Skip to content

Commentary

Every few weeks, a note that steps back from the single companies and reads the whole list together. It looks at the themes that connect them, which discounts are likely to close and which probably will not, and the dates worth watching next. It is information, not advice (see the disclaimer).

Updated

Lenders join the AI build-out as 20 picks get ready to report

More of the money for AI chips and data centers now comes from loans, guarantees and new shares, and less from the spenders' own cash. The consumer staples raise prices again and pay more to borrow. Several recent results were helped by one-time gains, and 20 picks report between October 13 and October 29.

Lenders join the AI build-out

The AI build-out now leans on borrowing and guarantees, and less on the spenders' own cash. Broadcom backs a financial partner's leases of AI racks to one customer. If that customer does not pay, Broadcom pays the difference between 85% of the amount owed and the sale value of the racks, up to about $29 billion. The same customer can also issue up to $42 billion of convertible notes to Broadcom. On October 7 Bloomberg reported early talks on about $30 billion of debt for OpenAI's Broadcom chips, and the Wall Street Journal put the figure above $50 billion. Nvidia guarantees up to $105 billion of lease and power payments for an OpenAI data center campus in Ohio, starting in fiscal 2029. Its CFO said that AI labs that Nvidia backs with its balance sheet will be roughly a quarter of its business next year.

The spenders borrow too. In the first half of 2026 Alphabet issued $51.8 billion of notes, and in June it raised $49.6 billion net from new common and preferred stock. In August it sold $25.0 billion more of notes. Amazon held about $219 billion of Anthropic and OpenAI stock and notes at June 30, 2026. Press reports say Moody's expects Microsoft's free cash flow after dividends to be about $4 billion negative in total over fiscal 2027 and 2028.

Demand itself still looks strong. TSMC's third-quarter revenue was about NT$1.49 trillion, about 50% more than a year earlier and above its own guidance. At its October 6 Investor Day, Marvell raised its fiscal 2029 custom chip outlook to more than $12 billion, from more than $10 billion. Synopsys rose 12.8% on October 1 after its fiscal 2027 guidance came in above the analysts' consensus, and it now trades only 1.9% below its fair value. The risk has moved to the credit markets: if lenders pull back, orders can slow even while end demand stays strong.

Companies: AVGO, NVDA, GOOGL, AMZN, MSFT, TSM, MRVL, SNPS

Courts and regulators set the calendar

For several picks, the next big event is a court or a regulator, not a quarterly report. Meta expects about $10 billion of legal expense in its third quarter for its settlement with 52 state and territory attorneys general. In a second New Mexico case, the state asked a judge for $35 billion to $40 billion in penalties, and the judge is expected to rule in late October, close to Meta's report.

AppLovin now faces four legal fronts at once: a securities class action filed in September 2026, an older one from 2025, a suit that San Diego County filed on October 5, and its arbitration with Unity. Its fair value stayed at $470 while the stock fell about 43% in the third quarter.

Alibaba has been on the US Defense Department's list of Chinese military companies since June 8, 2026, and it is suing to get off it. Good results cannot remove that part of its discount. In California, the Assembly adjourned on September 1 without a vote on the wildfire reform bill, PG&E suspended its five-year plan on September 2, and Fitch moved PG&E's outlook to negative on September 16. The state regulator's proposed decision on the Kincade and Dixie fire costs, due on November 13, is PG&E's next hard evidence.

Companies: META, APP, BABA, PCG

Fair values moved both ways

The fair value analysts changed several values in the past six weeks, and not always in the direction of the price. They raised Marvell to $360 from $300 after the October 6 Investor Day, their second raise in six weeks, so Marvell's discount widened to 23.5% even though the stock rose. They raised McDonald's to $295 from $270 after its September 23 Investor Day, while Wells Fargo, Guggenheim, KeyBanc and Bernstein cut their targets in the first week of October. On October 6 they raised their uncertainty rating for Copart to High, which widens the range around the $35 fair value.

