The companies we rate Attractive, best first. The bar shows the price against what we think the business is worth.
One of the two strongest businesses here, at about half our value, with nothing reporting soon. Taiwan and two large customers are risks the value cannot price, which keeps conviction medium.
Most of its score comes from the room before the Neutral line.
The same news moves it with AVGO, NVDA. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #1
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseEvery AI chip maker needs the newest node and the packaging beside it, and only TSMC ships both at volume; the case needs AI spending to keep rising through 2027.
Demand for the newest chips keeps growing past 2027, and margins stay well above half of sales despite the overseas fabs.
About half our value, near the bear case and tied to no other name here: the cleanest step away from the AI chip bet. The rack arm's use of cash is the doubt.
Most of its score comes from the room before the Neutral line.
The formula alone would place it #7
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseCompanies and cloud providers are buying computers for AI faster than ever; TD SYNNEX must turn the stock it built for that demand back into cash.
Hyve's profit is worth less than distribution's: it needs more inventory for each dollar earned, and its buyers can walk.
Under half our value and close to the bear case, with defense work steadying the AI boards. It moves with Celestica and Sanmina, and it is the soundest of the three.
Most of its score comes from the room before the Neutral line.
The same news moves it with CLS, SANM. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #2
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAI servers need boards with more layers and defence budgets favour missiles and radar; TTM must deliver its new boards on time and beyond two customers.
A strong year built on one new board for few customers, in a business rivals can copy, now priced below management's own target.
Wide room on a solid business, and the latest report raised the outlook. Its largest new customers lose money, its bet is priced for success, and with Taiwan's chipmaker it is one AI bet.
Most of its score comes from the room before the Neutral line.
It loses most on the bet.
The same news moves it with TSM, NVDA. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #9
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseThe biggest AI builders want chips designed for their own work, and Broadcom builds them over years together, so each new data centre they power up buys more from it.
Next year's sales are more contracted than the price credits: Broadcom has bound itself to $52.7B of 2027 parts, and Google to 2031.
The strongest of the contractors at about three-fifths of our value. Customer prepayments flatter its return, but the bear case sits near the price; EMCOR repeats this bet below.
Most of its score comes from the business.
The same news moves it with EME, FPS. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #6
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseData centres are growing while other building shrinks, and each needs far more piping and cooling; Comfort Systems must keep its margin and prepayments as that growth slows.
The argument is about the wrong line. Growth is visible in the booked work; how much of the margin is customer cash is not.
About half our value and below the bear case, and it moves with nothing else on the page. China, a large new rival mine and Brazil's government are real risks, already in the price.
Most of its score comes from the room before the Neutral line.
The formula alone would place it #3
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseCheap new copper from mines Vale already runs adds a growing business the iron ore price does not touch.
Fair at today's price: the ore business is good but under pressure, and nothing yet shows its costs or premium turning.
About half our value with the business accelerating, but a large share-funded purchase is pending and custom chip work rests on management's word. Wide room, ranked below the cleaner cases.
Most of its score comes from the room before the Neutral line.
The same news moves it with TSEM, FN. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #10
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseCloud companies are building their own AI chips and faster links between them, and Marvell designs both; it must keep winning each new generation against Broadcom.
Next year is priced in; the year after, now at a discount, rests on programs re-won against Broadcom, with upside owed to Google and Lumentum's holders.
A little over half our value and tied to no other name here, with costs steady while sales jumped. It swings with chip spending, so the room has to carry a cycle.
Most of its score comes from the room before the Neutral line.
The formula alone would place it #16
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAI chip demand is driving chipmakers to buy more factory equipment, which carries MKS parts; its profits must hold up when that spending stops rising.
The profit jump looks mostly like fixed costs meeting a spending peak; Q3 and the Q4 guide fit a lasting change too, and the price counts on the peak lasting.
Payroll software customers rarely leave, asking only modest growth. Less room than the names above, but a steady business and a latest report that met our test.
Most of its score comes from the business.
The same news moves it with ADBE. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #24
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAmerican employers are adding almost no jobs, so Paycom gains only by replacing HR paperwork with its software; its client count must keep growing.
The margin step and share cut were one-time and are now in the price, while the company's own forecast shows growth slowing again.
Record signed work, no debt and cash matching profit, but a thinner cushion than Comfort Systems above and the same news moves both. A second helping of one bet, not a new one.
Most of its score comes from the business.
The same news moves it with FIX, FPS. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #12
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseData centres are the one part of American building still growing while the rest shrinks, and EMCOR moves its electricians and pipefitters to wherever that work is.
A machine that moves bonded crews to whichever market grows, priced about fair on growth that fades and a margin a little below today's.
A solid business under two-thirds of our value, ranked lower only because its report is days away. Read the print first; it shares the AI bet with the two chip names above.
Most of its score comes from the room before the Neutral line.
