50 new hedge fund pitches — Q2-2026 fund-letter compilation (SSR distribution)

Arista makes the switches that move data between the thousands of chips inside an AI data centre. Brown Advisory's argument is that this is really a software business wearing hardware clothes: one operating system, EOS, runs across the whole range, so customers get better monitoring, automation and reliability, and end up paying less over the life of the equipment than they would with a rival.

They expect Arista to keep taking share as AI pushes networks to be rebuilt more often, with revenue growing more than 20% a year and the best profitability in its industry. They bought during a dip caused by short-term supply problems — the sort of wobble that has nothing to do with the long-run case.

Accton is the Taiwanese company that actually builds many of the network switches used in large data centres, often badged for someone else. ClearBridge calls it a high-quality compounder: it owns few factories relative to its sales, so it needs little capital to grow, and it has long-standing ties to the handful of customers who buy switches by the shipload.

The reason to own it now is that AI moves enormous volumes of data between machines, which forces network gear to be upgraded on a shorter cycle than before. ClearBridge's point is that you get that growth without paying a scarcity premium — the valuation is still attractive.

Sumitomo Electric makes cable: the very high-voltage kind that carries electricity long distances, the undersea kind that links offshore wind farms and countries, the optical fibre that carries data, and the wiring inside electric vehicles. Artisan bought it as a bottleneck play — not because cable is exciting, but because the world cannot make enough of it.

The specific mechanism they name is a queue. Utilities and renewable developers are competing for cable production slots years ahead, which hands pricing power to the few firms with capacity. Every grid upgrade, every data centre connection and every electrified car adds to the same order book.

Broadcom designs the custom AI chips that big technology companies use instead of buying everything from Nvidia — Google's TPUs are the best-known example — and it also sells the networking silicon and, through VMware, the software that manages large fleets of servers. Baron bought the shares into a post-earnings sell-off.

Their case is that the obvious bear argument — that customers will eventually design these chips themselves — underrates how hard the job has become. Modern AI systems require the compute, memory, chip-to-chip links and networking to be designed together, and Broadcom is at the front in all of them. A company whose revenue depends on having the best chips cannot afford to hand the work to a second-best in-house team.

The evidence they point to is contractual rather than rhetorical: Google extended its agreement with Broadcom to 2031, OpenAI got its first inference chip designed and taped out in nine months, and Apple signed a multi-year deal spanning several product generations. Broadcom's chief executive expects custom chips to match graphics-processor volumes next year.

Samsung makes memory chips — the components that hold data while an AI processor works on it — as well as phones, screens and, through its foundry, chips for other companies. Baron's thesis has four parts, and the fourth is the punchline: at roughly four times earnings, you are paying for the memory business and getting everything else free.

On demand, the argument is that an AI system's output is limited by how fast it can pull data out of memory. Agent-style AI, which reasons internally before answering, needs far more of it — one estimate they quote is that the amount of text a model holds in working memory has grown thirtyfold in a year. On supply, the high-bandwidth memory used beside AI chips consumes more silicon per gigabyte than ordinary memory, flash capacity can no longer be converted to make up the difference, and a new factory takes two to three years.

Their third claim is the most contested: that memory is becoming less of a boom-and-bust industry, because customers now sign long-term contracts and the chips are increasingly designed jointly with the processors they serve, which makes them harder to swap for a rival's. The free option on top is the foundry — currently loss-making, but a hedge if anything ever disrupts Taiwan.

SK hynix is the Korean memory maker that leads in high-bandwidth memory, the stacked chips that sit next to every AI accelerator. Two separate funds in this compilation pitch it, which makes it the most-agreed name in the sample.

Artisan's version is about technical leadership: HBM is what lets AI chips process data quickly without wasting energy, and SK hynix is co-developing the next generation with Nvidia while spending on the packaging and testing capacity that stacking requires. They attach an honest caveat — memory prices have risen so fast that customers may eventually balk.

Buffalo's version is broader and simpler: memory is what makes agent-style AI possible at all, and agent workloads need not just the exotic stacked memory but ordinary memory and flash storage as well. SK hynix's particular advantage is that its HBM is built to order for each customer rather than sold as a commodity.

Western Digital makes hard disk drives — the spinning magnetic storage that remains the cheapest way to keep very large amounts of data. Most of its sales now go to cloud providers rather than consumers, which makes it a supplier to the data-centre build-out rather than to the PC cycle.