At Honeywell Aerospace the two views stay far apart. After the August guidance cut, most analysts rate it Hold, with recent targets between $170 and $205, while the fair value is $256.

Experian's fair value on this site was wrong. The fair value analysts cut it to $42 from $45 per ADR on June 10, 2026, and cut the moat to Narrow on the same day, but the site showed $55 from the day the pick was added. At $35.48, Experian trades 15.5% below the correct value, not about 35% below. The October 1 note now uses the correct value.

Companies: MRVL, MCD, CPRT, HONA, EXPGY

Staples raise prices and pay more to borrow

The consumer staples, where analysts are least hopeful, now face higher costs on both their goods and their debt. PepsiCo confirmed on October 9 that it will raise prices on some snacks and drinks by single-digit percentages, and named higher fuel, aluminum and farm costs. Its prices stay below their level at the start of 2026. After its third quarter, brokers cut their targets to between $126 (Barclays) and $165 (Goldman Sachs).

Campbell's issued junior notes at 8.5% on October 5, rated below investment grade, to repay 5.20% notes due in March 2027 and commercial paper. Moody's cut it to Baa3 with a negative outlook on September 23, and the board had already cut the dividend by 36%. At 53.7%, CPB still has the largest discount on the list.

Clorox guides fiscal 2027 organic sales growth of about 3.5% to 4.5%, but more than 3.5 points of that come from an easy comparison with fiscal 2026, when an ERP change disrupted orders. Its fiscal 2026 dividends of $602 million were larger than its free cash flow of $405 million. Procter & Gamble's fiscal 2026 growth came only from price, with flat volume, and it expects first-quarter core EPS to fall 5% or more. HSY reports on October 29 and CLX about November 2.

Companies: PEP, CPB, CLX, PG, HSY

One-time gains helped recent results

Several recent results looked better than the business underneath. At AutoZone, tariff refunds added 145 basis points and a LIFO gain 105 basis points to fourth-quarter gross margin, more than its whole rise of 182 basis points. Without them, the margin fell about 68 basis points, so the site's checkpoint for that quarter now reads failed. Tariff refunds also added 4 percentage points to PepsiCo's third-quarter core operating profit growth, and they stop after the third quarter.

UnitedHealth's medical care ratio improved to 86.7% in the second quarter of 2026, from 89.4%, but $860 million of favorable changes to earlier reserves helped. Its third-quarter test is the ratio against its full-year guide of about 88.1% without large reserve releases. Novo Nordisk's first-quarter profit included a one-time DKK 26.8 billion reversal of a US rebate provision, and Novo has warned of a possible non-cash impairment in the third quarter after a failed trial.

Companies: AZO, PEP, UNH, NVO

Rates move against the borrowers

Rates kept rising for the picks that need cheaper loans. The Freddie Mac 30-year mortgage rate rose from 6.43% on July 2 to 7.40% on October 8, and the 10-year Treasury yield was 5.24% on October 9. ICE bought Black Knight for a mortgage recovery that has turned the other way, but its exchanges carry the quarter: third-quarter trading volume was 31% higher than a year earlier. Pool sells to pool builders, and new in-ground pool construction in 2025 was just below 60,000 units, about half the 2021 peak of about 117,000. Brookfield and its asset manager fell about 20% from mid-August to late September.

Schwab earns more on client cash when rates rise, but its stock fell about 6% on September 22 on fear that AI assistants such as Meta's Muse could weaken client loyalty. Its August core net new assets were 46% higher than a year earlier. Capital limits its buybacks more than cash does: its adjusted Tier 1 leverage ratio of 6.8% sits at the bottom of its 6.75% to 7.00% target.

