly, as of the most recent Fed meeting with Warsh as chairman, he states the Fed’s plan to put together “task forces”. These task forces include the following: a group that seeks methodological changes in data gathering for inflation measures, a team that seeks to review the inflation framework (including how the target is defined), and a team that aims to review how AI will transform productivity (and in my opinion support a disinflation narrative).

So, there is clear pressure, degraded data quality, a fired BLS head, and a Fed chair implementing a taskforce that seems designed to keep both perceived inflation and nominal rates suppressed. Whether or not the CPI print is “accurate,” that policy bias, combined with deficits that bias inflation higher over time, means the system is built to deliver falling and likely negative real yields. This stays true for as long as the repression machinery can hold nominal yields below where a free market would set them.

The important implication here is inflation stays sticky, and you can make an argument around this. Some sub-components of PCE services’ inflation are ramping, while leading indicators for wage inflation are also on the rise.

4. Closing the escape hatches

In and after WWII, the government didn’t just cap interest rates, and create captive buyers, it found a way to “seal the exits” so to say. The 1933 gold confiscation had already made it illegal for Americans to own monetary gold (a ban that ran until 1974), and Bretton Woods–era capital controls made moving savings abroad difficult. Trapped in domestic bonds and bank deposits paying below-inflation returns, savers had nowhere to hide while the government quietly inflated its debt away on their backs. That playbook can’t be rerun today: the US runs an open capital account, and Americans can freely hold gold, foreign currencies, and even crypto. Banning any of these would be politically explosive and, in a digital world, very difficult to enforce. This is exactly why this is gold-bullish; repression only works if savers can’t escape, so with the exits wide open and the other levers (capped rates, captive buyers, inflation) driving real yields negative, capital flows straight to the one asset the state can neither print nor confiscate. The pressure the 1940s bottled up inside the bond market now leaks directly into the gold bid. It’s the pressure release valve.

Gold as a Debasement Hedge

By this point, hopefully I’ve made the case that secular monetary debasement is a real risk, and that the current administration is actively planning around this debasement strategically.

Now onto the actual asset, why gold, why not just short government bonds or buy crypto?

For government bonds, QE is a real risk, despite Warsh’s claims to want to shrink the balance sheet. If I’m right, inflation will chip away at the real yield of government bonds, but having an active buyer with infinite demand nonetheless makes it difficult to be short, or at the very least caps the upside of the trade.

As for crypto, we’ve seen the way bitcoin reacted last year, which stood in stark contrast to Gold’s performance. I’m not a crypto expert, and I don’t have a strong view. I love the idea of decentralized currency that exists outside of the purview of Sovereigns, but from my understanding of human nature, the powerful seek control above all else. Sovereigns are more organized than the masses and have significant resources at their disposal. An organized minority will always beat out a disorganized majority, in my opinion. This makes it difficult for crypto to become a reserve asset, but I’ll leave my argument there.

What I can with more confidence, is that the characteristics of gold make it attractive as a reserve asset. For one, it’s physical. It can’t be hacked and the only way to steal it is through military force, the same thing that ultimately backs fiat. Sovereigns have a monopoly on domestic violence, and so preservation of the asset is all but guaranteed outside of the context of war.

Second, supply has remained highly inelastic despite price appreciation over time. The value of the asset cannot be easily inflated, and unlike fiat, it most certainly cannot be printed.

We’ve established 2 things. Monetary debasement is real, and gold is a good hedge against debasement. Let’s study what really drives the price of gold. If we look at gold throughout different regimes, we can get a feel for it.

Real yields don’t really seem to have a direct cause and effect relationship. The dollar on the other hand does. The chart below makes the relationship clearer.

We can see that gold has increasingly turned into a dollar story from the 2000s onward. A weak dollar means higher gold.

How do we get a weaker dollar? We’ve laid out the case for lower rates, and now with a potential resolution to the Iran conflict, we could see the Strait of Hormuz open and oil prices alleviate. It seems to be what we’re seeing in the price action for crude.

Lower oil reduces the risk of an inflation re-acceleration, which eases the pressure for further Fed tightening and tilts the path toward cuts; a cutting Fed means a softer dollar. An end to the conflict likewise strips a geopolitical bid out of the dollar. Both channels resolve to the same thing: a weaker dollar, which is what carries gold in this regime.