Alger's thesis is about industry behaviour, not demand. Only two companies still make these drives at scale, and both have chosen to increase how much data fits on each disk rather than build more factories. That restraint keeps supply tight, which is why prices and margins came in above the company's own guidance last quarter.

Samsung Electro-Mechanics makes multilayer ceramic capacitors — tiny components that smooth and filter electrical current. Every electronic device contains hundreds or thousands of them; an AI server or an electric car contains far more, and of a higher grade.

Hood River's edge was a call on the cycle before others made it, and specifically on why it was turning. The market read the recovery as customers restocking phones and laptops; their research said the real driver was demand for the higher-value parts used in AI servers, data centres, driver-assistance systems and electric vehicles. Because those parts sell for more, both volumes and the mix improved together — and subsequent results confirmed it, which is why they judge the upturn broader and longer-lasting than a normal consumer bounce.

Disco makes the machines that cut, grind and polish silicon wafers — an unglamorous step that every advanced chip, and especially every stacked memory package, has to pass through. Buffalo holds it as a top-ten position for exposure to trends they consider durable: AI, electrification and defence.

Their stated criteria are worth more than the write-up itself, which is portfolio commentary rather than a company case: businesses with a real competitive advantage that generate consistent free cash flow, run by managers who earn returns above their cost of capital. Disco qualifies because its tools are hard to replace once designed into a production line.

Wedgewood's pitch is narrow and specific rather than a general defence of Meta. Investors punished the company for raising its 2026 spending plans by about $10 billion, blaming the rising cost of memory chips.

What Wedgewood says everyone missed is an asset sitting on Meta's own books. As part of a chip-supply arrangement struck with AMD in February, Meta received warrants — the right to buy AMD shares at a fixed price — which they estimate are now worth close to $90 billion. That is roughly nine times the extra cost investors were worried about. It is not a hedge in the accounting sense, but economically it more than offsets the problem, and the offset is invisible in the reported numbers.

Snowflake sells software that lets a company keep all its data in one place in the cloud and ask questions of it. Spyglass calls it a top contributor and a core position.

The detail they single out is unusually concrete for an AI claim: Snowflake raised its guidance and said explicitly which new AI product was responsible. Most software companies talk about AI in general terms; being able to attribute a guidance raise to one product is evidence rather than narrative. They trimmed the position because the shares had run, but still think the long-term opportunity is misunderstood.

Guidewire makes the core software that property and casualty insurers use to write policies, handle claims and bill customers — the system of record, which almost never gets replaced once installed. Baron used an 18% drop in the quarter to hold on rather than sell.

The reason is that the hard part is over. Guidewire spent years shifting customers from software installed on their own servers to a cloud subscription, a transition that depresses reported growth while it happens. That is now substantially complete, insurers are migrating faster, and the money previously spent on building cloud infrastructure can be redirected to new products it can sell to an installed base that has nowhere else to go. Profitability on the subscription business has already improved noticeably.

TechnologyOne sells administrative software — finance, payroll, HR, asset management — to Australian and British councils, universities and government agencies. It has now reported record recurring revenue for seventeen consecutive years, with no customers leaving at all last period.

The reason LHC Capital highlights it is what its new AI product reveals. "Plus" sits on top of the existing software and lets staff ask questions across every system at once. Crucially, it only works well if a customer has put more of its data into TechnologyOne — so the AI product sells more of the underlying software rather than replacing it. Most early Plus deals came bundled with extra modules or the removal of a rival's system, and one university bought the entire suite on a ten-year contract simply to give Plus more to work with.

On top of that, customers pay for AI usage separately, creating a new recurring revenue stream above the subscription — and early usage has run well ahead of what management expected.

EPAM supplies software engineers to large companies rebuilding their technology. It has been the worst position in White Falcon's portfolio, and they say so plainly, quoting the line about there being no difference between being early and being wrong.

The shares have fallen to about four times earnings before interest, tax and depreciation, with more cash than debt. The cause is straightforward: firms are postponing big technology projects until they understand what AI will change.

White Falcon's argument is that the delay is temporary and the direction is favourable. The bottleneck in AI has moved from building models to actually getting them working inside real companies with old systems and messy data — which is precisely the kind of engineering EPAM has spent decades organising. Their evidence that the market is mispricing it: a smaller rival was just bought at more than double EPAM's valuation.

