QUALCOMM QCOM

Reference table

1. What the business is, and the concentration the vendor cannot show

Chief executive Cristiano Renno Amon; 52,000 employees; NASDAQ-listed; reporting in USD; fiscal year ends the last Sunday of September.

Two reportable segments, and seg_prod nearly ties. Fiscal-2025: QCT $38,367M + QTL $5,582M = $43,949M against reported revenue of $44,284M — a $335M (0.8%) residual for QSI and other.

Note QTL: $6,358M, $5,306M, $5,572M, $5,582M across four years while QCT swung from $37,677M to $30,382M and back to $38,367M. The licensing line is an annuity and the silicon line is a cycle. That is exactly what the knowledge base's independent claim describes — "Licensing (QTL) is ~15% of revenue but ~30% of operating income at 85-90% gross margins, fueling the rest of the business" — and the revenue share checks out at 12.6%, close to the claimed 15%.

Customer concentration, from the 10-Q filed 2026-07-29 — there is no vendor field for any of this:

> "A significant portion of our revenues are concentrated with a small number of customers/licensees of our QCT and QTL segments."

Two counterparties are 43% of total revenue in the June quarter. The largest rose five percentage points in twelve months. A third dropped out of the 10% disclosure entirely, having been 13% a year ago. Both directions are risks: increasing dependence on one buyer and demonstrated attrition of another. The knowledge base's own bearish claim from 2023 — "Biggest risk: customers rolling their own chips" — is now observable in the filing's own table.

seg_geo STOPS AT FISCAL 2023 and is three years stale. The most recent entry available is fiscal 2023:

Qualcomm books nearly two-thirds of its revenue in China and 3.5% in its home country, and the most recent data available to us is from three years ago. This is the single largest unquantified risk in this file. We report the fiscal-2023 figures with the staleness stated and draw no conclusion about the current mix, because we cannot.

2. The quarterly deterioration

Three sequential declines from the December peak: $12,252M → $10,599M → $9,947M, a 18.8% fall in six months. Year-on-year growth turned negative in March and stayed negative in June.

Operating margin has collapsed from 27.5% to 16.3% in two quarters. Gross margin fell 250 basis points over the same period. Revenue down 4% produced operating income down 41% — that is negative operating leverage, and it is what a semiconductor business looks like when volume falls against a fixed research base of $9,042M a year (20.4% of revenue).

The GAAP net income column is unusable and the 10-Q explains exactly why:

> "In the fourth quarter of fiscal 2025, tax reform legislation included in the One Big Beautiful Bill Act (OBBB) was enacted in the United States... [we recorded a] $5.7 billion valuation allowance on our federal deferred tax assets in fiscal 2025."

> "In the second quarter of fiscal 2026, the U.S. Department of Treasury and the Internal Revenue Service issued Notice 2026-07, which... allows us to reduce CAMT by certain previous [amounts, producing a] $5.7 billion income tax benefit in the second quarter of fiscal 2026... a $5.7 billion benefit in the second quarter of fiscal 2026 from releasing our valuation allowance on our federal deferred tax assets."

That single item explains both anomalies: the minus $3,117M fourth quarter of fiscal 2025 and the $7,370M second quarter of fiscal 2026 on operating income of $2,309M. It also explains the fiscal-2025 income statement, where tax expense of $7,122M on pre-tax income of $12,663M is a 56.2% effective rate and cut net income to $5,541M from $10,142M — against a fiscal-2024 rate of 2.2%. The nine-month fiscal-2026 income tax line is a BENEFIT of $4,136M. Deferred tax assets on the balance sheet went from $743M to $5,679M.

A second, smaller earnings-quality item in the June quarter. Investment and other income was $1,014M, of which $726M was "net gains on marketable securities... primarily... net unrealized gains related to certain marketable equity securities." The quarter's $2,002M of net income therefore includes $726M — 36% — of unrealised marks. On a clean basis the quarter earned closer to $1,430M, and the nine-month figure carries $605M of the same. This is the GOOGL defect class in miniature and it is why we anchor entirely on the non-GAAP series.

Restructuring is underway: $68M in the quarter and $97M in nine months, "substantially all of which related to severance costs."

3. Balance sheet

Correction 1 — netDebt omits $4,635M of short-term investments and is nine months stale. The vendor's $8,530M is total debt less cash and equivalents only. Corrected at the current date: total debt $15,270M less cash $4,533M less marketable securities $3,771M = $6,966M. The vendor reports $8.530B; the filing supports $6.966B; we use $6.966B — an 18.3% overstatement corrected. We do NOT net the $2,323M of restricted cash, which is new in the June quarter and appears associated with the acquisition activity that took goodwill from $11,358M to $14,274M.

Correction 2 — the vendor's own cash figure does not match the filing. The vendor's fiscal-2025 balance sheet shows cashAndCashEquivalents of $7,843M; the 10-Q's comparative column for the same date shows $5,520M. A $2,323M difference — exactly the restricted-cash line the filing now discloses separately. The vendor appears to include restricted cash in cash and equivalents. Recorded; it does not change the corrected net-debt figure because we exclude restricted cash on both sides.