The same news moves it with TSM, AVGO. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #4
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseClouds and AI labs are building data centres faster than chips can be supplied, and most of them run on NVIDIA's computers, so its sales keep climbing.
The shares are cheap only if next year's growth is real demand; we think most of it is, but NVIDIA is financing part of it.
About half our value and on the bear case, but the value needs two TV operators to resume paying and the timing is unknowable. Cheap waiting, ranked below cases that do not wait.
Most of its score comes from the room before the Neutral line.
It loses most on the bet.
The formula alone would place it #28
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseChipmakers are moving to hybrid bonding to stack AI chips, and if they need Adeia's patents to do it, chip royalties could grow several times.
The recurring base is smaller and more fought over than the headline, and the price already pays for the recovery and the chip target.
About two-thirds of our value and near the bear case, but a rival leads both its markets and its main drug's protection runs out within years. Priced for decline, not for a reversal.
Most of its score comes from the room before the Neutral line.
The formula alone would place it #23
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseMore people will treat obesity with medicine for years; if Novo's next drugs work better than Wegovy, it keeps patients it would otherwise lose to Lilly.
The fade is real but already in the price; Novo still gains patients, so the decline may be slower than feared.
Wide room on a mixed business with record profit, but the bear case sits under the price because write-downs on loans it keeps return in a downturn. It moves with Corpay and PayPal.
Most of its score comes from the room before the Neutral line.
The same news moves it with CPAY, PYPL, SOFI. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #17
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseIf banks keep handing their declined borrowers to an outside model funded by investors, Pagaya becomes the plumbing behind a large share of US consumer loans.
The models have earned an outside rating agency's upgrades, and the price holds even if profit falls well below today's.
A good business with little room, and the take on each payment dollar still falling. It moves with PayPal; of the two its business is the better and its cushion the thinner.
Most of its score comes from the business.
The same news moves it with PGY, PYPL, SOFI. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #20
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseBusinesses are moving supplier payments off paper cheques onto cards and electronic rails; Corpay must keep its volume growing without the fee on each payment dollar falling further.
Both halves are true about different things: volume is compounding, but buyers pay 19% less for each dollar moved.
The widest room on the page and under even our bear case, but extra sales barely add cash, shareholders were asked for money and three customers carry it.
Most of its score comes from the room before the Neutral line.
It loses most on the business.
The same news moves it with TTMI, SANM. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #5
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseCloud giants are filling new AI data centres with switches and server racks Celestica builds; their spending must keep rising and its extra sales must start bringing in cash.
The growth is real. The profit it is priced on is flattered by a share-price swap, and the growth now costs far more capital.
Its value rose with the forecasts and the cushion is now wide, but one rack programme carries the growth. It moves with TTM and Celestica above, so it ranks last of the three.
Most of its score comes from the room before the Neutral line.
It loses most on the bet.
The same news moves it with TTMI, CLS. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #30
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseIf AMD's rack wins buyers from NVIDIA's, someone in America must build and test thousands of them a year, and Sanmina owns the factories AMD used.
The market is right about the thin margin. At November's price it already pays for the forecast falling short, which the forecast itself does not.
Below our bear case because the price assumes this year's windfall fades. Aircraft metal is the lasting gain, but a profit lifted by a tariff and a rival's fire ranks below cleaner cases.
Most of its score comes from our conviction.
The formula alone would place it #14
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long casePlane makers are building more jets after years of shortage, and Constellium's aircraft metal rides that; most of its sales are cans and cars, which this recovery does not reach.
The earnings are a tariff, not a franchise. The market half-knows it, so the question is how long the tariff lasts.
Just past our margin of safety after a report that beat and raised its outlook. It sells into the same power build-out as the contractors above, so it adds little that is new.
Most of its score comes from the business.
The same news moves it with FIX, EME. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #29
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAmerica is building data centres for AI and the grid to feed them, and Forgent builds their power gear fast; orders must keep coming once rivals add factories.
The year ahead is bought and the next is nearly free at this price; the years after depend on orders outlasting a shortage everyone is adding capacity to end.
A strong, debt-free business priced as if Google's AI answers end its growth, but the cushion to the line is modest and the bear case sits below the price. Tied to no other name.
Most of its score comes from the business.
The formula alone would place it #11
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAdvertisers keep moving money to platforms that prove sales, and Reddit's human conversations are the audience a chatbot cannot replace.
Search visitors were the least valuable users; the signed-in users who bring most of the money are still growing.
A price that assumes growth has stopped, on a business still growing, but the price sits only at our bear case and a report is close. Worth reading after the report, not before.
Most of its score comes from the room before the Neutral line.
The same news moves it with PAYC. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #13
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseBusinesses are moving work into AI tools, which threatens where tasks begin; Adobe wins if people still finish their documents, images and videos in its apps.
So far the slowdown is a pricing choice, not lost demand: both customer groups still grew faster than 10% last quarter.