Baron held SpaceX privately for years under a confidentiality agreement and can finally explain the position now that it has completed the largest stock-market listing ever, raising more than $85 billion. They describe it as a company with no comparison — "N=1".

The foundation is reuse. Where the rest of the industry throws a rocket booster away after one flight, SpaceX flies one up to 35 times, so it needs three boosters for a hundred launches instead of a hundred. That has cut the cost of putting a kilogram into orbit roughly tenfold from the industry standard, with the next vehicle, Starship, aimed at another huge reduction.

Cheap launch built Starlink, the satellite internet service with more than 12 million subscribers, attacking a $1.5 trillion global connectivity market where each ten percentage points of share is worth roughly $150 billion of revenue at very high margins.

The newest and least understood part is artificial intelligence. SpaceX has signed its first agreements to host computing for Anthropic and Google, together worth about $26 billion a year, and built a 100,000-chip cluster in 122 days against an industry norm of about two years. The longer-term idea is to put data centres in orbit, where sunlight and space are free: Baron's arithmetic suggests the launch cost for a gigawatt of orbital computing could be under $2.3 billion against $10-15 billion to build and run the equivalent on the ground. Everything beyond that — chip manufacturing, point-to-point transport, mining, the Moon, Mars — they explicitly value at zero.

This pitch argues the opposite of Jay Singh's own position. Singh is short AST SpaceMobile on valuation and execution risk; Crossroads Capital owns it and thinks the company has crossed from research project to operating business.

AST is building satellites that let an ordinary, unmodified mobile phone connect directly to space, so a normal handset works where there is no mast. The evidence Crossroads cites is procedural rather than promotional: American regulators have authorised commercial service for a network of up to 248 satellites, the company holds about $3.7 billion of cash — enough, they argue, to finish the constellation without asking investors for more — and manufacturing has passed 500,000 square feet with parts for forty satellites already built.

The change they consider decisive is how satellites get to orbit. Until recently AST launched one at a time; in June it flew three together on a single rocket, which they describe as launch maths shifting "from additive to multiplicative". Twelve are now in service, with about forty-five targeted by early 2027.

Commercially it has partnerships with more than sixty mobile operators covering over three billion subscribers, and a growing defence business where they claim AST is the only bidder on earth able to demonstrate the capability. On the failed launch that destroyed a satellite, their view is that the fault was the rocket provider's, the loss was about $125 million and partly insured, and nothing about it changed the plan.

FTAI overhauls jet engines — specifically the CFM56, the most common engine on commercial aircraft. Crossroads bought it eighteen months ago precisely because a short-seller campaign had crushed the shares, and they now describe it as having graduated from a special situation into a genuine compounder.

The business works by buying old engines, stripping them, and rebuilding them into swappable modules using its own replacement parts. Because airlines can exchange a module in days rather than wait months for a full overhaul, FTAI sells speed in a market where a grounded aircraft earns nothing. It is also shifting from owning aircraft to managing other people's money against them, which needs less of its own capital.

Two events during the quarter matter most. The engine's original manufacturer signed a multi-year materials agreement with FTAI — a competitor it might have been expected to squeeze — which Crossroads reads as proof the incumbent does not feel threatened enough to retaliate. And a joint venture signed a five-year supply deal with a large cloud provider, with an opening order of $1.465 billion for mobile generators built from converted jet engines. Data centres wait years for new turbines; a jet engine that already exists can deliver power now.

APA drills for oil and gas in the Permian Basin of west Texas and in Egypt. The shares fell during the quarter because oil retreated on hopes of a settlement in Iran, not because anything went wrong at the company.

Hotchkis & Wiley own it as a valuation case. It generates strong spare cash, helped by getting better-than-average prices for its natural gas, and it has drilling prospects in Suriname, Egypt and possibly Alaska that they think the market ignores. The main bear argument — that its best Permian acreage will run out sooner than rivals' — is acknowledged rather than dismissed; their answer is that the shares already price it, since APA trades cheaply against the cash it produces while carrying an investment-grade balance sheet.

Permian Resources drills in the Delaware Basin, part of the Permian. Conestoga started a new position on three qualities rather than a view on the oil price: it costs them less than peers to produce a barrel, management has been disciplined about what it buys and builds, and cash generation has held up through different price environments.

The point of a low-cost producer with an investment-grade balance sheet and a deep inventory of good wells is optionality: it can keep drilling when prices are weak and others cannot, which is when acreage and companies get cheap.