Correction 3 — a $2,916M increase in goodwill in nine months is unexplained by any vendor field. Goodwill rose from $11,358M to $14,274M and intangibles from $1,148M to $1,510M, with inventories up $1,853M. Qualcomm has made a material acquisition and the payload shows no trace of it.

The rebuilt enterprise value:

> Market cap $170.804B (1,050M shares × $162.67)

> + total debt $15.270B − cash $4.533B − marketable securities $3.771B

> = corrected enterprise value $177.770B

The vendor reports $181.541B, overstated by $3.771B (2.1%) — precisely the omitted short-term investments. Note also that the vendor's EV implies net debt of $10.737B against its own netDebt field of $8.530B: the file is internally inconsistent by $2.2 billion.

Interest coverage is 14.9x and return on invested capital is 17.4%. This is a lightly levered balance sheet and the risk in this name is not financial.

4. Cash flow — the strongest thing in this file

A 28.9% free-cash-flow margin on capital expenditure of 2.7% of revenue, and 98.3% of that free cash flow returned to shareholders. The 7.37% total shareholder yield is the highest in this batch by a wide margin — the next is Schwab's 6.4% and most are under 3%.

The buyback more than doubled in fiscal 2025 to $8.791B — 5.1% of the current market capitalisation — and it is working: diluted shares fell from 1,130M (FY2024) to 1,069M in the June 2026 quarter, a 5.4% reduction in seven quarters. Shares outstanding at 2026-07-27 were 1,050 million, per the 10-Q cover.

Dividend. The vendor reports lastDividend: 3.59 and dividendPerShareTTM: 3.59, yielding 2.21%. That $3.59 is the trailing-twelve-month total — the same mislabelling as everywhere in this batch. Dividends paid in fiscal 2025 were $3,805M on roughly 1,090M average shares, or $3.49 per share. Payout is 41.3% of trailing GAAP earnings and roughly 31% of trailing non-GAAP.

The knowledge base predicted this capital-allocation pattern in 2023 and was conservative: "With ~$6-10B annual free cash, Qualcomm will keep returning capital via buybacks and dividends plus selective bolt-ons (Nuvia), avoiding big transformative M&A." Actual fiscal-2025 free cash flow was $12.820B — above the top of that range — and the pattern held. (Note that the $2.9B of goodwill added in fiscal 2026 suggests a larger bolt-on than the claim anticipated.)

5. Valuation — priced in or room?

At $162.67 (market cap $170.804B, corrected enterprise value $177.770B):

Note the shape: the forward multiple RISES from 14.3x trailing to 15.7x on fiscal 2027, because earnings fall in both intervening years. That is the opposite of the rolldown every other name in this batch shows, and it is the single most honest statement of the situation. You are not buying a cheap growing business; you are buying a cheap shrinking one that consensus expects to resume growing in fiscal 2028.

est.ebitAvg is BELOW est.netIncomeAvg in fiscal 2029 and 2030 — arithmetically impossible — and both EBIT and EBITDA rows are REJECTED. FY2030 shows ebitAvg of $21.153B against netIncomeAvg of $27.548B; FY2029 shows $18.315B against $19.391B. A company cannot earn more net income than earnings before interest and tax while paying both. The ratios are additionally fixed: ebitdaAvg / revenueAvg = 34.04% and ebitAvg / revenueAvg = 29.69% in all eight years, to four significant figures. Per the data contract, all forward valuation runs on epsAvg.

FY2029 and FY2030 rest on 4 and 5 analysts and are excluded. The FY2028 EPS line carries 9 analysts with an $11.17 to $15.45 range and is used only for the bull case, with coverage disclosed.

Peer context. The vendor peer set is mixed: Uber and Intuit are not semiconductor comparables, and Sony and Arm are only loosely so. The defensible names are Texas Instruments ($259.1B at $283.67), Intel ($509.1B), Applied Materials, Lam Research and KLA — the last three being wafer-fab equipment rather than fabless design. MediaTek, which the knowledge base names as Qualcomm's direct competitor with an "unspoken truce dividing the market by price tier," is not in the payload. No peer multiples are supplied, so no peer-multiple comparison is drawn. Note that ADI, also covered in this batch, is a semiconductor name but not a comparable — analog content-per-unit economics have nothing in common with handset silicon plus licensing.