On the line of the margin of safety with the bear case close, and the checkout button losing ground. Cash flow and buybacks are paid for; growth is not.
Most of its score comes from the room before the Neutral line.
The same news moves it with PGY, CPAY, SOFI. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #19
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseIf shoppers let AI assistants buy for them, those assistants need a trusted way to pay, and PayPal already holds the accounts and the merchants.
We agree the button is slipping, slowly, and think the price already assumes it, so neither cheap nor a trap.
Just past the wider cushion we asked of a cyclical whose first big order is unsized. Room is thin, so it waits below names with more.
Most of its score comes from the business.
The formula alone would place it #27
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAs AI racks switch from air to water cooling, every tray needs sealed couplings; Helios can win a share if buyers approve it beside makers already shipping.
The debt repair is the lasting gain. The earnings recovery is mostly done, and the data-centre option is weaker than its billing.
Only just past the deeper cushion we set for it, after its view changed three times in a month. A lender whose cheapness needs the next report to hold.
Most of its score comes from the business.
It loses most on the bet.
The same news moves it with PGY, CPAY, PYPL. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #31
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseMembers who move their main account to SoFi keep adding products, so each new one, from crypto to business accounts, sells to people already there.
The price is not doubting management's return target. It has already paid for it in full.
Below our bear case with orders back, but much of what it makes is interchangeable and the last upturn ended with the margin down. Cheap for a reason the cycle may repeat.
Most of its score comes from the room before the Neutral line.
It loses most on the business.
The formula alone would place it #15
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAI data centres, factories and cars use more power parts as customers restock; Vishay must show those orders are real use, not stockpiling, and margins must climb.
The upturn is real and partly stockpiling, but after the fall the price no longer pays for its whole best case.
The car parts pay for the price and the data-centre turbine comes free, but sales are flat and insiders have been cashing out. A cash-return case, not a growth one.
Most of its score comes from the room before the Neutral line.
The formula alone would place it #21
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseData centres short of grid power are building their own, and buyers will take a new turbine maker if it can deliver sooner than the queue.
The car parts are worth about what buyers paid before the turbine; at today's price the turbine is free, and growth beyond it is unproven.
Capacity prepaid by customers and net cash, but the price sits just under the line, a funded rival is coming, and it moves with Fabrinet and Marvell in optics.
Most of its score comes from the business.
The same news moves it with MRVL, FN. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #26
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAI data centres need more optical links, silicon photonics replaces assembled parts, and Tower's prepaid capacity makes most of those chips.
The capacity is prepaid through 2028 and the margin is rising as the fabs fill; at the price, the next two years are underpriced.
Profit only has to hold, but printing shrinks, computers are bought on price and a report is close. Little room for a business losing units fastest.
Most of its score comes from our conviction.
It loses most for a report or test this week: we would wait for it.
The formula alone would place it #32
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseIf memory prices ease in 2027 and businesses replace ageing PCs with AI models, HP's PC margins recover while ink and toner keep paying the dividend.
Passing costs on cost HP the most buyers of any big maker; flat 2027 profit needs that to stop.
Customers rarely leave and the latest report met every test, but the price has only just reached our margin of safety and the threat from AI site builders is unmeasured.
Most of its score comes from the business.
The formula alone would place it #25
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAs more one-person firms start up and build with AI, GoDaddy can sell them Airo at the moment they buy their name.
GoDaddy's cash per share keeps rising for several years on modest growth and buybacks, even if customer numbers stay flat.
A sound fleet with almost no room before the line, and Texas power prices softening. It ranks near the foot until the room comes back or the value rises.
Most of its score comes from the business.
The formula alone would place it #18
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAI data centres, factories and a growing population are raising power demand in Texas and the eastern grid faster than in decades, and Vistra's existing plants sell into it.
The reported swings are mostly hedge accounting. On what the business actually earns, the market has not paid for the AI story at all.
Engine repair alone could cover the price, but debt is heavy against thin equity and much recent profit came from asset sales. The weakest balance sheet of the names with room.
Most of its score comes from the room before the Neutral line.
It loses most on the business.
The formula alone would place it #22
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseData centres cannot wait years for the grid or a new turbine, so they build their own gas power, and a rebuilt jet engine ships now.
The engine business is real. The two-thirds of the company around it is consuming what that business earns.
A strong, fast-growing business held Attractive on a thin cushion, since customers stay and it holds net cash. One large customer and a stalled module line keep it near the foot.
Most of its score comes from the business.
The same news moves it with MRVL, TSEM. They sit in one Sectors group, so on this page they are one bet, not several.
The formula alone would place it #8
From our fair value: the price at 80% of the base case, where the view turns Neutral, against the morning review's close.
The long caseAI data centres buy far more optical links per server; Fabrinet's own AI module line must grow again for it to share in that.
The growth is real but it is not mainly AI modules, and even the lower price needs steady double-digit growth.
Room to the Neutral line, on the close of
Order judged on