The biggest discounts are the slowest to close

AI ties much of the list together

Nearly half the list depends on AI in one way or another: 23 of the 48 picks have at least one AI theme. Five are spenders: GOOGL, MSFT, META, AMZN and BABA. Nine are suppliers: NVDA, AVGO, TSM, MRVL, SNPS, QCOM, CEG, SPGI and BN. Ten could lose business to AI tools: NOW, SAP, INTU, CSU, RELX, EXPGY, BR, SCHW, APP and GOOGL. GOOGL appears twice, because it spends heavily on AI while AI answers can also take clicks away from Search.

The spenders and the suppliers stand on the same ground. The data center budgets of MSFT, AMZN, GOOGL and META drive much of the AI demand at NVDA, AVGO, TSM and MRVL. SPGI earns rating fees when companies sell bonds to pay for data centers, and BN and CEG sell power and sites to AI customers. Together the two groups make up 14 of the 48 picks, and a cut in the spenders' budgets would reach all 14 at the same time. The list as a whole does best if AI spending continues and pays back.

The market is far more worried about the companies AI could replace than about the ones paying for it. The median discount is 16.9% for the spenders and 15.9% for the suppliers, but 26.6% for the companies that AI tools could hurt. RELX and BR carry some of the largest discounts in that group, and only several years of steady growth can prove the fear wrong.

Companies: GOOGL, MSFT, META, AMZN, BABA, NVDA, AVGO, TSM, MRVL, SNPS, QCOM, CEG, SPGI, BN, NOW, SAP, INTU, CSU, RELX, EXPGY, BR, SCHW, APP

Where the fair value and Wall Street analysts disagree

Analysts mostly like these stocks. On the LSEG scale, which runs from 1 (strong buy) to 5 (strong sell), the median rating is 1.97, and 24 of the 47 rated picks sit below 2. Nine picks are rated 2.5 or higher, closer to hold than to buy.

Six of those nine also trade at least 25% below their fair value: CPB, CLX, HONA, HSY, POOL and PEP. Four of them are consumer staples. CPB and CLX are the only picks rated above 3, and they have the largest and the third largest discount in the list. The fair value analysts see wide moats and a temporary problem. The Wall Street analysts see falling volumes, high debt and, at CPB, guidance for a second year of decline. A split this wide can point to a value trap, where the price stays low because earnings keep falling. Unit volume is the test. PEP reports on October 8, followed by HSY (about October 30), CLX (about November 3) and CPB (about December 3).

On six picks the two sources agree, with a discount of at least 30% and an analyst rating below 2: AVGO, APP, BABA, WYNN, RELX and YUMC. Agreement does not make a gap close faster. Only AVGO and APP have a gap that a dated event can close. BABA and YUMC carry country risk, and WYNN and RELX need several years.

At the other end, NVO and QCOM trade at a small discount or a premium, and analysts rate them near hold. Neither source sees much upside there.

Companies: CPB, CLX, HONA, HSY, POOL, PEP, AVGO, APP, BABA, WYNN, RELX, YUMC, NVO, QCOM

Discounts that will probably not close

A large discount, on its own, says little. Of the eight largest discounts in the list, only AVGO's and HONA's can close through a dated event within about two years. Among the picks below fair value, the median discount is 21.2% where the gap can close, 31.5% where it closes slowly, and 40.2% where it will probably not close. A gap can also shrink from the other side: in recent weeks the fair value analysts cut the fair value for four picks (CEG, CPRT, POOL and VNOM) as their prices fell.

Eight picks have a gap that will probably never close: BABA, YUMC, TSM, PCG, BTI, BN, SUGR and COST. For these, a ranking by discount makes the opportunity look larger than it is.

Country risk is the most common reason. BABA and YUMC depend on China, and US-China tension can move their prices whatever their results show. BABA also trades through a VIE structure that gives ADR holders no direct ownership. TSM makes most of its advanced chips in Taiwan. Strong results can close the part of its gap that comes from execution, but not the part that comes from Taiwan-China risk.

The other reasons are legal or structural. PCG faces open-ended wildfire liability in California, and on September 1, 2026 the Assembly adjourned without a vote on the reform bill. BN is a holding company with many layers, and the layers remain after its merger. Many investors do not own tobacco, so the market will probably always pay less for BTI's cash flow. At SUGR, the founder controls the company, the stock trades very little, and minority holders cannot block a bid from Beta San Miguel.