5a. What today's price assumes (the inversion)

At $162.67 — 15.3x FY2026 consensus and 15.7x FY2027 — the price embeds:

• Two down years and then a recovery. Non-GAAP EPS of $10.627 (FY2026, −10.8%) and $10.346 (FY2027, −2.6%) before $13.122 in FY2028 (+26.8%). (Consensus; 18, 21 and 9 analysts.) The recovery year rests on nine analysts and spans $11.17 to $15.45.
• Revenue reaches $51.709B in fiscal 2028 — a 16.8% increase on fiscal 2025's actual $44.284B, off a base currently shrinking 4% a year. (Consensus; 24 analysts.) This is the most fragile assumption in the price, and it can only be automotive, internet-of-things and edge diversification working. There is no segment disclosure in the payload that lets a reader verify progress toward it.
• The two customers at 43% of revenue stay. (Filing figure; our assumption on stability.) One rose from 18% to 23% in a year; a third fell from 13% to below 10%. The nearest check is the concentration table in the fiscal-2026 10-K.
• The buyback continues at roughly $8.8 billion a year. (Our annualisation of fiscal 2025.) That is 5.1% of market capitalisation and it is doing roughly half the work of holding per-share earnings up while revenue falls.
• The market keeps paying 15-16x on the forward year. (Our number.) At 12x FY2027E the stock is $124 — the 52-week low; at 20x it is $207, essentially the street's target. The multiple has spent the last year traversing that entire range.

5b. The return bridge (why the multiple moves)

Qualcomm's return bridge is unusual and worth stating carefully, because earnings go DOWN before they go up. Over roughly two years: non-GAAP EPS change (−13.1%, from FY2025's actual $11.908 to FY2027E $10.346) + multiple drift + shareholder yield (+7.37% a year, ~15% compounded).

Almost all of the expected return over the next two years is the shareholder yield, not the earnings. At a 7.37% yield — $8.8 billion of buyback plus $3.8 billion of dividend against a $170.8 billion capitalisation — a holder is paid roughly 15% over two years while earnings decline 13%. That is the whole reason this can work at all, and it is entirely dependent on the buyback continuing.

Our base assumes modest multiple EXPANSION, which is unusual for us and needs justification. We use 17.4x fiscal-2027 consensus against 15.3x on fiscal 2026 today. The reason is not optimism about the business: it is that 15.3x on a trough year, for a company with a 7.4% shareholder yield, a 0.54x levered balance sheet and a contractual licensing annuity, is a distressed multiple that historically resolves upward once the earnings trough is visible. If instead the multiple simply HELD at 15.3x on FY2028E the price would be $201 (+23%); if it compressed to 12x on FY2027E the price would be $124 — exactly the 52-week low.

The bull case at $260 is 16.8x the FY2028 consensus HIGH of $15.45 — requiring both the diversification to work and no de-rating. The bear case at $122 is 13x the FY2027 consensus LOW of $9.42, and sits 1.6% above the street's own $120 low target and 1.7% below the 52-week low of $124.07.

5c. Variant perception (where we differ, what would surprise)

• We differ from the street's mean while agreeing with its dispersion. Consensus is $204.05 but the individual targets run $120 to $400 — a 3.3x spread and the widest in this batch — with 34 holds against 30 buys and 5 sells, producing the only "Hold" consensus rating in these twelve names. When the sell side cannot agree within a factor of three, the mean is a statistic rather than a view. Our $180 sits below it deliberately.
• We reject the entire trailing GAAP earnings series and the multiple built on it. The $5.7 billion valuation allowance taken in Q4 fiscal 2025 and released in Q2 fiscal 2026 produces a minus $3,117M quarter and a $7,370M quarter that between them make any GAAP trailing figure meaningless. The June quarter additionally contains $726M — 36% of net income — of unrealised marks on equity securities. All valuation here runs on non-GAAP.
• We think the customer-concentration disclosure is the most under-priced fact in this file. Two counterparties at 43% of revenue, with the larger rising five points in a year and a third falling out of the 10% table from 13%. This appears in no vendor field, no ratio and no screen. Watchable number: the concentration table in the fiscal-2026 10-K. Customer (x) above 26%, or customer (y) below 17%, would each be material.
• We think the balance sheet is better than it screens and it barely matters. Corrected net debt of $6.966B at 0.54x EBITDA against a vendor figure of $8.530B at 0.797x. Qualcomm's risk is commercial, not financial, and the correction changes nothing about the verdict.
• We can adjudicate three 2023 knowledge-base claims and two have aged remarkably well. "Sales are cyclical around new-standard launches; with 5G largely built out, an elongated pause until 6G (late 2020s) slows growth" — correct, and it is precisely what is happening. "Biggest risk: customers rolling their own chips" — visible in the concentration table, where a 13% customer has fallen below 10%. "With ~$6-10B annual free cash flow, Qualcomm will keep returning capital via buybacks and dividends" — correct on the pattern, conservative on the amount ($12.820B).
• Positive surprise that would force a re-rate: the 2026-11-04 print beating $10.232B with a December-quarter guide showing sequential growth, which would mark the revenue trough; automotive or internet-of-things revenue disclosed separately and growing; the concentration table showing diversification rather than concentration; or a buyback acceleration above $9 billion.
• Negative surprise that would break the thesis: fourth-quarter revenue below $9.5 billion or a December guide below $10.5 billion; customer (x) or (y) reducing; gross margin below 52%; foundry access constrained as one independent claim argues; or the buyback slowing. At 15.3x with revenue falling, the last of those would be the most damaging, because the yield is currently doing most of the work.