Through this cycle gold has decoupled from real yields. Gold has risen even as real yields climbed after 2022, so the 10-year TIPS yield enters the regression weak and wrong-signed, and it is in any case highly collinear with the dollar, which dominates the explained variance. Rather than lean on an unstable, double-counted rate term, I let the dollar absorb the combined rate-and-currency channel. Overall gold is trading as a dollar/debasement hedge.

My dollar forecast here, and the slight weakness we can expect in my house and bull cases, are based off falling rates and debt monetization. As I mentioned earlier, even if Warsh’s plan is to monetize deficits through the private sector, that remains inflationary (and arguably more than directly through the Fed’s balance sheet). This directly contributes to money supply.

With the macro drivers out of the way, let’s turn to central banks. One thing to keep in mind is that there’s no contemporaneous volume/price correlation, meaning central bank buying does not immediately lead to higher gold prices in the same period.

However, price-inelastic buying removes float and shifts the regime over time. We are in a central bank buying regime, and central bank purchasing decisions affect institutional purchases. CBs are effectively responsible for setting the trend, but they do not lead institutional buying. They are the price insensitive buyers that just keep buying. We can see from this chart, while institutions buy and sell, CBs have just kept buying since ~2008.

To visualize this trend better, we can derive a premium by locking in regression coefficients in the net-selling era (1990 – 2000). This isn’t a perfect analysis, but it gives you a sense of how traditional macro drivers in the net selling era do not fully explain the move in the 2010s to today. Part of this is explained by the increasing importance of the US dollar, and part of it, in my view, is driven by CB purchases and institutional follow on.

What are CBs saying now? The latest survey from the World Gold Council provides bullish data:

We can see from these charts, the trend in buying is expected to continue. Interestingly, the first 2 charts show an expectation in continuing a shift from USD reserves to gold reserves. Charts 3 and 4 are simply more directional confirmation that gold buying will increase.

The same structural tailwinds for Gold are still in place as far as central banks are concerned, the data confirms it. It does seem like peak bullishness from CBs is likely behind us however (95% expected an increase in 2025, down to 89% now), but the trend remains very positive.

The Set-Up

In the short term, we have some data that shows Gold’s sentiment is at its worst in the past decade. This is positive from a contrarian’s point of view.

However, as we showed on the dot-plot, the Fed seems to be relatively hawkish in the very short term.

The Iran war, however, appears to be coming to its conclusion. If the strait re-opens, this will be extremely positive for inflation. It seems a surplus was present going into the conflict, and the UAE left OPEC (effective May 1, 2026) specifically to maximize production. We could potentially see headline CPI falling, and rate expectations drifting alongside it.

However, this set-up makes it difficult for Gold to do well without a meaningful catalyst.

Catalysts

2 things need to happen for gold to play out correctly in my view.

For one, Warsh’s plan needs to become clearer, and we need to see the mobilization of Warsh’s taskforces. There will be clues that Warsh’s plan in fact designed to manipulate data and inflation methodology so that the Fed can build a compelling argument to cut rates.

Secondly, the end of the Iran war needs to be confirmed, and we need to see flows out of the Strait of Hormuz normalize. This is important since falling inflation will lead to a weaker dollar, which we have shown strengthens the case for gold.

A tertiary catalyst that plays into the Fed rate cut thesis is continued captivity of the Fed by the current administration. This means Lisa Cook’s termination is a key catalyst, and we could see this happen by the end of June 2026. Although, I wouldn’t assign a significant probability to this.

Macro Risks

The biggest macro risks in my view are twofold: a resumption of the Iran war and a pick-up in core inflation.

As I argued, core inflation, particularly services, could be sticky. The labor market has been more resilient than I would have expected, and that could continue to be inflationary.

Secondly, and the larger risk in my view, is that the current peace plan is another red herring designed to quell oil market pressure. Benjamin Netanyahu has put his political career on the line betting on the Greater Israel Project, and folding to the US’ demands is in my view unlikely. It is unclear how much influence Israel has over the current administration’s policies, but we have seen them enter a war that has so far seemingly undermined Scott Bessent and Stephen Miran’s plan to manufacture financial repression and stimulate economic growth. This is the most significant risk in my view, but we’ll gain clarity on the outcome of the deal in the next few weeks.

