QCOM

We still rate Qualcomm Avoid: at $178.40 the price is 22.7% above our fair value of $138, and the central case returns 4.5% a year, under the 10.83% cost of equity. This report corrects our 2026-10-15 report, which went out in the 17:13 publish at $141 and stays as published. The operating forecast, capital structure, cost of capital and view are unchanged. Two errors in the long fade are fixed, and the value falls $3.

How to read the cases. Each report values a company three ways. The central case is our best estimate. The bear case is a severe but plausible bad outcome. The bull case is the matching good outcome. The central case sets the view; the bear and bull show how far the value could move. Qualcomm's fiscal year ends in late September, so "fiscal 2031" ends on 2031-09-28. QCT is Qualcomm's chip business and QTL its patent-licensing business.

1. Two errors in the long fade lowered the value by $3

The fade is how the model values Qualcomm after the five forecast years: it lets the return earned on new capital, and growth, drift to a mature level over 15 years. Two of its inputs were wrong.

Error 1: the fade started at the blended return. The published model started the fade's return on new capital at the 57% the forecast years earn in total (after-tax operating profit up $4,684M from fiscal 2027 to fiscal 2031 on $8,254M of capital added). About 38% of that rise is margin gained on revenue Qualcomm already had (QCT excluding data center 27% → 29%, data center 2% → 21% on its $4B base, unallocated costs 9.5% → 8% of revenue, Arm's royalty cost starting). The fade holds margins flat, so it must start at the volume-only return: new revenue at fiscal 2031 segment margins, after unallocated costs and tax, over the capital added. The model now computes it per case: central 35% ($2,919M over $8,254M), bull 64%, bear negative (so the bear still starts and stays at its 15% mature return). An outside reviewer measured 34%; the point between is Arm's $100M a year from fiscal 2028, which we treat as a cost on the existing business, not volume ($86M after tax). The rule is in our valuation rules (valuation.md as of d3f682a17).

Error 2: the mature return counted securities as operating capital. The 22% came from fiscal 2016, when $13.7B of long-term securities sat in invested capital; without them fiscal 2016 earned 49%. On one basis for every year (equity plus debt less cash, short-term and long-term investments; operating income at 15% tax; market data statements), fiscal 2016–25 returns were 49%, 16%, 10%, 76%, 50%, 60%, 49%, 25%, 30%, 42%. The lowest outside the fiscal 2017–18 licensing-dispute years is fiscal 2023, the last handset trough, at 25.1%. The model now computes this from the statements and uses it for the central and the bull (whose blind 25% it matches); the bear keeps 15%, our judgment of a worse business.

Measured by switching each step in a scratch copy of model.py. Neither step rests on the price; together they move the value $3 further from it.

2. Our forecast gives $138; the outside-view checks sit higher

The operating cases are the published ones (business, results, forecast lines and their evidence: sections 1–3 of the 2026-10-15 report; assumptions.md is the blind record). Model model.py, output output.txt; after-price arithmetic unblinded.py (output).

Checks. The outside view (peer ten-year median GAAP P/E 29.4x on fiscal 2031 EPS) is $330, 89% above the fade. Most of the gap is rates: the peers' decade had a ten-year yield near 2.5% against 5.1% now, and moving the earnings yield by the difference gives 16.7x and $187, 7% above the fade. The rest is the peer set: AVGO, MCHP and ADI earn their multiples on amortisation-cut GAAP earnings and analog franchises. We believe the fade. The value-driver formula gives 33.9x because growth at exit (8.3%) sits near the cost of capital (WACC, 10.30%); the fade carries that growth down to 3%, and its 14.8x implies 4.8% for ever at the mature return. Qualcomm's own ten-year median P/E (18.7x, read after the price) on our fiscal 2031 EPS gives $159, still 11% below the price.

Cost of equity 10.83%: 5.10% ten-year Treasury (2026-10-15) + peer median Blume beta 1.36 × 4.20% Damodaran implied premium (2026-10-01). Own Blume beta 1.38: $137, purchase $131. Damodaran's semiconductor beta 1.52 (January 2026): $125, purchase $123.