Synthos fair values

Anchors are multiples of the fiscal-2027 and fiscal-2028 consensus non-GAAP EPS distributions (FY2027 mean $10.346, low $9.42, high $12.44 on 21 analysts; FY2028 mean $13.122, low $11.17, high $15.45 on 9 analysts — disclosed).

• Bear ~$122 — 13x the FY2027 consensus LOW of $9.42. Cross-check: 12.7x FY2026E; 1.6% above the street's own $120 low target and 1.7% below the 52-week low of $124.07; corrected EV of $135.1B on FY2027E revenue is 3.01x sales. The scenario: the revenue decline extends through fiscal 2027, one of the two large customers reduces further, gross margin stays below 53%, and the diversification story is deferred again. −25.0%.
• Base ~$180 — 17.4x the FY2027 consensus MEAN of $10.346. Cross-check: 16.9x FY2026E; 13.7x FY2028E; 15.8x trailing non-GAAP; corrected EV of $195.9B on FY2027E revenue of $44.912B is 4.36x sales. Sensitivity, stated openly: 15x FY2027E gives $155, 4.7% BELOW spot, and 20x gives $207, essentially the street's target. The whole answer lives in a 15-20x band. The scenario: fiscal 2026 and 2027 land near consensus, the revenue trough forms, the buyback continues at $8.8 billion, and the multiple recovers modestly from a trough level. +10.6%, plus about 15% of compounded shareholder yield over the period — which is where most of the return actually comes from.
• Bull ~$260 — 16.8x the FY2028 consensus HIGH of $15.45. Cross-check: 25.1x FY2027E; well below the street's $400 high, which we treat as an outlier. The scenario: automotive, internet-of-things and edge compute deliver the fiscal-2028 revenue step to $51.709B, the licensing annuity holds, concentration falls as diversification succeeds, and Qualcomm re-rates from a handset cyclical to a diversified edge-compute franchise. +59.8%.

Base is 10.6% above spot; asymmetry roughly 2.4:1 to the upside (25.0% down, 59.8% up). On asymmetry alone this is the most attractive setup in this batch. What holds it at Watch is that the base sits 11.8% below the street, revenue is currently falling, the fiscal-2028 recovery that carries the bull case rests on nine analysts, the entry includes a 7.32% one-day gap with the 50-day average 20.8% overhead, and the single largest risk — customer concentration — is worsening in the filing's own table.

6. Knowledge base — deep, two-sided, and heavily concentrated in two sources

Raw hits: 28 (21 entity, 7 text). Used: 14. Discarded: 14. Searches covered Qualcomm and Snapdragon.

Two attribution problems, both reported.

First: EIGHT claims dated 2022-04-27 from a single channel are QUALCOMM MANAGEMENT VOICE. Their content is operator framing, not analysis: "Snapdragon strategy is to be synonymous with premium Android — enabling, not competing with, Google"; "Qualcomm is expanding beyond phones into next-gen PCs (Snapdragon laptops, Lenovo ThinkPad 5G with ~28hr battery)"; "The largest addressable market is making 'everything' smart... a market that grows exponentially"; "Apple is investing to build its own modem, but Qualcomm's mobile strategy is Snapdragon-driven and not defined by the Apple modem relationship." These are first-person strategy statements from a long-form interview. All eight are excluded from the conviction pool, quoted at half weight, and are four years stale.

Second: NINE claims come from one channel on one date — 2023-05-27 — i.e. a single episode. That episode is by some distance the most useful material in the lane and it is genuinely balanced, but nine claims from it is one source, not nine, and the raw count overstates breadth by roughly a factor of three.

After both adjustments the lane is roughly four independent inputs, two of which are named 2026 speakers.

The 2023 episode's claims, and three are now gradeable:

> bullish · conviction 70 · "Qualcomm's IP is embedded in the global 3G/4G/5G standards, so nearly every phone pays it a royalty — a durable licensing moat."

> bullish · conviction 75 · "Licensing (QTL) is ~15% of revenue but ~30% of operating income at 85-90% gross margins, fueling the rest of the business." CHECKED: QTL was 12.6% of fiscal-2025 revenue, close to the claim, and its revenue has been $5.3-5.6B for four straight years while QCT swung $7.3B. The annuity characterisation is confirmed.

> bearish · conviction 65 · "Biggest risk: customers rolling their own chips — Apple wants to build its own modem to displace Qualcomm entirely." ADJUDICATED: the 10-Q's concentration table shows a customer that was 13% of revenue a year ago has fallen BELOW 10%. Directionally confirmed.