Miraial (4238.JP) is a semi wafer container manufacturer which benefits from continued demand of advanced wafers used for AI and the bottoming analog semi cycle. Leading indicators heavily suggest demand across analog semis is picking up. Miraial is trading at a p/e of ~8.5x on my FY 2028 numbers, whereas comps trade above 15x. The downside mitigation is high; the P/B is ~0.5x with cash worth 35% of the market cap. Capacity utilization is quite low, and assuming revenue grows in line with leading indicators, margins would expand and the upside on the stock could be ~55%+ at 12x P/E with limited downside within the next 2 years. As a Japanese microcap, the stock has very little analyst coverage and sits in an area of the semi value chain that is under covered. The stock is up ~15% in the days leading up to this report.

Thanks for reading! Subscribe for free to receive new posts and support my work.

Products

4238 produces FOUPs and FOSBs; which are used for the transportation of semiconductor wafers between and inside semi foundries. FOSBs carry wafers between foundries, whereas FOUPs carry wafers inside a foundry to be processed at different parts of the foundry. Collectively, FOUPs and FOSBs (wafer transport containers) account for ~70% of Miraial’s sales, largely skewed towards FOSBs. The remaining revenue consists of plastic resins and molding machines.

FOSB

FOUP

These are inferred ASP and gross margins. The unit economics are clearly more attractive for FOUPs given utility of the product and higher ASP. That said, Miraial has structurally higher exposure to FOSBs with secondary exposure to FOUPs. Note the ASPs in the table are estimates; used FOSB/FOUP data exists on ebay but brand new sku prices are not disclosed.

FOSBs

FOSB (Front Opening Shipping Box) is the inter-facility container. Wafer carriers used for transfer from the wafer manufacturer to the semiconductor manufacturer are called FOSBs. For example, the moment a wafer manufacturer like Sumco, Miraial’s biggest customer, sells its wafer to a fab, it will ship it in a FOSB.

Miraial is vertically integrated, and they own the majority of the FOSB manufacturing process. For one, it produces its own resins, which are used in the formulation of FOSBs. The resin needs to be designed in line with the required features of FOSBs: electrostatic dissipation, dimensional stability, and cleanliness. They also design and fabricate their own molds. The mold is a critical piece of IP in precision plastic injection; it encodes decades of trial-and-error learning about shrinkage, warpage, and dimensional control. Miraial owns these molds completely. They also do their own precision injecting, with equipment ranging from small injection molding machines with a clamping force of 30 tonnes to large 3,000-tonne injection molding machines. A 300mm FOSB has little tolerance for error, so these are also important. Some of this molding machinery is also created through a subsidiary they own called Sanjo Seiki. They effectively own the entire FOSB creation process, which creates high barriers to entry and keeps customers locked in.

The demand for FOSBs is driven by wafer volumes, specifically 300mm wafers. The mature-node segment, which includes analog, power, and automotive chips, shipped on both 300mm and 200mm wafers, has been the most depressed part of the cycle and has the most recovery potential. This means a recovery in FOSB volumes is heavily tied to the analog and mature-node recovery. Well, the good news is that virtually all leading indicators are pointing towards a recovery in analog wafer demand.

In 2021, analog semi demand was pulled forward since people were working from home and ordering high amounts of consumer electronics products. The cycle peaked in 2022 and seems to have bottom in the past year. We can prove this by looking at Miraial’s customers’ customers. That is 2 steps up the value chain; if their sales are higher, that means they’re drawing down more inventory, which implies demand will start to flow to wafer manufacturers and ultimately to Miraial.

We can see here guidance points towards higher analog sales from major producers into 2026, with the bottom around Q1 2025.

A few other indicators, Vishay (VSH), which produces analog semis, has seen book-to-bill rise to 1.47 for semiconductors in Q1 2026, meaning backlog is growing faster than shipments. Multiple analog companies including TI, ADI, and Infineon have also raised prices, typically a sign of improving supply-demand balance. Analog lead times are already 2-3x pre-COVID levels; analog lead times hit ~30 weeks during the 2022 shortage, and today’s high-end analog is already at 35 weeks vs the 10-14 week pre-COVID baseline.

We can see this in contracting DIO figures as well, as producers draw down inventory (which includes analog wafers), meaning they’ll eventually need to order more wafers.

