Barron's — Midyear Roundtable: 45 Stock Picks From Top Investors

DoorDash delivers about two-thirds of America's restaurant orders and is expanding into groceries and retail. When Anthropic's latest AI model spooked the market about anything digital, DoorDash got lumped in — but Ellenbogen argues its moat is physical, not digital: a dense network of drivers that AI can't replicate, which already makes each delivery ~$4 cheaper than in 2020. With Europe's Deliveroo accelerating, grocery catching on with younger customers, and a new tech platform about to boost margins, he sees earnings passing $10 a share by 2028 and the stock going from ~$188 toward $300.

An Italian software house that just went public (July 1). Its model: buy neglected, slow-growing apps cheaply (like Evernote), strip them down to a small team on a shared technology platform, and run them for cash — Evernote's profit margin went to ~60% with fewer than 20 people on the app. Ellenbogen invested pre-IPO and thinks free cash flow can compound ~30% a year; today's AI-driven fear of software companies is actually a gift, letting Bending Spoons buy better businesses cheaper. Target $45–55 vs ~$35 today. (The article gives no ticker symbol yet.)

Ferguson distributes plumbing, HVAC and building products — a boring-sounding middleman business where scale wins: bigger buyers get better prices from suppliers and can keep more inventory in stock, which is exactly what contractors pay for. Its warehouses look like Amazon's, and it keeps taking market share in both home construction and commercial projects (including data centers). At 18x next year's earnings — cheaper than the index — Ellenbogen sees mid-teens earnings growth and a move from ~$230 to $300–350 as the economy broadens beyond AI.

Rogers thinks the toy maker is worth $25–26 a share in a private sale versus a ~$13 stock price. The brands (Barbie, Hot Wheels, Fisher-Price, American Girl) are irreplaceable; what hurt the stock was a costly push into digital games and trading cards plus some flop movies. Now an activist investor is publicly urging a sale — possibly to Hasbro — and Rogers thinks today's merger-friendly regulators make this the moment. Meanwhile the company keeps buying back its own cheap shares.

JLL is one of the world's two big commercial-real-estate service firms (with CBRE) — leasing buildings, brokering sales, managing properties, running a money-management arm. The stock is flat because investors fear AI will replace brokers; Rogers sees no evidence of that in complex deals, and notes JLL is using AI to cut its own costs. At a ~35% discount to his ~$450 estimate of what the business would fetch privately, with double-digit earnings growth, he prefers it to the pricier CBRE.

The lawn-and-garden leader — #1 brand in every key category — lost a decade to a disastrous detour into hydroponics (cannabis-growing gear). That era just ended: longtime CEO Jim Hagedorn stepped down in June, and Rogers has met the new CEO Nate Baxter repeatedly and likes his back-to-basics plan. At ~14 times next year's earnings (a ~20% discount to what Rogers thinks the business is worth), with steady demand, cost cuts and more online selling ahead, it's his kind of unglamorous bargain.

Zebra makes the barcode scanners, RFID readers and rugged handhelds that run warehouses, stores and delivery fleets. Earnings are growing double digits — demand rises as retailers and logistics firms chase efficiency — but the stock's multiple collapsed from over 20x to ~14x on one fear: AI demand is sending memory-chip prices soaring, and chips go into everything Zebra sells. Rogers says the company is managing it (pricing, buying ahead, redesigns) and the stock now sits at a 40% discount to his $390–400 private-market value.

Part of Giroux's argument that small and mid-size biotechs are the market's best hunting ground: big pharma faces a $400–500 billion revenue hole as blockbuster patents expire, and can't invent enough internally — so it must buy. Cytokinetics is his proof the picks work even without a takeover: he recommended it in January in the $30s; a successful late-stage heart-drug trial that doubled its addressable market has it in the $80s. It remains on his list of seven biotechs he'd be "shocked" weren't acquired at 50–100% premiums.

Alnylam leads a field called RNA silencing — drugs that switch off the specific gene causing a disease. Its franchise treats a widely underdiagnosed heart condition (TTR amyloidosis); it has the best drug today and an even better one coming in 2028–30, with patent protection out to 2045, plus a pipeline that could produce several more blockbusters. At $37 billion it's a big bite for an acquirer, but Giroux says if he ran Merck or Lilly he'd "buy it in a heartbeat" — and even unacquired, he expects it to be one of the decade's best-performing large pharma stocks.