Companies: ICE, POOL, BN, SCHW, META

The next two and a half weeks

From October 13 to October 29, 20 of the 44 picks report, and all but two of those dates are confirmed. UNH opens on October 13 with its first report since the 2027 Medicare Advantage star ratings. Analysts at J.P. Morgan and Stephens estimate that the share of its members in plans rated 4 stars or more fell to about 67%, from about 78%. Copart's offer for ACV Auctions expires on October 15, two days after the US antitrust waiting period ends, and TSM and SCHW report the same day.

SAP (October 21) and RELX (October 22) are again the first tests of the fear that AI tools will replace software and data. PG, PCG and POOL also report on October 22. October 28 is the busiest day: MSFT, GOOGL, NOW, HONA and OTIS report, META is expected to, and the Fed announces its rate decision. Microsoft guides Azure growth of 44% to 45% in constant currency, on a new definition that leaves out GitHub cloud and some security and healthcare products. Meta's profit will carry its legal charge, so its cash flow gives the clearer reading. ICE and HSY report on October 29, AMZN is expected the same day, and MarketAxess holders vote on ICE's offer.

Most AI suppliers report later, after their customers have set their plans: NVDA about November 18, MRVL about December 1, SNPS about December 9 and AVGO about December 10. All four dates are estimates.

Companies: UNH, CPRT, TSM, SCHW, SAP, RELX, PG, PCG, POOL, MSFT, GOOGL, NOW, HONA, OTIS, META, ICE, HSY, AMZN, NVDA, MRVL, SNPS, AVGO

Themes

AI disruption risk

10 companies

AI tools could replace part of what customers pay this company for.

APP, RELX, BR, INTU, SAP, SCHW, GOOGL, EXPGY, NOW, CSU

Median discount
25.4%
Average LSEG
1.96
Structural gap
0 of 10

Weak consumer

10 companies

Sales depend on households that are spending less on big or optional purchases.

CPB, CLX, YUMC, WYNN, HSY, PEP, MCD, EXPGY, YUM, PG

Median discount
27.4%
Average LSEG
2.42
Structural gap
1 of 10

New management

10 companies

A new chief executive or a reset plan must prove itself.

CLX, OTIS, POOL, HSY, CPRT, UNH, CSU, NVO, YUM, WM

Median discount
21.0%
Average LSEG
2.42
Structural gap
0 of 10

Interest rates

9 companies

Earnings or valuation move with interest rates and the housing market.

BN, WYNN, POOL, SPGI, SCHW, MCD, ICE, YUM, MRX

Median discount
22.0%
Average LSEG
2.08
Structural gap
1 of 9

AI supplier

7 companies

Sells the chips, networks or tools that AI data centers buy.

BN, AVGO, NVDA, MRVL, SPGI, TSM, SNPS

Median discount
23.5%
Average LSEG
1.84
Structural gap
2 of 7

China and geopolitics

7 companies

Exposed to China, Taiwan or export rules between governments.

BABA, YUMC, OTIS, WYNN, NVDA, TSM, SNPS

Median discount
36.3%
Average LSEG
1.86
Structural gap
3 of 7

Tariffs and trade

6 companies

Tariffs or trade rules change its costs or its sales.

CPB, PEP, AZO, TSM, PG, SUGR

Median discount
17.8%
Average LSEG
2.42
Structural gap
2 of 6

AI spender

5 companies

Spends heavily on AI data centers; the market asks when that spending pays back.

BABA, GOOGL, META, AMZN, MSFT

Median discount
15.5%
Average LSEG
1.80
Structural gap
1 of 5

Weight-loss drugs

5 companies

GLP-1 weight-loss drugs change what people eat, buy or need from health care.

CPB, HSY, PEP, MCD, NVO

Median discount
25.5%
Average LSEG
2.76
Structural gap
0 of 5

Power and energy

1 company

Earns from electricity, gas or oil demand, including power for data centers.

PCG

Median discount
36.0%
Average LSEG
2.39
Structural gap
1 of 1

Gap types

Not every discount can close. Tap a company to read why its gap is of its type.