What does this mean for Miraial? Ideally these indicators translate into higher sales. Note the MSI (wafer shipment index) here is rising in Q1 2026, and is the highest Q1 figure since the cycle peak in 2022.

FOUPs

FOUP (Front Opening Unified Pod) is the intra-fab container. A FOUP is designed for in-fab transport, moving wafers between process tools within a cleanroom via automated material handling systems (AMHS). Once the wafers arrive at the fab, they get transferred into FOUPs, which then ride overhead rail systems from tool to tool throughout the entire manufacturing process. This includes etching, deposition, lithography, etc. The wafer is processed, returned to the FOUP, and the cycle repeats (~up to 500-1000 times) over the course of fabricating a chip, all without the contact of a human. It’s sort of an internal shuttle for the wafer.

FOUPs often have sophisticated features such as interlocks, sensors, and RFID tags that help ensure the proper handling and tracking of wafers within the cleanroom environment and are an integral part of automated material handling systems. They are higher value and carry higher margin vs FOSBs.

Demand wise, since FOUPs are used inside foundries, their demand follows fab capacity utilization as well as new installed capacity. Naturally, foundry capex is in a structural bull market, but utilization is also recovering from a 2023 low after peaking in 2021/2022.

In the chart below, Shin Etsu Polymer (SEP), one of Miraial’s core competitors, reports a precision molding segment that reflects both FOUP and FOSB sales. SEP, however, is more heavily weighted towards FOUPs. So, the chart below shows a decent signal for the relationship between FOUP sales, fab utilization and installed capacity. Note their precision molding segment also includes other products, so the signal isn’t perfect.

Market Structure

I derived market share estimates for the entire FOSB & FOUP market using bottom-up methodology from the firms’ reported segments. This should be directionally correct, but is by no means perfect. Note there are some reports that Miraial has ~40% FOSB market share, but it’s difficult to confirm this data. They are, however, the dominant FOSB producer alongside Shin-Etsu. FOSBs are by no means a commoditized product, however Entegris and Shin-Etsu are more focused on producing FOUPs given the higher margin and complexity of the product. This, alongside the technical barriers mentioned earlier, reduces the competition Miraial faces in the FOSB market, while providing it decent optionality to FOUPs through oligopoly market structure it operates in.

Upside

Miraial has been aggressively adding capacity over the past 2 years. So, it’s effectively operating at its lowest utilization in years. The Company doesn’t report a utilization rate, or production volumes, but we can proxy this with Sales / PP&E, with the assumption ASPs stay relatively stable. Despite the company looking to increase their exposure to new segments long term, their goal with the capex was largely FOSB / FOUP focused: “The company is investing in production capacity expansion and automation under the mid-term growth strategy [to] 2028, targeting semiconductor market growth and share gains”.

I’m growing their sales at ~20%. Given where we seem to be in the analog cycle bottom, and the structural tailwinds for FOUP demand, this seems reasonable. Importantly, this is below management guidance, which implies a 24% CAGR on their revenue target to FY 2029, ending January 2029. Even with that growth, utilization ramps slowly. To FY 2029, it gets to 140% sales / PPE vs the former 180% we had at peak utilization in FY 2022 / CY 2021. I’m modeling a recovery in margins to achieve 11% net margins by 2028. Given the operating leverage, I think this is reasonable; incremental EBIT margins were over 50% between FY2021 and FY 2022 as utilization ramped. My model has incremental margins reaching around the same levels. Q1 2027 guidance is 20% y/y top line with operating margins moving to 5% (from 3% last year). The incrementals aren’t huge here, but management has a history of sandbagging guidance, and there are some front-loaded costs in Q1 (mostly related to SG&A) that might be dampening the ramp. As my returns point to below, this is a FY 2028 (CY 2027) story, and so it’ll take the next ~1.5 years for the thesis to play out. Management agrees: “In the Plastic Molding segment, while demand for advanced products in the semiconductor market is robust, a certain amount of time is still required for recovery in legacy products — a gradual recovery trend is expected to continue.”

We can see this pick up in demand from the backlog (for its plastic molding segment; FOSBs and FOUPs) as well; the most recent figure grew 35% y/y.