Exco is a small natural gas producer traded over the counter rather than on a main exchange, with the Canadian insurer Fairfax owning nearly half of it. Cedar Creek's pitch is arithmetic, not narrative.

They bought at $17.83 a share against an accounting book value of about $24.77 and debt of only $81 million. At the purchase price that was under four times their estimate of annual profit and about two times cash earnings. The leverage to the gas price is dramatic: every dollar rise in the price per thousand cubic feet adds roughly $3.00 to earnings per share, on a stock trading around $22.

Their conclusion is the sort a private buyer would reach rather than a stock-market one: in a sale they think the assets are worth $60 to $80 a share.

McDermott designs and builds large energy facilities. It went through a painful period taking on contracts that lost money; those are now essentially finished, so it should be profitable again. The remaining obstacle is its balance sheet — it owes more than it owns on paper, which stops customers awarding it big contracts and keeps the shares cheap.

The fix is a $500 million rights offering: existing shareholders get the right to buy new shares at a fixed price. What makes this one unusual is that the price is set far below where the shares trade, and there is no facility to buy more than your allocation. Anyone who does not take part is therefore massively diluted, and the shares they decline go to four large investors who have agreed to underwrite the deal. Alluvial will take up its rights in full — declining would simply hand value away.

Afterwards the company is far safer and, on their numbers, trades at about three times its 2027 earnings before interest, tax and depreciation, or under four times 2028 earnings. The manager compares it directly to Garrett Motion's 2021 rights offering, down to the same backstop structure, which turned into a very large winner after a couple of dull years.

Green Dot is being taken apart deliberately: shareholders have approved selling its technology operations and merging its bank with CommerceOne Financial. Only a regulatory sign-off remains, which Alluvial expects shortly.

Shareholders will receive a large cash distribution when the deal closes, and are left holding shares in the combined bank. Even after a decent run, those shares still trade below the accounting value of the bank's assets — and Alluvial's view is that the people who will run it are capable operators who would buy back stock rather than tolerate that discount. Adding the payout to the residual value, they see 50-70% upside over a few years.

Beasley owns radio stations and the buildings they sit in. It has been a disaster — the shares are down more than 90% from their peak a decade ago, and its bonds were trading at 25 to 30 cents on the dollar because it borrowed too much.

Then a distressed-debt investor offered to swap those bonds for new debt at half their face value, cutting Beasley's net borrowings by nearly $100 million. The catch, and the reason Kingdom Capital is interested, is a clause: if the family that controls Beasley does not repay the new debt on time, the lenders take 95% of the company.

That single provision changes what management is trying to do. For the first time the family's incentive is to sell assets and pay down debt — exactly what an outside shareholder would want — because the alternative is losing the business. Kingdom's estimate is that the property and radio licences could support up to around $200 a share once the debt is cleared, though that depends entirely on selling assets at decent prices and on time. They built the position under $6 a share.

Pershing Square bought both Visa and Mastercard earlier in the year, and this write-up is really a defence of a business model against three fashionable objections.

The model first: the networks take about 0.2% of each card transaction in exchange for making it work instantly, anywhere, with fraud protection and a way to dispute a charge. They lend nothing and carry no credit risk, and because their fee is a percentage, inflation raises their revenue automatically. Card payments still account for only about half of what consumers spend globally, and services sold alongside the network now make up roughly 40% of Mastercard's revenue while growing two to three times faster than payments.

The shares had fallen to 22 times expected earnings on three worries. On stablecoins — digital tokens pegged to the dollar — Pershing argues they will grow where cards are weak (business payments across borders, remittances, savings in unstable currencies), not where cards are strong, because a stablecoin payment cannot easily be reversed and offers no credit or rewards. On AI agents doing the shopping, they argue that agents need exactly what the networks provide: proof the buyer really authorised the purchase, spending limits, and someone to complain to. On regulation, the proposals that scared investors have stalled, and would not matter much anyway.

ICE runs exchanges — most importantly the market for Brent crude oil and European gas — plus financial data and mortgage software. It was GreensKeeper's worst performer, down nearly 22%, even though revenue rose 20% and profits 34%.

Two fears did the damage. One is that earnings have peaked after a volatile period. The other is a new product called perpetual futures, popular in cryptocurrency, which never expires and might in theory pull business away from traditional contracts.