Can close

28

64% of rated picks

A dated event that someone can check can close the gap within about two years.

  • AVGO
    Why AVGO: 44.4%

    Each quarterly report checks the FY2027 AI chip revenue target, starting with the December 2026 report.

  • APP
    Why APP: 41.1%

    The early November report and the next quarters show whether the AI video tool ships and e-commerce growth holds, even while the lawsuit runs.

  • HONA
    Why HONA: 38.3%

    Quarterly reports show when the parts shortage eases and the aftermarket backlog turns into revenue.

  • OTIS
    Why OTIS: 37.2%

    Service margins show in quarterly reports, and Judy Marks retires by July 31, 2027 at the latest, so a new CEO must start by then.

  • POOL
    Why POOL: 30.5%

    The October 22 report on the peak pool season and the next quarters show whether maintenance demand holds while construction stays weak.

  • INTU
    Why INTU: 29.6%

    The January to April 2027 tax season shows whether TurboTax keeps its price-sensitive filers.

  • SAP
    Why SAP: 28.9%

    Quarterly cloud growth and the 2027 decisions of on-premises customers give a dated test of the seat-loss fear.

  • RBGLY
    Why RBGLY: 26.3%

    A sale of Mead Johnson or a clear end to the NEC lawsuits can remove the main distraction within about a year.

  • NVDA
    Why NVDA: 26.0%

    The November 2026 report and the next two quarters show whether demand comes from many customers and whether gross margin stops falling.

  • MRVL
    Why MRVL: 23.5%

    The Q3 fiscal 2027 report near December 1, 2026 and the ramp of new custom chip programs over the next quarters test the targets Marvell raised at its October 6 Investor Day.

  • SPGI
    Why SPGI: 22.3%

    Year-end issuance data and the next quarters show whether AI-related bond issuance and private credit ratings continue.

  • SCHW
    Why SCHW: 22.0%

    Net interest margin and net new assets show each quarter whether the cash-sorting and AI-loyalty fears are real.

  • CPRT
    Why CPRT: 21.9%

    The loss of the large insurance customer laps by mid-2027, and volume after that date shows whether the moat holds.

  • AZO
    Why AZO: 20.5%

    The next three or four reports show whether domestic same-store sales and margins recover without one-time items.

  • MCD
    Why MCD: 20.3%

    US guest counts each quarter show whether the value menu and the franchisee plan bring customers back in 2027.

  • UNH
    Why UNH: 20.1%

    The medical care ratio shows each quarter, and a DOJ settlement with a limited penalty can remove the main overhang.

  • GOOGL
    Why GOOGL: 18.8%

    The US ad tech ruling removed the US breakup risk, so quarterly Cloud and Search growth through early 2027 can close the gap.

  • META
    Why META: 15.5%

    The Q3 2026 report near October 28 and the next quarters show whether Meta One and Muse earn revenue and free cash flow recovers.

  • NOW
    Why NOW: 14.6%

    Q3 2026 earnings on October 28 and the next two quarters show whether AI products add revenue or replace paid seats.

  • ICE
    Why ICE: 14.2%

    The October 29, 2026 report, shareholder vote and antitrust deadline, then the MarketAxess closing in the first half of 2027, give dated checks.

  • AMZN
    Why AMZN: 12.5%

    Quarterly AWS margins and backlog conversion can show the AI spending pays back before the 2027 antitrust trials.

  • MSFT
    Why MSFT: 10.8%

    Azure growth against rising capital spending is visible each quarter, starting with the October 28 report.

  • NVO
    Why NVO: 8.0%

    The FDA decision on CagriSema in Q4 2026 and its launch in 2027 give a dated test of whether Novo can take share back from Lilly.

  • YUM
    Why YUM: 6.6%

    Taco Bell US sales after the outbreak show in the next reports, and Yum must refinance its 2027 debt.

  • WM
    Why WM: 5.1%

    Healthcare Solutions synergies and the CEO handover in January 2027 show in the next few quarterly reports.