My base case is a P/E of 12x on FY 2028’s earnings, which gives Miraial an implied share price of ~2,500 yen; about 55% higher than the current price of ~1,600 yen. The bull case is a re-rate to around 15x. Comps trade anywhere from 15 – 30x P/E, but Miraial is smaller, so I think closer to 12x is warranted.

Management provides guidance to FY 2029; sales of 24Bn yen, operating margin of 4.7Bn yen, and ROE of 11%. On the current capital base, their implied net income for FY2029 is ~2.5Bn yen. On a P/E of 12x, this implies an upside of ~84%, or an IRR of 28% over the next 3 years. I’m being a bit more conservative, but the trajectory of my forecasts is similar.

Risk / Reward

My downside case is modest revenue growth and a flat 8% net income margin. I think this is unlikely, but if I’m completely wrong on the cycle it is a possibility. Margins would stay depressed given the PP&E / D&A overhang. My probabilities are rather conservative in my view, as I think the downside case is likely lower than 25%.

On top of that, the stock has a real margin of safety to it. If we look at the firm’s liquidation value, the EQ is relatively fairly priced as is. Notably, most of the company’s land is severely undervalued on its books, as it was acquired decades ago.

Overall, I think the risk / reward is quite attractive.

Risks

There are risks. I could be wrong on the analog cycle as mentioned, however I don’t think I am. The market isn’t always right, but it appears the market is betting on the cycle recovering. Stocks like TXN, ADI, MPWR are relatively pure analog signals, and are all up considerably in the past few months.

Miraial also recently completed an M&A as part of their strategy to diversify away from their FOUP / FOSB market as it wants to reduce the business’ cyclicality. They acquired a company called Nunotani Nautical which manufactures and distributes marine navigational and engine control instruments. The target is relatively small (~2 billion yen in sales; ~1/6 the size of Miraial) and not overly profitable. Management states there’s shared technical / engineering knowledge between both firms, and they intend on replacing some of Nunotani’s metal components with components made by Miriaial’s plastic molding processes. They also see this as a good way to expand into social infrastructure, since Nunotani supplies major domestic and international shipyards. The main risks I could see here are the usual M&A problems; integration risks, weaker margins post M&A if synergies don’t materialize, etc. But given the small size, I don’t see this as thesis killer even if the deal doesn’t work out.

Lastly there is some customer concentration risk. Top 3 customers = ~30% of sales. These are Sumco, Samsung Japan, and GlorySemi. This has come down historically from ~45% in FY 2023 to <30% as of FY 2026.

GLORYSEMI has the most opaque customer relationship among Miraial’s disclosed counterparties. Incorporated in Singapore in October 2021 and registered at Golden Mile Complex, an address shared by hundreds of trading entities, its principal activity is wholesale trade without a dominant product, with electronic components as a secondary line. This profile is consistent with a procurement intermediary routing Japanese semiconductor-related goods to Chinese end-buyers in ways that are harder to trace or restrict, a structure that has become increasingly common as US export controls have tightened.

The relationship is relatively less important now though; GLORYSEMI has contracted from ~11% of revenue at its peak to ~6% in FY2026, likely reflecting both the broader semiconductor downturn and growing scrutiny of Singapore-based intermediaries serving Chinese buyers. At current levels it is a smaller direct concern, but if export controls expand to cover semiconductor handling equipment such as FOUPs and FOSBs, Singapore-routed procurement could be a channel that attracts increased scrutiny.

More importantly, GLORYSEMI is part of larger Chinese exposure. Miraial’s own geographic breakdown shows China at 22% of FY2026 consolidated revenue, and that number could contract depending on the direction of geopolitics between the US and China. As we saw with the Russian oil sanctions however, things tend to get re-routed, and we could see more “GLORYSEMIs” popping up in different geographies.

Lastly, a notably sizeable customer in the latest filings was Samsung Japan. Samsung Japan is buying FOSBs (not FOUPs) since it’s shipping wafers from its Japanese sales subsidiary which ships wafers to Samsung’s fab in Korea (to make more memory). A green flag; Samsung’s capex is rising materially given the structural demand pull for memory from AI. This means more FOSBs.

Disclaimer: I own shares in Miraial and may buy or sell at any moment without disclosure. This write-up is not financial advice and is for entertainment purposes only.

Thanks for reading! Subscribe for free to receive new posts and support my work.