2025 was a good year, though not the kind that shows up in short-term performance tables.

Energy remains the anchor, representing more than half of our consolidated net asset value. This isn’t by design; it just followed the exorbitant returns from our investments in the Canadian oil patch between 2018 and 2021.

In June, we added to this exposure, buying shares in Brazil’s national oil company Petrobras right at market bottom. Petrobras pumps 2.7 million barrels a day—eighth in the world, wedged between Chevron and TotalEnergies—with proven reserves good for a decade-plus, and prime offshore assets that break even below $30 a barrel.

Yet earlier this year it traded like a company with one foot in bankruptcy court and a pro forma dividend yield so high it looked like someone misplaced a decimal. Picture an oil major worth $71 billion, promising to pay out $45 to $55 billion in dividends over five years, and maybe $65 billion if the stars behave: no spreadsheet required to see the math falters. For us, this may be déjà-vu all over again.

The reality is that nobody invests in Petrobras anymore—the risks are too popular. The main one has four letters and an office in Brasília. The government—read Lula—owns 37% and runs the show. It’s bad press when the top man, who used to praise Mao and Chávez, is back at the helm of Brazil’s crown jewel. Plus, five CEOs in three years say plenty about job security there.

Lula has bled Petrobras before, forcing it into every kind of side hustle. Diworsification, fuel subsidies, and price caps cost the company tens of billions of dollars. They bought votes while Petrobras paid the bill. And that’s before his successor, Dilma Rousseff, presided over the greatest corruption scandal in corporate history.

But every saint has a past, and every sinner a future. After a major cleanup, the company reemerged on solid footing, with a leaner balance sheet and disciplined governance. It sold off dead wood, doubled down on its wildly profitable offshore oilfields, and started acting like a serious business again. International creditors had made their point clear.

Or had they? Widely respected across the board, former CEO Jean Paul Prates wanted to continue that path. His reward was being shown the door. Then Magda Chambriard stepped in, and publicly declared that the new plan is to make Petrobras « as big as possible. » Equity investors have heard that tune before. Small wonder they fled.

However, behind closed doors, the state wants its dividends, and it probably won’t butcher its own cow a second time. That, in essence, is our bet.

Besides, and remarkably, despite the government’s nefarious interference in the past, Petrobras has delivered an average of $22 billion in annual free cash flow between 2014 and 2024—a decade marked by two brutal downturns in the oil business.

Against a market cap of $71 billion, these fundamentals translate into a valuation of about three and a half times earnings, and a dividend yield of around 15% excluding special distributions. That’s assuming Petrobras performs just as badly in the next decade as it did in the last, if « bad » is how one would describe a business that spits out $22 billion a year and keeps growing production despite its already massive scale.

With that said, our investment in Brazil pales in comparison to our new commitments in Constellation Software’s spinoff Lumine, as well as the purchase of additional shares in its other offshoot Topicus—which, before its valuation collapse this year, was the second best performing investment we held across two of the three investment vehicles we manage. Taken together, Topicus and Lumine are now our largest investment at cost.

It used to be that Constellation Software and its progeny could do no wrong. Now they can do no right. The three took a serious beating this year, with Constellation down 30%, Topicus 35%, and Lumine 45%, all in just three months. The reasons are that Mark Leonard stepped down for health reasons and artificial intelligence—of course—is coming to eat everyone’s lunch, including vertical software businesses. Or so the story goes.

Although out of fashion, the three are still in business. The matrix Leonard designed was made to outlive him. As for AI, it still needs hands, heads, and smarts. Rollout, training, maintenance, security: none of it runs itself. So the plumber shouldn’t fear the new wrench. If anything, it should be a gift to the niche, entrenched, mission-critical businesses these companies run.

We’ve followed Constellation Software for fifteen years, always saying we’d bet big on it—or its offspring—if a pullback ever came. It never did, and we watched from the sidelines as Constellation’s market value grew nearly a hundredfold. Now that the stars have aligned, we pounced at scale, exactly like we did a first time with Topicus three years ago.

Over the last twelve months, we have also increased our position in consulting group Alten, already discussed at length in last year’s letter, and initiated a new position in Groupe SEB, the superbly-run, family-controlled French maker of small domestic appliances and cookware, home to brands like Tefal, SEB, Rowenta, and Moulinex.

Groupe SEB saw its valuation collapse below book value and to a single-digit multiple of operating earnings in late 2025—a record low in its long history—after it revised its growth objective. This, in our view, does very little justice to its enviable record.

Again in France, we acquired shares in BNP Paribas two weeks after the bank got sanctioned for its alleged dealings in Sudan. Bad news travels faster than arithmetic and everyone loves a good beating, especially when it’s a bank taking the punches.

As a result, Europe’s second-largest lender traded like an institution that forgot its own balance sheet, selling at the time of our investment for 72% of tangible book value and an 8% dividend yield, despite earning double-digit returns and maintaining a healthy capital base. Numbers like that are seen only when panic takes the pen. In the case of BNP, last time was during peak Covid.

Meanwhile, competition in Europe struts around in shinier suits. HSBC and UniCredit get twice the respect—literally twice the multiple on book, and half the dividend yield. Even Deutsche Bank—yes, that Deutsche Bank—draws a friendlier crowd these days. Not to mention US lenders, which are getting back to multiples last seen fifteen years ago, before the subprime crisis.

Since 2012, by all accounts a brutal stretch for European banking, BNP’s book value compounded at 4.4% annually, while returns to shareholders grew 10.2% per year on average. Not JP Morgan territory, but nothing to blush at either. Plus, the French bank is doing what must be done. Its branch network is shrinking—a third gone by 2030—and its cost-to-income ratio is dropping to match Europe’s leanest operators.

In other news, it grabbed AXA’s asset management arm. A month before, it scooped up HSBC’s private banking arm in Germany. BNP is building serious muscle in a business where scale is religion. The AXA deal turned it into a heavyweight in the European asset management game, just behind Amundi and UBS. Insurance and asset management will soon make up a fifth of the group’s consolidated income, cushioning the swings in banking revenue.

So at €66 per share, we got €91 of tangible equity throwing off double-digit returns, plus an 8% yield. The discount was steep and in our view abnormal, which may explain why it has since narrowed considerably.

These new purchases were partly funded by selling Tessenderlo, which continues to face impossible headwinds despite its astute management, Daily Journal Corporation and Richelieu Hardware. If circumstances allow, we would gladly welcome the opportunity to once again becoming shareholders of these outstanding companies. Standard also took on negligible leverage for the first time in its history, with low-cost debt equaling 4% of total asset value on December 31st.

The least pleasant news in 2025 was the write-off of our venture investment in Cambridge-based Generation Bio, which cost us 1.1% of our asset value. We knew the risks but had higher hopes than the lamentable end the company faced. This mistake won’t be repeated.