COST is the opposite case. It trades above its fair value, and investors have paid that premium for many years in return for its predictable membership profit. The premium can get smaller if growth slows, but nothing in the research suggests the market will stop paying for that certainty.

NVO and HSY also have controlling owners, a foundation and a trust, but their discounts come from business problems that can change, so they are not in this group. WYNN (Macau), CEG (a Crane restart that can slip to 2031) and VNOM (Diamondback decides the drilling) should close slowly, yet part of their discount may also last.

Companies: AVGO, CPRT, POOL, BABA, YUMC, TSM, PCG, BTI, BN, SUGR, COST, NVO, HSY, WYNN, CEG, VNOM

Higher rates test the borrowers

The Federal Reserve raised rates in September 2026, and the 10-year Treasury yield went above 5%. That hurts the picks that carry a lot of debt, and the ones that investors value like bonds.

Debt is a main risk for 12 picks, and leverage is highest at SUGR, CPB and CLX. SUGR's leverage ratio is 5.5 times, and most of its bank lines are uncommitted. CPB's is about 4.9 times on Moody's measure (about 4.3 times on its own), and Moody's keeps a negative outlook. CLX's net debt is about 4.2 times its fiscal 2026 EBITDA after the GOJO purchase. All three sell to price-sensitive shoppers, so weaker sales push the debt ratio up.

Some picks must refinance soon, at higher rates. WYNN still has about $1.35 billion of debt due in 2027 after its September refinancing, and holders can put back $600 million of convertible bonds in March 2027, and YUM has about $1.6 billion due in 2027. YUMC must refinance a bridge loan of about $1.2 billion within about a year of August 2026. HONA carries about $16 billion of debt from its spin-off and pays no dividend yet. META and BN now borrow at higher rates than before, and BN's real estate business has $64.8 billion of property debt with an average term of only 2 years.

Rates also change value directly. Higher rates lower what investors will pay for the steady royalties of MCD and YUM. They slow new mortgages at ICE and new pools at POOL, and they lower the value of BN's listed holdings. Two picks gain, because SCHW and MRX earn interest on client cash. On balance, more picks lose than gain when rates rise.

Companies: SUGR, CPB, CLX, WYNN, YUM, YUMC, HONA, META, BN, MCD, ICE, POOL, SCHW, MRX

The next five weeks

Between October 1 and November 5, 2026, 30 of the 48 picks report earnings. The companies have confirmed 15 of these dates, and the other 15 are estimates. By the end of November, 42 picks will have reported.

PEP opens the season on October 8 with an early view of the US consumer. UNH follows on October 13, then TSM and SCHW on October 15. TSM is the first AI supplier to report, and what it says about AI demand and 2027 prices will be a first signal for the whole supplier group. SAP (October 21) and RELX (October 22) are the first tests of the fear that AI tools will replace software and data. PG, PCG and POOL also report on October 22.

The last week of October is the busiest, with 12 reports from October 27 to October 30. It includes the four US AI spenders: GOOGL and MSFT about October 27, and META and AMZN about October 29, all four dates estimates. Their capital spending plans for 2027 matter more to this list than any other numbers, because the AI suppliers depend on them. NOW and HONA report on October 28, and ICE on October 29, the same day as the MarketAxess shareholder vote.

From November 2 to November 5, nine picks report, mostly companies that sell to consumers: YUMC, CLX, WYNN and MCD, along with VNOM, BR, OTIS, NVO and APP. By November 5, eight of the ten weak-consumer picks will have reported.

Most AI suppliers report later, after their customers have set out their plans: NVDA about November 17, SNPS about November 26, MRVL about November 27 and AVGO about December 10. Two other dates fall in the window: the MRVL Investor Day on October 6, and the FDA decision on NVO's CagriSema, due in Q4 2026.

Companies: PEP, UNH, TSM, SCHW, SAP, RELX, PG, PCG, POOL, GOOGL, MSFT, META, AMZN, NOW, HONA, ICE, YUMC, CLX, WYNN, MCD, VNOM, BR, OTIS, NVO, APP, NVDA, SNPS, MRVL, AVGO