> neutral · conviction 60 · "Qualcomm sales are cyclical around new-standard launches; with 5G largely built out, an elongated pause until 6G (late 2020s) slows growth." ADJUDICATED and the single best call in this lane: revenue is down 4.0% year on year and consensus expects two down years. This claim, made in May 2023, describes 2026 exactly.

> neutral · conviction 55 · "Qualcomm and Mediatek have an unspoken truce dividing the market by price tier, but Mediatek has closed the engineering gap and could contest higher-end phones."

> bearish · conviction 55 · "Apple's application processor is ~two years ahead of Snapdragon, and rival OEMs competing with Apple pressures Qualcomm's SoC business."

> bullish · conviction 60 · "Building a baseband modem is very hard and different from an app processor; Apple won't have its own ready this year or next, protecting Qualcomm near-term." Made in 2023; "this year or next" has now passed.

> bullish · conviction 55 · "Under CEO Cristiano Amon, Qualcomm repositioned from a wireless company to edge/connected-devices, with automotive an important new growth market."

> neutral · conviction 60 · "With ~$6-10B annual free cash, Qualcomm will keep returning capital via buybacks and dividends plus selective bolt-ons (Nuvia), avoiding big transformative M&A." ADJUDICATED: correct on pattern, conservative on amount — fiscal-2025 free cash flow was $12.820B and $12.596B was returned. Note the $2.9B of goodwill added in fiscal 2026 is a larger bolt-on than the claim anticipated.

The two 2026 named-speaker claims, which carry the most current weight:

> 2026-01-18 · bullish · conviction 63 · thesis · speaker: Jordi Visser · role: independent · skill 2.0 · entity: QCOM

> "Qualcomm underperformed (down ~6% YTD) but fits advanced-packaging group and enterprise on-prem buildout; benefits from iPhone AI and agentic enterprise upgrade cycle."

> The highest-skill independent voice in the file, and the freshest bullish input. Note that it was made when the stock was down 6% year-to-date; it is now 35.2% below its 52-week high, so the call is currently underwater.

> 2026-04-07 · conviction 70 · thesis · speaker: Dylan Patel · role: independent · entity: TSM

> "AI chips are buying up all TSMC 3nm and 2nm capacity, squeezing Apple, Qualcomm and MediaTek off leading-edge nodes and toward Intel foundry."

> Tagged bullish on TSMC and BEARISH in substance for Qualcomm. It is the most specific supply-side risk in the lane and there is nothing in Qualcomm's filings addressing it.

One further claim used as colour: a 2025-11-04 observation that "US silicon leaders (AMD, Nvidia, Qualcomm on edge) are the partners of choice for building Gulf AI infrastructure."

Discards: a 2013 claim about integrated LTE modems (thirteen years stale); a 2023 claim about foundry concentration where Qualcomm is incidental; a 2023 claim about a mixed-reality headset (product-level); a 2018 trade-friction claim (eight years stale); a 2024 claim about InterDigital that mentions an ex-Qualcomm executive (different security); and the eight management-voice claims held out above.

Net conviction: mixed. Breadth 14 raw, roughly 4 effective after de-concentrating. Two claims carry named independent speakers; eight are management voice; the remainder are channel-attributed. No promoter-class voice appears. This lane's distinguishing feature is that it contains three claims from 2023 that are now testable against the filing, and two of the three called the current situation correctly — which is the best evidence in this batch that the knowledge base can produce gradeable foresight.

7. Data integrity — what we rejected from the vendor file and why

Ten findings.

1. epsdiluted is NULL in every income-statement row — annual and quarterly. All six annual and all nine quarterly rows return None. Every GAAP EPS figure here is computed as net income divided by weightedAverageShsOutDil and cross-checked against priceToEarningsDilutedRatioTTM of 18.5696, which at $162.67 implies trailing GAAP EPS of $8.76. This is the same zeroed-field defect found on UNP in this batch.

2. The trailing GAAP earnings series is dominated by a $5.7 billion tax item taken and released within three quarters — REJECTED as a valuation basis. The 10-Q: a $5.7 billion valuation allowance on federal deferred tax assets in the fourth quarter of fiscal 2025 following the One Big Beautiful Bill Act, released as a $5.7 billion income tax benefit in the second quarter of fiscal 2026 following Treasury Notice 2026-07. The effect: fiscal-2025 tax expense of $7,122M on pre-tax income of $12,663M (a 56.2% rate, against 2.2% in fiscal 2024), a Q4 fiscal-2025 net LOSS of $3,117M on operating income of $2,918M, and a Q2 fiscal-2026 net income of $7,370M on operating income of $2,309M. Deferred tax assets on the balance sheet went from $743M to $5,679M. The nine-month fiscal-2026 tax line is a net BENEFIT of $4,136M. All valuation in this dive runs on non-GAAP.

3. The June quarter's net income contains $726M of unrealised equity marks — 36% of the total. Investment and other income was $1,014M, of which $726M was "net gains on marketable securities... primarily... net unrealized gains related to certain marketable equity securities." Net income was $2,002M on operating income of $1,626M. On a clean basis the quarter earned closer to $1,430M. This is the GOOGL defect class and it is a second reason to reject the GAAP series.