  • PG
    Why PG: 2.4%

    The restructuring ends in mid-2027, and each quarter shows whether unit volume grows again.

  • SNPS
    Why SNPS: 1.9%

    Falling debt and Ansys savings show in quarterly reports, but a new China export rule can stop the process.

  • MRX
    Why MRX: -2.0% (AI estimate)

    The Q3 2026 results and the next quarters of normal volatility show within about a year which profit level is the true base.

Slow

9

20% of rated picks

The gap can close, but only over several years of execution.

  • CPB
    Why CPB: 53.7%

    Management does not expect margins to recover before fiscal 2028, and legacy soup and snack demand may keep falling.

  • CLX
    Why CLX: 45.6%

    Clorox needs a new CEO, an earnings recovery and debt repayment through the end of 2027, so the gap closes only step by step.

  • RELX
    Why RELX: 36.6%

    Any new AI product from a rival can reopen the fear, so Legal and STM must outgrow it for several years.

  • WYNN
    Why WYNN: 36.3%

    Macau must recover and the UAE resort must open in September 2027, and Macau country risk will probably keep part of the discount.

  • BR
    Why BR: 33.8%

    Recurring revenue and bookings must hold for several years to show that AI and tokenization do not route work around Broadridge.

  • HSY
    Why HSY: 29.2%

    Management itself calls 2026 and 2027 years of profit repair, so volume and margins return to balance only from 2028.

  • PEP
    Why PEP: 25.5%

    The margin target runs over three years, and North America volume must recover while the US consumer stays weak.

  • EXPGY
    Why EXPGY: 15.5%

    Only years of steady organic growth can show that lenders still need bureau data in the age of AI scoring.

  • CSU
    Why CSU: 14.2%

    The new President must prove the acquisition model over a full cycle without the founder, and that takes several years.

Will probably not close

7

16% of rated picks

A lasting reason holds the price below fair value, so the discount overstates the opportunity.

  • BABA
    Why BABA: 46.2%

    The VIE structure, the US military-company listing and US-China tension keep a discount on Alibaba whatever its cloud business does.

  • BN
    Why BN: 44.6% (Management)

    The multi-layer holding structure stays after the merger, so part of the gap to plan value will probably never close.

  • YUMC
    Why YUMC: 44.3%

    US-China tension and a weak Chinese consumer keep a country discount on Yum China that its own results cannot remove.

  • PCG
    Why PCG: 36.0%

    California has no cap on wildfire liability, the 2026 reform bill failed, and one bad fire can erase years of profit.

  • TSM
    Why TSM: 15.1%

    Strong results can close the execution part, but the market will probably keep a discount for Taiwan-China risk that TSMC cannot control.

  • BTI
    Why BTI: 11.9%

    Many investors do not own tobacco, and regulators keep adding rules, so the market will probably always pay less for BAT's cash flow.

  • SUGR
    Why SUGR: 0.5% (AI estimate)

    The founder controls the company, the stock trades very little, and minority holders cannot block a bid, so the price will probably stay below reported value.

Fair value vs analyst ratings

Both see value

8

Discount above the median, and analyst ratings are positive.

  • BABA46.2% · LSEG 1.83
  • AVGO44.4% · LSEG 1.77
  • YUMC44.3% · LSEG 1.77
  • APP41.1% · LSEG 1.94
  • RELX36.6% · LSEG 1.67
  • WYNN36.3% · LSEG 1.74
  • NVDA26.0% · LSEG 1.71
  • MRVL23.5% · LSEG 1.87

Only the fair value sees value

13

Possible value trap: discount above the median, but analyst ratings are not positive.

