Pieter Slegers — Buy-Hold-Sell List: May 2026

The problem set out here is that nobody knows which artificial intelligence company will end up winning, and the honest answer is to stop guessing. The analogy used is the California gold rush: most prospectors found nothing, and the people who reliably made money were the ones selling picks, shovels and boots, because they got paid no matter who struck gold.

Brookfield is the shovel seller. It owns the physical things AI has to run on — land, power stations, buildings, cables. It has launched a $100 billion programme to build AI infrastructure, signed a $5 billion deal with Bloom Energy to install up to a gigawatt of power generated right next to the data centres that need it, and acquired 350,000 square metres in Sweden that lets one site grow from 300 megawatts of capacity to 750.

None of that depends on a particular model or chip winning. Whoever builds the data centre needs the land, the electricity and the fibre.

The shares have returned nearly 3,000% since 1997, about 12.6% a year, and on the firm's own model they are worth roughly twice the current price.

Adobe makes Photoshop, Illustrator, Acrobat and the rest of the software creative professionals use, sold as subscriptions. Revenue has grown about 15% a year for a decade.

On this issue's own numbers it is the single most attractive company on the whole list: the shares cost under 11 times next year's profits, the model puts fair value at $602.90 against a market price of $250.70, and the expected return is 20.8% a year. It is described as trading at its cheapest valuation ever.

And then it is declined. The reason is stated plainly: all of that only holds "under the assumption you don't believe Artificial Intelligence will disrupt their business model" — and if image-generating AI makes professional design software less necessary, the cheapness is a trap rather than an opportunity. "For me personally, Adobe is in the 'too hard' pile."

That gap between the model output and the judgement is worth noticing. Anyone following the published Buy list mechanically would be buying something its author will not.

Nine days after being listed among the purchases made "because it was cheap" and given a "not sure" on whether it was worth owning for a decade, Novo Nordisk is rated a Strong Buy here — and the reason is that the news changed.

Profit, revenue and operating profit all came in ahead of forecasts and management raised its guidance for the year. The pill version of Wegovy is described as the most successful drug launch ever, with more than two million prescriptions written and the Wegovy brand now taking 65% of all new weight-loss prescriptions in America — which is the number that answers the "losing to Eli Lilly" worry. It has also signed a partnership with OpenAI to speed up drug discovery.

The shares are at their lowest valuation ever: about 12 times next year's earnings with a 4.3% dividend, against a model fair value roughly double the market price, giving an expected return of 19.2% a year.

The two views are not contradictory, and it is worth understanding why. The archive rates every holding twice — once on the quality of the business, once on the price. Novo remains a Medium conviction on quality. It is simply very cheap.

FICO is upgraded from Hold to Buy here, and unusually the upgrade comes with the objections attached rather than removed.

The positive case is the price: the shares are down 37% this year, at one of their cheapest valuations ever, against a model fair value about 9% above the market and an expected return of 11% a year. Over ten years the shares have still compounded at nearly 26% annually.

The two risks are named without hedging. First, the company pays its staff heavily in shares, which flatters the reported profit and therefore the apparent cheapness. Second, and more seriously: "the fact that FICO might lose its monopoly" — a US housing regulator has approved a rival credit score for mortgages, which is what caused the fall.

Holding both at once is the point. This is an upgrade on valuation while the central question about the business remains genuinely open, and the write-up says so.

Fairfax moves from Hold to Buy on this month's list, which is the step that leads to it actually being bought in August.

The numbers behind the rating are unusual and worth reading carefully. The shares cost 9.1 times next year's earnings — but the "fair exit" multiple, the multiple the model assumes you would eventually sell at, is 8.0. That is lower than today's, and it is the only company on the entire 49-stock list where that is true. The model is therefore not expecting the valuation to improve at all; the whole 10.7% expected return comes from profits growing and dividends being paid, with the multiple assumed to shrink.

That is a conservative way to underwrite an insurance holding company whose earnings swing about with investment results, and it produces a modest answer: fair value 2,352 against a price of 2,168, only 7.9% of upside. Which sits oddly beside the estimate published a week earlier of an intrinsic value of 3,000 Canadian dollars.

3i Group joins the watchlist this month and arrives already rated a Buy. It is a UK-listed investment company whose value comes almost entirely from Action, the European discount retail chain.

The model puts fair value at 50.8 against a market price of 25.7 — the shares priced at roughly half what the underlying business is reckoned to be worth — with an expected annual return of 15.8%, helped by a 3.1% dividend.

Its arrival completes a sequence worth noting: priced at a small discount to net asset value on 21 April, named a buy candidate on 28 April, ranked the month's second-best idea on 3 May, and formally added to the rated list on 7 May. Four appearances in seventeen days, each one a step closer to a purchase.

Alphabet is added to the watchlist this month and rated a Buy: the model puts fair value at $665 against a market price of $382, an expected return of 13.6% a year, on assumed profit growth of 15%.

It is also one of the three big technology companies whose April surge is used, earlier in the same issue, to illustrate how concentrated and AI-driven the market has become — up 21.7% in the month on AI and advertising strength.

There is a tension a careful reader should notice. Sixteen days earlier the stated willingness to buy Alphabet was at 18 times earnings, or about $210 a share, against a market price of $336 — a refusal to pay up. The model here reaches a Buy at $382 by assuming the shares should eventually trade at 25 times rather than 18. The company has not changed; the assumed exit multiple has.

Games Workshop is the best illustration in this issue of why the portfolio is rated on two separate axes.

On quality it is a Very Strong conviction — the top tier, the Warhammer maker with its own shops, its own manufacturing, a devoted community and the ability to raise prices every year without losing customers.

On price it is the only holding in the entire portfolio that is overvalued, and by a long way. The shares cost 33 times next year's profits against a "fair" 23, the expected return is just 6.4% a year — below every one of the 49 names on the buy list — and the model puts fair value at 124.5 against a market price of 189.5, so it is trading about 51% above what it is judged to be worth.

Hence the Hold: a wonderful business, currently at a price that does not offer a sensible return. Nothing is being sold; nothing is being added either.

Nathan's Famous appears in this archive for the first time, upgraded from Sell to Hold and described only as a "branded food and licensing company."

The business model is worth knowing even without a full write-up: Nathan's is best known for hot dogs, but most of its profit comes from licensing its name to a meat producer and collecting a royalty, plus franchise fees from restaurants it does not own. That makes it far more capital-light than a restaurant chain, which is presumably why it is on a quality-focused watchlist at all.

No figures, valuation or thesis are given here — only the rating change, which moves it from "would not own" to "would own at the right price."