Cencora (formerly AmerisourceBergen) is one of the giant middlemen that move drugs from manufacturers to pharmacies. Here's the twist: the same patent cliff that terrifies big pharma is Cencora's best friend — when branded biologics go off-patent, distributing the cheaper "biosimilar" copies is where Cencora makes its fattest margins. A string of one-off issues knocked the stock down ~15% this year and its valuation from 20x to 14x earnings, but Giroux sees earnings doubling in five years — with growth accelerating in 2028–29 as a wave of Medicare blockbusters loses protection. Bonus: it's one of the most defensive stocks in the market, and insiders are buying.

Aurora builds the self-driving system for long-haul trucks. Giroux thinks the industry is finally at its tipping point: FedEx says the tech is ready, Volvo has published targets of 25,000 driverless trucks by 2030 and 220,000 by 2035. The economics are blunt — no driver means ~30% cheaper operation, trucks that run twice as many hours, and much fatter margins on long routes (most of the market). Aurora partners with nearly every truck maker. His back-of-envelope: if driverless takes 20% of a 200-billion-mile market and Aurora keeps half, that's $20 billion of software-margin revenue by 2035 — potentially a $300 billion company, or "a 25-bagger." A venture-style bet with venture-style risk, sized accordingly.

ServiceNow sells the workflow software big companies run their internal operations on. It's down nearly half since mid-2025 on the fear that AI will let companies replace such software. Cohen's rebuttal: most enterprises don't want to build and babysit these systems themselves — they want a vendor responsible for them, and ServiceNow has now shipped its own AI product suites. At ~20 times 2027 earnings, she's sticking with her January pick.

Two panelists defend the German software giant that runs the back office of much of the world's commerce. The stock fell ~30–48% (currency depending) in the AI-kills-software panic, but the actual business grew: cloud revenue up 27%, order backlog up 25% in the first quarter — and only half of SAP's customers have even moved to the cloud yet. Rossbach's point: a company this embedded in its customers' operations is positioned to SELL them AI, not be replaced by it. At 14–18x forward earnings, both call the selloff a chance to buy an industry leader cheap.

A work-management software platform (think flexible project-tracking building blocks) down 74% in a year and sitting near the very bottom of its 52-week range. Cohen's tell: management is aggressively buying back its own stock at ~$73 average — a signal the people with the best information think it's too cheap. At ~13x 2027 earnings with thin analyst coverage, it's her contrarian software pick alongside the big two.

Cohen's AI-energy theme, taken abroad: data centers need power, and this Shanghai-listed company operates China's giant hydroelectric dams (it's a spinout of the Three Gorges parent). The dams are already built — so almost no new capital is needed, and profit margins run ~41%, ten-plus points above U.S. utilities. At ~18x earnings with a 3.5% dividend, she frames it simply: buy this one for the cash flow.

The sibling spinout is the growth bet: solar, wind and tidal projects under construction in China's deserts, Spain, Brazil and Peru — activities in over three dozen countries. Margins are much thinner (~13%) because it's spending heavily on the buildout, and the stock has done little since its 2021 IPO, but the projects complete over 2027–28 and the state-owned parent stands behind it. Buy Yangtze for cash flow, this one for growth.

Kao is Japan's household-products staple — detergents and personal care, plus growing international brands (Molton Brown, Oribe, John Frieda, Jergens). Cohen calls it "an early-stage Procter & Gamble" — both started as 19th-century soap makers. Her macro angle: unlike squeezed American households, Japan's middle class is finally seeing real income growth and rising confidence. Modest 4% revenue growth has been producing 12%+ earnings gains; ~20x 2027 earnings with a 2.4% yield. (U.S. investors can use the ADR, KAOOY.)

Korea's defense champion, and Cohen's way to play global rearmament. Its edge over Lockheed-style Western primes isn't margin — it's speed: Hanwha builds artillery, vehicles and aerospace systems on much shorter timelines, which matters enormously when countries are racing to restock arsenals. It manufactures in Korea, Australia and Europe, revenue is growing rapidly, and returns on equity run 21–22%. The trade-offs: thinner margins (~15%) and a token dividend versus the entrenched, long-contract Western contractors.