3. The price needs 26% more profit than we forecast

At the 10.83% cost of equity: central NOPAT ×1.26 in every year (fiscal 2031 NOPAT $15.9B against our $12.6B; fiscal 2031 EPS $14.17 against $11.21), or an exit of $243, 20.2x fiscal 2032 NOPAT against the fade's 14.8x. The Street's fiscal 2027 revenue matches ours within 0.5%; by fiscal 2029 ours is 13% lower on revenue and about a third lower on EPS, the gap between our data center and IoT paths and management's Investor Day targets (unblinded-output.txt; the vendor's fiscal 2030 row is not a real estimate and is left out).

4. Three results settle whether we are wrong

1. Fiscal 2027 data center revenue against $5B, and the first accelerator silicon in customers' hands (mid-2027 on the company's plan).
2. Fiscal 2027 non-handset growth against the company's "greater than 60%" (our central +50%, $20.0B from $13.3B).
3. Android after the memory shock: whether fiscal 2027 handset revenue excluding Apple grows (our +7%) or falls again.

Bear ($48): Android flat then shrinking, data center stuck at $2–5B, QTL renewals 15% lower in fiscal 2028, QCT margins held at 24%, Arm costing $400M a year, tax 16.5%. Why the bulls could be right ($230): Qualcomm has beaten its own guide in 7 of 8 quarters, beat its 2021 automotive target by 60%, and has Amazon tied to up to $60B of purchases over ten years; if data center reaches the $15B fiscal 2029 target it is a second franchise.

5. Catalysts and what would show us wrong

• 2026-11-04, after the close: Q4 fiscal 2026 results (Robinhood earnings calendar, marked verified; not confirmed on the company site). Our forecasts are unchanged: revenue $10,350M (range $10,000–10,700M), non-GAAP EPS $2.22 ($2.10–2.35), QCT handsets $5,304M, automotive $1,685M, QTL $1,319M.
• Q1 fiscal 2027 results (early February 2027, expected, not confirmed): first Amazon custom-silicon revenue and the Apple step-down.
• 2027-04-01: the renewed Apple licence takes effect (terms not disclosed).
• What would show us wrong: fiscal 2027 data center at or above $5B with accelerator silicon at customers (move toward the bull); fiscal 2027 non-handset revenue above $21.3B; fiscal 2027 Android handset revenue down again (weight the bear); a disclosed QTL renewal at a lower rate.

6. What the reviewers questioned, and how it came out

A checklist review of the work raised or cleared these points.

• The move from the blind $129 to $138 (+$9), a medium concern. The largest change since the blind step is the fade starting at the forecast's own return (now volume-only, +$1.6 against the blind 30% start; the published report's blended start was +$5.9). Every change rests on a method rule or a filing line, none on a management claim or a Street figure; the review answers are in section 9 of the published report.
• Damodaran's semiconductor beta, a medium concern. At 1.52 it is harsher than the peer median and not used; at 1.52 the value is $125 and the view is unchanged.
• The outside view is 89% above the fade, a low concern, kept as a check. Section 2 splits the gap between rates and the peer set.
• The central fiscal 2029 is 13% below the Street on revenue and below management's fiscal 2029 targets, a medium concern, with the dated reasons in the published report (2021 IoT miss; no accelerator or CPU silicon at customers, CEO 2026-07-29; handsets down 20% on memory, CFO 2026-07-29).
• Checked and clear: the Amazon warrant and the 17.83M Modular shares are in the share count (8-K 2026-07-31 saved here); the bridge lines from enterprise value to equity are from the 10-Q for Q3 fiscal 2026 saved here.
• Not raised: the view is the blind view; the value is 23% below the price; the purchase price is $132 at peer beta and $131 at own beta; the value is 9% from the blind value.

7. Sources

As in the published report (section 12 there), plus: market data balance sheets for fiscal 2016–25 for the mature return (data/balance-sheet-statement-annual.json), an outside reviewer's rerun of 2026-10-15, and our valuation rules (valuation.md at d3f682a17).