4. netDebt omits $4,635M of short-term investments and is nine months stale — REJECTED and rebuilt. The vendor reports $8,530M; the filing supports $6,966M at 2026-06-28 (total debt $15,270M less cash $4,533M less marketable securities $3,771M). Consequential: netDebtToEBITDATTM 0.797x → 0.54x; enterpriseValueTTM $181.541B → $177.770B. We do not net the $2,323M of restricted cash.

5. The vendor's cash figure exceeds the filing's by exactly the restricted-cash balance. Vendor cashAndCashEquivalents for fiscal 2025 is $7,843M; the 10-Q's comparative column shows $5,520M — a $2,323M difference matching the restricted-cash line the filing discloses separately. Recorded. It does not change the corrected figure because restricted cash is excluded on both sides.

6. The vendor's enterprise value and net-debt fields disagree with each other by $2.2 billion. enterpriseValueTTM of $181.541B less market cap of $170.804B implies net debt of $10.737B, against netDebt of $8.530B in the same file. Neither is right; the corrected figure is $6.966B.

7. est.ebitAvg is BELOW est.netIncomeAvg in fiscal 2029 and 2030 — arithmetically impossible — and the EBIT/EBITDA rows are REJECTED. FY2030: ebitAvg $21.153B against niAvg $27.548B. FY2029: $18.315B against $19.391B. The ratios are additionally fixed at 34.04% and 29.69% of revenue in all eight years. All forward valuation runs on epsAvg.

8. seg_geo stops at fiscal 2023 and is three years stale — reported with the staleness stated. The most recent entry shows CHINA at $22,382M of $35,820M (62.5%) and the UNITED STATES at $1,259M (3.5%). This is the single largest unquantified risk in the file and there is no newer data anywhere in it. We report the fiscal-2023 numbers and draw no conclusion about the current mix.

9. A $2,916M increase in goodwill in nine months appears in no vendor field. Goodwill went from $11,358M to $14,274M, intangibles from $1,148M to $1,510M, inventories from $6,526M to $8,379M, and restricted cash from nil to $2,323M between 2025-09-28 and 2026-06-28. Qualcomm has made a material acquisition that the payload does not reflect. The extracted 10-Q does not name it in the sections retrieved, so we record the balance-sheet evidence and do not name the transaction.

10. lastDividend: 3.59 is the trailing-twelve-month total, not the last declared dividend. Fiscal-2025 dividends paid were $3,805M on roughly 1,090M average shares — $3.49 per share.

Share count — checked and verified exactly clean. The 10-Q cover states "The number of shares outstanding of the registrant's common stock was 1,050 million at July 27, 2026." Market cap of $170,803,500,000 divided by $162.67 gives 1,050,000,000 — an exact match to the million. No share-class or partial-unit error. The June-quarter diluted weighted average of 1,069M is the normal wedge.

quote.yearHigh/yearLow ($259.92 / $121.99) disagree with tech.hi52/lo52 ($251.02 / $124.07) by 3.4% and 1.7% — the second-largest such discrepancy in this batch. We use tech.

Vendor composite rating — partially accepted. rating gives A− / 4, with 5s on return on equity and return on assets, 4 on discounted cash flow, 3 on price-to-earnings and 2 on price-to-book and debt-to-equity. The return-on-equity 5 is computed on a GAAP earnings figure distorted by the tax item and is rejected as a lens; returnOnInvestedCapitalTTM of 17.4% is the honest measure and supports a high score by a sounder route. The valuation sub-scores are reasonable.

Non-equity tripwire — checked and passed. QCOM is NASDAQ-listed common stock. Price $162.67 is not par-like; beta 1.638 — the highest in this batch; the dividend is variable with a long record of increases; volume 15.32M shares (~$2.5B of turnover, the highest in this batch); the 52-week band of $124.07 to $251.02 is a 102% range. This is common equity.

8. Technicals

• Price $162.67. −35.2% from the 52-week high of $251.02; +31.1% above the low of $124.07. Position within the annual range: 30th percentile. The 102% range between high and low is the widest in this batch and is itself the story.
• Moving averages, both overhead: −17.2% below a 50-day average of $196.48 and −3.7% below a 200-day of $168.96. The 50-day is far ABOVE the 200-day, which is the classic broken-uptrend configuration — the stock fell fast enough that the short average has not yet crossed down.
• RSI 38.1 — weak. MACD −10.49 — the second-most negative reading in this batch.
• Maximum drawdown from peak over the trailing year: −35.2% — tied with the deepest here.
• Relative performance: 3-month −3.4% vs SPY +7.6%; 6-month +6.6% vs SPY +11.1%; 12-month +9.8% vs SPY +24.3% and QQQ +30.8% — a fifteen-point twelve-month deficit against the index and twenty-one points against the technology index.
• Sentiment: 1 strong buy, 29 buy, 34 hold, 5 sell across 69 analysts — the deepest coverage in this batch and the only "Hold" consensus. Target $204.05 (+25.4%), median $195, high $400 (+146%), low $120 (−26.2%) — a 3.3x spread.