Black's classic value screen — high return on equity, low absolute P/E, gushing free cash — lands on the world's #2 online travel agency (Expedia, Hotels.com, Vrbo, Orbitz; together with Booking they control 46% of online travel). Strip out the cash on its balance sheet and you're paying under 12 times this year's earnings for a company growing profits 17–18% and buying back stock by the billions. Booking is the better business — but not so much better that it deserves a P/E half again as high.

Black likes that Urban Outfitters reports honest, unadjusted numbers — "what you see is what you get." Three store brands target three demographics (Urban Outfitters for young adults, Free People for bohemian 18–34s, Anthropologie for higher-income women), and Nuuly — a clothing-rental subscription with over 500,000 subscribers — is the growth kicker. His model has earnings growing 20%+ while the stock trades at ~11.5x, with excellent inventory management and up to $400M of free cash flow this year.

Jain's arithmetic for why the big oil companies are the market's cheapest quality assets: at $70 oil, TotalEnergies pays a 6% dividend, retires 4% of its shares each year — a 10% annual return before any growth — and still trades at just 8 times earnings. His view is oil goes higher (the Iran war effectively gives Tehran control of the Strait of Hormuz, the U.S. strategic reserve is at a 43-year low, shale output is plateauing), but the point of the math is that the stocks work even if he's wrong. Energy also doubles as portfolio insurance: it's the sector that benefits from exactly the inflation that would hurt everything else.

The largest listed U.S. water utility — and Jain's favorite among his defensive keeps precisely because it has nothing to do with the power-generation frenzy. Water rates grind higher, earnings compound 8–9% a year like clockwork, and at 20–21x earnings he calls it "almost like a bond" — the sleep-at-night asset for a market he thinks is led by unsustainable cyclicals.

Vulcan sells crushed stone, sand and gravel — the literal foundation under roads, bridges, homes and data centers. The moat is geology plus geometry: rocks are too heavy to truck far, and new quarries are nearly impossible to permit, so owning quarries near growing cities is a local monopoly ("America's tollbooth on concrete"). Profit per ton has climbed from $7 to $11 since 2021 and Ahlsten sees $15 by 2028, because Vulcan can raise prices faster than costs. With state highway programs, data centers and reshoring all pouring concrete, he targets $400+ in three years vs ~$313.

Linde supplies industrial gases — oxygen, nitrogen, hydrogen — piped "over the fence" into customers' plants on long-term contracts where customers pay whether they take the gas or not. That makes it a de facto utility, without the regulator, spread across chips, healthcare, food and clean energy. The quirky kicker: rocket launches. Linde's space business has quadrupled in three years and could pass $1 billion by decade-end — "if you missed the SpaceX IPO, there is Linde." Never cheap (28x) because it never disappoints; Ahlsten's target is $654 by end-2028 vs ~$540, an 11%+ annual compounder.

Ahlsten's clever AI angle: Alphabet's AlphaFold can now predict protein shapes, which massively speeds up drug discovery — but every AI-predicted protein still has to be tested in an actual lab. Danaher sells the lab tools and biologic-manufacturing equipment that testing requires, so more AI-generated drug candidates mean more Danaher customers, and possibly a bottleneck in exactly what it sells. The sector is just emerging from a funding trough, the stock trades at only 20x versus its history, and he sees a re-rate to 25–30x — $194 to $290+ in three years.

Arista makes the network switches and software that shuttle data between chips inside giant data centers. Rossbach's sequencing insight: hyperscalers have been consumed by permits, construction, chips and memory — but once those clear, the constraint moves to the network, because latency wastes everything else you built. Arista already serves Meta and Microsoft and is expanding with Anthropic and Oracle; guidance of ~28% revenue growth is supported by building deferred revenue and purchase commitments. It's expensive (41x) but he argues 35%+ earnings growth justifies it.

The French counterpart to Eaton: Schneider makes the electrical gear and software that manage power — increasingly for data centers, where it's one of the world's largest suppliers (and it bought control of Motivair, a liquid-cooling specialist, since AI chips now run too hot for air). Beyond AI it rides electrification, grid renewal and factory automation; its weak spots (residential, automation) are cyclical and give extra upside when rates fall. Rossbach expects 13.5% annual earnings growth for five years and considers 25x fair for that. (U.S. ADR: SBGSY.)