GreensKeeper's rebuttal is about who ICE's customers are. They are not speculators but oil companies, utilities and institutions hedging real exposures, and they need contracts that settle on known dates, are standardised, are guaranteed by a clearing house and are recognised by regulators — none of which perpetual futures prioritise. And if institutional demand ever did appear, ICE has the technology, clearing and customer relationships to launch its own version faster than a start-up could.

Equitable sells retirement products, manages money and advises wealthy clients. Harris Associates started a position because the company has quietly changed shape.

Traditionally an insurer makes money on the spread — it invests your premiums and keeps the difference between what it earns and what it owes you — which requires holding a lot of capital and carries market risk. Equitable has shifted so that more than half its distributable cash now comes from fee-earning businesses that need little capital. Fees are steadier and worth more per dollar than spread income, but the shares are still priced as though it were an old-fashioned insurer.

They also read the pending merger with Corebridge as a genuine combination of equals that adds scale and should increase earnings. At under six times their estimate of 2027 distributable cash flow, they think the price understates the quality of what is now being earned.

Schwab is the American brokerage and custody business where millions of investors and thousands of advisers hold their accounts. Baron lists it among its top five positions, in a very concentrated fund where the ten largest holdings are nearly three-quarters of assets.

The published rationale is generic rather than specific: these are businesses with strong brands, deep technical expertise or proprietary data that cannot easily be copied, and with large markets still to penetrate. For Schwab the relevant asset is the platform itself — once an adviser's clients are custodied somewhere, moving them is painful — which is why it keeps gathering assets year after year.

Marsh is the world's largest insurance broker: it arranges cover for large companies with complicated risks. Artisan added aggressively when the shares fell to fourteen times earnings, a level they consider absurd for the business.

The shares fell on a story rather than a result — that AI agents will let companies buy insurance directly and cut brokers out. Artisan say they have looked for evidence and found none, only assertions.

Their counterargument uses history. Direct selling did displace agents in car and home insurance, but only because those policies are essentially identical products with mandated coverage levels, easy to compare on price. A global company's insurance is nothing like that: property in dozens of countries, multiple jurisdictions and regulators, layered policies built to order. That is why direct selling has existed for decades and never taken hold in commercial insurance — and why even in car and home cover, agents still handle most of the market.

They add one more asset: Marsh knows what every major underwriter has charged, and on what terms, across decades and countries. That information is private, so no AI start-up can scrape it and learn from it.

Edwards makes replacement heart valves that are threaded up through an artery rather than implanted in open-heart surgery. It dominates the market for aortic valve replacement, treating a narrowing that eventually kills if untreated.

Baron initiated because three separate things are moving the same way. The market itself should grow from about $7 billion to more than $10 billion as regulators approve the procedure for more patients; one competitor has abandoned the field entirely and another published poor results, both of which push surgeons toward Edwards; and Medicare has widened the set of hospitals allowed to perform the operation.

The second act is the mitral and tricuspid valves, which are structurally much harder to treat and where Edwards has spent over a decade developing both repair and replacement options. Baron thinks that market can grow from $1.5 billion to as much as $8 billion. With 78% gross margins, the combination supports double-digit revenue growth and mid-teens earnings growth for a long time.

Jazz began as a sleep-disorder specialist — its main drug treats narcolepsy — and has been rebuilding itself as a rare-disease and cancer company. Aristotle's point is that the rebuild has already worked, and that this quarter simply confirmed it.

Revenue grew 19%, spread across four different products rather than concentrated in one, and the specific catalysts Aristotle had identified beforehand arrived on schedule: the cancer drug Zepzelca approved for wider use in small-cell lung cancer, continued growth in rare epilepsy, and the sleep franchise still holding the only approved treatment for one condition. A larger cancer launch is being prepared. Cash generation funds both the pipeline and further acquisitions.

CSL is the world's biggest plasma business: it collects blood plasma from donors and separates it into proteins sold as medicines for a range of serious conditions. It is also one of three large flu-vaccine makers and owns Vifor, which treats iron deficiency and kidney disease.

Brown Advisory's point is about price rather than quality. They had admired CSL for years without ever being able to buy it at a sensible valuation. Three things changed that: a badly judged acquisition (Vifor), weaker demand for certain plasma products, and growing hostility to vaccination in the United States.

The reason the underlying economics survive is subtle: because CSL sells more different proteins from each unit of donated plasma than its competitors, it earns more from the same raw material. That advantage is structural and does not go away because the share price has fallen.