Today's move and the entry

QCOM closed 2026-08-04 at $162.67, up 7.32% or $11.10 from $151.57. It opened at $156.60 — a 3.3% gap — traded $155.31 to $163.48, and closed within 0.5% of the day's high on 15.32M shares, the heaviest turnover in this batch. The entire semiconductor complex rallied on this session. No company-specific news appears in this file for 2026-08-04; the last event was the 2026-07-29 fiscal-third-quarter release six days earlier, which missed non-GAAP EPS by two cents.

The honest read. Like KLA in an adjacent batch, QCOM gapped from a deep drawdown into resistance rather than out of it: the 50-day average sits 20.8% overhead at $196.48 and the 200-day 3.9% overhead at $168.96. In pure chart terms this is a bounce inside a downtrend, and one strong session does not resolve it.

So the gap is not the reason to decline. The revenue trend is. Buying today means paying a 7.32% one-day move six days after a miss, ninety-two days before a print consensus already expects to be down 25% year on year.

What would change the answer:

• A price near $135 — roughly 13.0x FY2027E and 12.7x FY2026E — which would put our $180 base at +33% and the bear case only 10% below. That is 17% under today's close and inside the range this stock traversed in the last six months.
• Or the 2026-11-04 print beating $10.232B with a December guide showing sequential growth, which would mark the revenue trough and validate the fiscal-2028 recovery the bull case needs. At that point paying 15x is defensible.

9. Insiders

The insider file contains no material open-market activity in either direction and is reported as neutral. There are no open-market purchases and no discretionary open-market sales of size. With the stock 35.2% below its 52-week high for much of the last year, the absence of insider buying into that drawdown is a mild negative and no more than that.

The company itself has been the buyer, and at enormous scale: $8.791 billion of repurchases in fiscal 2025 — 5.1% of the current market capitalisation — taking diluted shares from 1,130 million to 1,069 million. That is the most economically significant insider transaction available and it is in the cash-flow statement rather than the Form 4s. It is also the single largest support under the per-share numbers while revenue falls, and its continuation is a named watch item.

10. Verdict, kill-criteria and flip conditions

Watch.

What is genuinely strong: $12.820 billion of free cash flow on $1.192 billion of capital expenditure — a 28.9% margin and a 7.5% yield; $12.596 billion returned to shareholders in fiscal 2025, a 7.37% total yield, the highest in this batch; corrected net debt of $6.966 billion at 0.54x EBITDA with 14.9x interest coverage and 17.4% return on invested capital; a QTL licensing annuity that produced $5.3-5.6 billion of revenue in each of four years at 85-90% gross margins while the silicon business swung $7.3 billion; 14.3x trailing non-GAAP earnings, the cheapest in this batch; a share count down 5.4% in seven quarters; and a 2.4:1 base-case asymmetry, the best here.

What we are declining to pay for: revenue down 4.0% year on year and 18.8% sequentially over two quarters; operating margin from 27.5% to 16.3% in six months; consensus non-GAAP EPS declining in both fiscal 2026 and fiscal 2027, with the fiscal-2028 recovery resting on nine analysts and an $11.17-$15.45 range; two customers at 43% of revenue, the larger up from 18% to 23% and a third fallen from 13% to below 10%; a geographic disclosure that stops at fiscal 2023 and shows China at 62.5% of revenue; a GAAP earnings series made meaningless by a $5.7 billion tax item taken and released; $726 million of unrealised marks in the most recent quarter's net income; a 7.32% one-day gap with the 50-day average 20.8% overhead; and a base fair value of $180 that is 10.6% above spot and 11.8% below the street's $204.05.

The distinction that matters. Qualcomm is genuinely cheap and the cheapness is not an illusion — the free cash flow is real, the balance sheet is clean, and the shareholder yield is being paid. What we are declining is to buy a shrinking business six days after a miss, ninety-two days before a print consensus already expects to be 25% down, with the entry including a 7% gap. In the Synthos frame, a name where our fair value sits below consensus and the operating trend is negative is a Watch — but this is the one name in this batch where the fill we want is close enough and the asymmetry good enough that a modest decline converts it.

Pre-registered KILL criteria — what would take this to Avoid:

• Fiscal fourth-quarter revenue below approximately $9.5 billion (against $10.232B consensus), or a December-quarter guide below $10.5 billion, which would mean the sequential decline is continuing rather than troughing.
• Either large customer reducing. Customer (x) at 23% or customer (y) at 20% falling materially would be a revenue event of a size the valuation cannot absorb.
• Gross margin below 52%. It has fallen from 55.6% to 53.1% in four quarters.
• The buyback slowing below roughly $6 billion annualised. It is doing roughly half the work of holding per-share earnings up and the yield is most of the expected return.
• Confirmation that leading-edge foundry access is constrained, as one independent knowledge-base claim argues — that would cap the fiscal-2028 recovery the bull case needs.