Witmer's favorite kind of setup: a good company whose stock fell for mechanical reasons, not business reasons. Brink's (armored trucks, cash logistics, smart safes) agreed to buy NCR Atleos (ATM networks) — a strategically sound, earnings-boosting deal. But the stock sold off anyway: dividend-and-buyback investors left when capital returns paused for the deal, and merger arbitrageurs are shorting Brink's stock as part of the deal math. That's forced selling, not a verdict on the business.

Meanwhile the underlying story improves: the Smart Safe subscription business (shops deposit cash into an on-site safe and get instant bank credit) is higher-margin and growing, and merging the two companies' ATM networks should save $200M+ a year. Witmer models $18–24 of free cash flow per share by 2029 — against a ~$101 stock. Even at a modest 10–11x multiple that's $180–250 in about two years.

The world's big glass-bottle maker, at under $10 a share. The turnaround: a new CEO has slashed costs and closed excess plants — and competitors are closing capacity too, so the whole industry is running fuller, which matters enormously in a fixed-cost business. Glass is winning some business back from aluminum cans now that its costs are lower. This year's guidance cut was mostly the Iran war's energy costs, which get passed to customers with a lag. A tax-law change (interest deductibility) means O-I will generate more cash than its reported earnings for years — cash that pays down its $4.7B debt, which lowers interest, which raises earnings. Witmer sees ~$2 EPS next year rising past $3 by 2029, and a $25–30 stock within a couple of years.

The rare name argued both ways at the same table. Rossbach picks it: Eaton's power-management gear is indispensable to data centers (orders up 240% in a year), it's spinning off its vehicle unit, co-designing power systems with Nvidia, and rides decades of grid renewal and reshoring beyond AI. Giroux names it among the AI-derivative industrials he'd avoid: the market is paying peak multiples on what will later look like peak (2029–30) earnings, and when AI capex growth slows, those multiples compress. Own it for the decade-long electrification story only if you can stomach the AI-capex air pocket Giroux expects.

The private AI company is the article's gravitational center. Ellenbogen: since Claude Opus 4.5 shipped in late November, Anthropic's revenue exploded from ~$9 billion to over $50 billion in six months, and it has confidentially filed for an IPO — which would finally let investors own the AI-model layer directly instead of through proxies, and force the market to judge which AI companies have durable cash flows. Giroux adds a twist: Anthropic built the world's best model on cheap Google and Amazon chips, proving Nvidia is optional. Jain supplies the caution: Anthropic and OpenAI's spending props up much of the AI buildout, and open-source models now serve two-thirds of all usage — if the money engines sputter, the whole capex chain feels it.

Giroux's valuation autopsy: Caterpillar spent a decade trading at ~16 times earnings; it now trades at 38 times because investors love its data-center generator business. Unpack the math and the market is paying roughly 230 times that business's current earnings — and about 70 times what he thinks will prove to be its PEAK earnings around 2030, since Caterpillar, Cummins and others are all adding capacity into the boom (capacity that arrives just as AI capex growth slows). "This won't end well."

The panel's most consequential negative. Giroux's argument: Nvidia's chip monopoly is ending — Anthropic built the world's best AI model on cheap custom chips from Google and Amazon, and as AI shifts from training models to running them (inference), those cheaper chips compete head-on with Nvidia's GPUs. The "neocloud" companies whose whole business was renting out Nvidia GPUs lose their reason to exist as big customers bring the work in-house. Ellenbogen notes the stock has already lagged other AI names for a year as the model builders diversified. Black's lone counterpoint: at 22x this year's earnings it isn't expensive — the debate is about what happens to the "22x" part.

Desai closed her gold recommendation, and the reasoning matters more than the call: gold's great run was powered by the "dollar debasement" trade — the belief that the Fed would let inflation run and erode the currency. A genuinely hawkish Fed chair (Warsh, whom she compares to Volcker) undercuts exactly that thesis, and profit-taking has begun after years of outsize gains. She still grants gold some value as a diversifier — just no longer a favorable setup.

A model of how to close a losing pick. Ahlsten recommended Boston Scientific in January; two things broke: Medtronic's new products took share in the heart-ablation business he thought Boston would dominate, and a key clinical trial for its Watchman stroke-prevention implant didn't produce the hoped-for result. That means a quarter of the company that was growing 20%+ stopped delivering, and the market permanently marked down the multiple. Rather than hope, he names what he got wrong — "we got the industry structure wrong, a trial didn't work" — and removes the pick.