  • CPB53.7% · LSEG 3.38
  • CLX45.6% · LSEG 3.10
  • BN44.6% (Management) · LSEG 2.17
  • HONA38.3% · LSEG 2.56
  • OTIS37.2% · LSEG 2.44
  • PCG36.0% · LSEG 2.39
  • BR33.8% · LSEG 2.30
  • POOL30.5% · LSEG 2.56
  • INTU29.6% · LSEG 2.36
  • HSY29.2% · LSEG 2.67
  • SAP28.9% · LSEG 2.04
  • RBGLY26.3% · LSEG 2.33
  • PEP25.5% · LSEG 2.65

Only analyst ratings are positive

17

Analyst ratings are positive, but the discount is at or below the median.

  • SPGI22.3% · LSEG 1.86
  • SCHW22.0% · LSEG 1.96
  • AZO20.5% · LSEG 1.89
  • UNH20.1% · LSEG 1.97
  • GOOGL18.8% · LSEG 1.80
  • EXPGY15.5% · LSEG 1.67
  • META15.5% · LSEG 1.83
  • TSM15.1% · LSEG 1.65
  • NOW14.6% · LSEG 1.82
  • ICE14.2% · LSEG 1.81
  • CSU14.2% · LSEG 2.00
  • AMZN12.5% · LSEG 1.80
  • BTI11.9% · LSEG 1.50
  • MSFT10.8% · LSEG 1.72
  • WM5.1% · LSEG 2.00
  • SNPS1.9% · LSEG 1.85
  • MRX-2.0% (AI estimate) · LSEG 2.00

Neither

5

Discount at or below the median, and analyst ratings are not positive.

  • CPRT21.9% · LSEG 2.21
  • MCD20.3% · LSEG 2.24
  • NVO8.0% · LSEG 2.85
  • YUM6.6% · LSEG 2.41
  • PG2.4% · LSEG 2.54

How the groups are drawn: the fair value sees value when a pick's discount to fair value is above the median of the picks with both (22.3% today). Fair values are the analysts' unless a source is named next to the discount. The LSEG analysts are positive at a rating of 2.0 or lower (1 is strong buy, 5 is strong sell). An LSEG rating of 2.5 or higher, in red, means the LSEG analysts lean to hold or sell. A large discount that analysts do not share can be a value trap: the market may know something the fair value does not.

Earnings by week

The next 6 weeks of reports, with the themes of the companies that report.

  1. Week of Oct 12: 2 reports — AI supplier ×1, AI disruption risk ×1, China and geopolitics ×1, Interest rates ×1, Tariffs and trade ×1

  2. Week of Oct 19: 5 reports — AI disruption risk ×2, Regulation and courts ×2, Weak consumer ×1, Interest rates ×1, High debt ×1, New management ×1, Tariffs and trade ×1, Power and energy ×1

  3. Week of Oct 26: 12 reports — Regulation and courts ×5, AI spender ×4, High debt ×3, New management ×3, AI disruption risk ×2, Interest rates ×2, AI supplier ×1, Weak consumer ×1, China and geopolitics ×1, Weight-loss drugs ×1

    2 of 12 dates are estimated (last report plus three months), not confirmed by the company.

  4. Week of Nov 2: 10 reports — Weak consumer ×5, Interest rates ×4, Regulation and courts ×4, New management ×4, AI disruption risk ×3, High debt ×3, China and geopolitics ×2, Weight-loss drugs ×2

    7 of 10 dates are estimated (last report plus three months), not confirmed by the company.

  5. Week of Nov 9: 1 report — AI supplier ×1, Interest rates ×1, High debt ×1

  6. Week of Nov 16: 5 reports — Regulation and courts ×3, AI disruption risk ×2, AI supplier ×1, Weak consumer ×1, China and geopolitics ×1, High debt ×1, New management ×1, Tariffs and trade ×1

    4 of 5 dates are estimated (last report plus three months), not confirmed by the company.

  7. Week of Nov 23: 1 report — AI spender ×1, China and geopolitics ×1, Regulation and courts ×1

    1 of 1 date is estimated (last report plus three months), not confirmed by the company.

Back to top