Progyny manages fertility benefits for large employers that pay their own healthcare costs. Employees get access to a curated network of clinics; the employer gets better outcomes and, because treatment works more often, fewer repeat cycles.

Buffalo's case has two halves. The demand side is demographic and slow-moving: people in wealthy countries keep having children later, which increases the need for fertility treatment regardless of the economy. The financial side is that Progyny needs almost no capital to grow, holds more cash than debt, trades on an undemanding multiple, and is buying back its own shares quickly enough to shrink the share count meaningfully.

Welltower owns and runs senior-housing communities. Baron's central move is to stop thinking of it as a landlord: they describe it as a mixture of property, hardware and software, because the company has built its own operating platform and data analytics for running the buildings.

That matters because of how the economics work. A landlord who merely collects rent gets a fixed payment however well the business inside performs. Welltower operates the communities itself, so higher occupancy and better pricing flow straight to it — and the software is what raises occupancy and lets it charge for individual amenities rather than a single blended rate.

The backdrop is the strongest part. The population over eighty is growing 4-5% a year, more than double the rate after the financial crisis, while new supply is stuck: construction starts have collapsed, the economics do not work for developers at current costs, and it takes over five years to get a new community permitted and built. Baron sees a path to earnings more than doubling in five years.

Ventas owns senior-housing property, and Guinness holds it alongside Welltower for exactly the same demographic reason. Their numbers make the squeeze concrete: occupancy across the sector has climbed back above 90% for the first time since 2017, while the number of units being built has fallen to about 2% of existing stock, the lowest since 2012.

That combination — more residents arriving, almost nothing new being built — is what lets landlords raise rents. Ventas guided to 15-17% growth in income from its existing properties and raised its acquisition budget to $3 billion, taking advantage of private buyers who cannot borrow as cheaply.

Guinness is careful to flag the risk in their own thesis: hot property sectors usually attract enough capital to over-build and go cold. They do not see it yet, but the construction-starts figure is precisely what would signal the end.

Prologis owns warehouses — the distribution buildings that e-commerce and modern supply chains depend on. Baron holds it with two smaller industrial landlords for a straightforward reason.

The rents on existing leases sit roughly 20% below what the same space would fetch today. As those leases expire and are renewed at market rates, income rises without a single new building or a single extra tenant. On top of that sits slower-moving demand: online retail, more complex logistics, companies holding more inventory closer to customers, and manufacturing moving back toward North America.

Magnum is the ice-cream business Unilever separated at the end of 2025 — Magnum, Ben & Jerry's, Cornetto, Wall's, Breyers and more. It is the largest ice-cream company in the world with about 21% of the market, nearly double the next competitor, and everyone after those two has 2% at most. Two different funds in this compilation pitched it, which makes it the sample's most-agreed new idea.

Upslope came to it while looking for food companies less exposed to weight-loss drugs, since only about a quarter of Magnum's sales are American. Their case has four parts: dominant brands plus a frozen distribution network that is genuinely hard to copy; the chance to grow faster and more profitably now that ice cream is the only thing management thinks about; a direct comparison with its private rival Froneri, which earns about 20% margins against Magnum's 16% and was recently valued at 10-11 times earnings; and a valuation of roughly 9.5 times next year's earnings before interest, tax and depreciation with modest borrowings.

Aristotle emphasises the moat instead: three million freezer cabinets in shops around the world, more than thirty factories, two hundred warehouses and two thousand distributors. Frozen distribution is capital-intensive and unforgiving, and those cabinets are also what drive impulse purchases. About 80% of sales are premium products priced roughly two and a half times private label. They put the shares at about eleven times normalised earnings.

The honest risks both note: a very short history as an independent company, uncertainty about weight-loss drugs, currency exposure and, prosaically, the weather.

Murphy USA runs petrol stations and small convenience stores, mostly next to Walmart car parks, competing on being cheapest every day rather than on promotions. It buys fuel more cleverly than most and runs its sites with very low overheads.

Artisan had watched it for a while and only bought after a change of management, starting small and increasing the position as their research convinced them that a low-price strategy on fuel and tobacco would matter again to shoppers. Their view is that the company is entering a better stretch of the profit cycle: fuel margins are structurally improving, it keeps taking share, it is still opening stores, and the new management team is improving operations.

Viking runs river and ocean cruises for affluent travellers over 55, sold under one brand, marketed directly, on ships that are almost identical to one another. That sameness is the whole point: identical ships are cheaper to build, staff and operate, and one brand plus direct marketing means the company keeps the customer relationship rather than paying a travel agent for it.

The result is customers who come back repeatedly and, unusually for a consumer business, visibility years ahead — it is already taking bookings for 2027 and 2028. The cheaper river business acts as the recruiting ground for the more profitable ocean business, where Viking has the youngest fleet in luxury cruising and more ships on order.

Brown Advisory expect revenue to compound at low-double-digit to mid-teens rates, with profits growing faster because the fixed costs are already paid. They used the sell-off caused by Middle East tension and fuel-price fears to buy.

RCI Hospitality runs nightclubs and restaurants and, importantly, owns most of the buildings they occupy. The shares are held down by a legal dispute in New York.

Ace River's argument is that the market has fixated on the lawsuit and stopped valuing everything else: the cash the business actually generates in a normal year, the property portfolio, and a long record of management buying back stock and reinvesting sensibly rather than empire-building. On conservative assumptions they think the shares are worth materially more than the current price.

There are two routes to that value being recognised: continued buybacks steadily raising the value of each remaining share, and an eventual settlement removing the overhang altogether. Their conviction rose during the period, and it is one of the fund's biggest positions.

U-Haul is the do-it-yourself moving business — the orange trucks — plus a large and growing self-storage operation, across nearly 22,000 locations. Note that this pitch is for the non-voting share class.

Most of the profit comes from one-way rentals, where you pick a truck up in one city and drop it off in another. That is much harder to run than it looks, because trucks accumulate in the wrong places, and solving it requires both a dense network and sophisticated pricing to nudge demand back the other way. Hotchkis argue this is what has let U-Haul earn better returns than competitors for decades.

Two other sources of value: the self-storage business, where occupancy is rising and existing customers are paying below market rates, and the property itself, which is carried cheaply. The reason the shares moved is that investors are finally looking past the heavy depreciation charges from vehicles bought during the pandemic to normalised earnings, lower spending and cash returned to shareholders.

GFL is the fourth-largest waste company in North America. Ave Maria like it for the classic reasons in that industry: routes get cheaper per stop the more customers you have on them, the founder still runs it, and rubbish collection does not stop in a recession.

Its particular advantage is size. GFL is big enough to be efficient but small enough that buying a mid-sized regional operator still moves its numbers — something the industry giants cannot say, since no available target is large enough to matter to them.

The reason it appears in this letter, though, is as evidence for a different position. GFL agreed to buy Secure Waste Infrastructure, which disposes of water produced by oil wells in western Canada. Ave Maria's argument had been that such a business should be valued like a waste company rather than an oilfield-services company — and a waste company paying up for it proves the point, which should also help their stake in a similar private business.

Tencent owns WeChat, the app through which a billion Chinese people message, pay, shop, book taxis and read news, and it is also the world's largest games company, owning Riot, Epic and Supercell. Vision Capital's numbers convey how unusual its position is: WeChat has 1.4 billion monthly users and takes about 76% of social-app time, while Tencent's apps together take roughly 55% of all time spent on phones in China.

Their description is that it is WhatsApp, PayPal, Instagram, Uber, Amazon and your bank in one app you never leave — something no company anywhere has replicated. The defences reinforce each other: the network is more useful the more people use it, switching costs are high, and each new service launches into an audience Tencent already owns.

On the worry that Tencent was slow to artificial intelligence, they argue that is a matter of timing, not capability, and that Tencent is running its usual playbook of copying quickly then out-innovating. Its breadth across phone, PC and cloud is an advantage for AI assistants that need to work across many apps. Their model — mid-teens growth in spare cash, a range of exit multiples, and a 20% discount on Tencent's large portfolio of stakes in other companies — produces annual returns of 15-32%, with 24% as the midpoint. The founder still owns 8.8% and still turns up after 28 years.

Hoa Phat is Vietnam's largest steelmaker, with over 35% of the construction-steel market. Harding Loevner own it as a cost story rather than a commodity bet: it runs a fully integrated operation with modern, efficient blast furnaces, which makes it the cheapest producer in the country, and its plants sit near the biggest sources of demand with a dense distribution network attached.

In a commodity business, being lowest-cost means you keep making money when prices fall and smaller rivals do not — which is how Hoa Phat has kept gaining share. Growth is already banked rather than hoped for: a new complex completed in late 2025 lifted capacity by roughly 60%, arriving as Vietnam spends heavily on infrastructure. Vietnam has also just imposed tariffs on Chinese steel imports, which should let domestic producers charge more.

Their framing is that this is a rare thing — a quality growth business inside a cyclical industry — which also diversifies a portfolio full of the usual emerging-market technology names.

INOX India makes cryogenic equipment: the heavily insulated tanks and systems needed to store and move gases that only stay liquid at extremely low temperatures, including liquefied natural gas and industrial gases. It has around 60% of the Indian market.

Baron's first reason is energy security. India wants to import less oil from the Middle East and more gas from elsewhere, and every LNG cargo needs tanks, terminals and transport equipment at the receiving end. The second is exports: as global companies spread their supply chains beyond one country, an Indian manufacturer with a reputation for quality wins work, and INOX already supplies Air Liquide, Linde and Air Products.

Two recent contracts illustrate the direction — a small LNG terminal in the Bahamas, and a cryogenic storage contract with what Baron believes is SpaceX, which needs vast quantities of super-cooled propellant. They model 15-20% compound growth in revenue and profits over three to five years.

Precision Wires makes enamelled copper winding wire — copper wire coated in insulation, wound into coils inside transformers, generators and electric motors. It is the largest producer in India with about 30% of the market, selling into cars, aerospace and defence, power, electronics, appliances and infrastructure.

The appeal is that it is an unavoidable input into two things India is doing at once. The country is adding electricity generation capacity quickly, and every generator, transformer and substation consumes this wire. Separately, electric vehicles need a specialised grade of the same product, and India's EV market is growing from a small base.

Baron model 15-20% compound revenue growth over three to five years, resting on scale, the ability to develop new grades, and long relationships with the manufacturers that design it into their products.

Theon makes night-vision goggles and thermal-imaging equipment for soldiers — Greek-run, listed in Amsterdam. Amati bought it as a direct play on Europe rearming.

What makes it well positioned is where the money is going: not tanks and aircraft, but the equipment carried by individual soldiers, battlefield communications and drones. Those are the categories European governments have committed to fund over the coming decade, and there is a parallel political commitment to buying from European manufacturers rather than American ones.

The origination detail is worth noting. Amati had followed the company for some time without acting; it took a meeting at the Eurosatory defence exhibition in Paris to convince them that changes inside the business were being underestimated by the market. Following a name is not the same as owning it — something has to convert the interest.

Volati is a Swedish group that buys small industrial businesses and improves them. Remarkably, it has done this since 2003 without ever issuing new ordinary shares — so every gain has accrued to existing owners rather than being diluted away.

In June it split in two, spinning off Salix. The market gave most of the value to Salix, and REQ think what remains is badly mispriced. The reason is concentration: one business, Ettiketto, produces roughly three-quarters of the remaining profits, and REQ argue it alone could be worth as much as the entire company's current market value.

Ettiketto makes self-adhesive labels and labelling machines, and has grown from a Nordic to a European supplier by acquisition. It has the features REQ look for in a serial acquirer: a fragmented industry with plenty of targets, repeat demand, good returns on capital, and a demonstrated ability to improve what it buys — the Swedish business it acquired in 2012 went from low-teens margins to above 20%.

The rest of Volati is currently earning below its potential because agriculture and construction have been weak, not because it has lost customers. REQ's summary of the mistake: the market prices Volati as a cyclical industrial company, when it is really a high-quality compounder hidden inside one.

Infineon appears in this compilation as a long pitch, but the passage attached to it is the only genuinely two-sided argument in the entire document, and it is worth reading precisely for that.

The constructive half: every chipmaker, in logic and memory alike, is committed to expanding capacity and has the profits to pay for it. More interestingly, the three big memory makers have all said they will be disciplined — Samsung explicitly promised more restraint than in past cycles, Micron said its expansion would follow customer demand. Harding Loevner's inference is that prices and margins could therefore stay high for longer than history would suggest.

Then the reversal. The very demand producing those profits is also funding the new capacity, and capacity eventually arrives. "For the companies, such investments may be rational. For the stocks, it could prove more complicated." Scarcity has driven both the earnings and the enthusiasm; if supply catches up, both change. That is a direct challenge to the eight other memory and AI-hardware pitches in this same compilation, and the reason this row is marked Neutral rather than Positive.