Pre-registered FLIP conditions — what would take this to Buy — Tactical:

• Price at or below approximately $135 (roughly 13.0x FY2027E). This is the most likely path to a Buy and it is 17% below today's close, inside the range this stock traversed in the last six months. At $135 the base case is +33% and the bear case only 10% below — genuinely asymmetric.
• The 2026-11-04 print beating $10.232B with a December guide showing sequential revenue growth — the trough confirmed.
• Automotive or internet-of-things revenue disclosed separately and growing at scale, which would give the fiscal-2028 consensus of $51.709B an observable basis rather than a hope.
• The concentration table showing diversification — customer (x) falling back toward 20% while total revenue grows.
• A reclaim of the 200-day average at $168.96 together with one of the above.

Where QCOM fits in the Synthos Framework Portfolio. The semiconductor and edge-compute sleeve, at 0% today with a 1.5-2% target on a fill near $135 or a confirmed November trough. On batch overlap: QCOM and ADI are the two semiconductor names here and they are not the same trade — ADI is an analog content-per-unit business in a violent cyclical upturn at 30.7x forward; QCOM is a handset and licensing franchise in a downturn at 15.3x. Owning both is a genuine barbell within the sector rather than a duplication, and if only one is to be owned, the choice is between paying up for the upturn and being paid 7.4% to wait through the downturn. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $162.67.

Single biggest risk: customer concentration, and it is worsening in the filing's own table. One customer was 23% of total revenue in the June quarter, up from 18% a year earlier. A second was 20%. Together they are 43% of the company. A third has fallen from 13% to below 10%. The knowledge base's 2023 claim that the biggest risk is "customers rolling their own chips" is no longer a forecast — the attrition of one 13% customer is visible in the disclosure. There is no vendor field, no ratio and no screen that shows any of this. At 15.3x forward with revenue already declining 4%, the loss or reduction of either large counterparty is not absorbed by the valuation, and the $122 bear case is what that looks like.

Most fragile assumption in the price: that fiscal 2028 revenue reaches $51.709 billion. That is a 16.8% increase on fiscal-2025 actual revenue of $44.284 billion, off a base currently shrinking 4% a year — and it can only be automotive, internet-of-things and edge diversification working at a scale that does not yet appear in any segment disclosure the payload contains. The fiscal-2028 EPS line that carries our entire bull case rests on nine analysts and spans $11.17 to $15.45. If fiscal 2028 revenue lands at $46 billion rather than $51.7 billion and EPS at $11.20 rather than $13.12, then at 15x the stock is $168 — three percent above today's price, two years from now, and the only return would have been the dividend and the buyback. Which, at 7.37% a year, is not nothing — and that is the honest reason this is a Watch rather than an Avoid.

Provenance & disclosures

• Traceability: 28 raw knowledge-base hits (21 entity, 7 text); 14 used; 14 discarded; breadth 14 raw and roughly 4 effective after de-concentration. Two attribution problems, both reported. First: EIGHT claims dated 2022-04-27 from a single channel are QUALCOMM MANAGEMENT VOICE — first-person strategy framing ("Snapdragon strategy is to be synonymous with premium Android", "Qualcomm is expanding beyond phones into next-gen PCs", "Apple is investing to build its own modem, but Qualcomm's mobile strategy is Snapdragon-driven") from a long-form interview. All eight are excluded from the conviction pool, quoted at half weight, and are four years stale. Second: NINE of the remaining claims come from ONE channel on ONE date (2023-05-27) — a single episode — so the raw count overstates breadth by roughly a factor of three. Two claims carry named independent speakers, both dated 2026: Jordi Visser (skill 2.0, 2026-01-18, bullish 63, currently underwater since the stock has fallen further) and Dylan Patel (2026-04-07, tagged bullish on TSMC but bearish in substance for Qualcomm on leading-edge foundry access). THREE 2023 claims are ADJUDICATED against the current filing and two aged remarkably well: "sales are cyclical around new-standard launches; with 5G largely built out, an elongated pause until 6G slows growth" — correct, and a precise description of 2026; "biggest risk: customers rolling their own chips" — directionally confirmed by a 13% customer falling below 10%; and "with ~$6-10B annual free cash, Qualcomm will keep returning capital via buybacks and dividends" — correct on pattern, conservative on amount ($12.820B actual). Discards: a 2013 modem claim, a 2018 trade-friction claim, a 2023 foundry claim where Qualcomm is incidental, a 2023 mixed-reality product claim, a 2024 InterDigital claim, and the eight management-voice claims. No promoter-class voice appears. Searches: Qualcomm, Snapdragon. All quotes are verbatim from the stored claim text.
• Data as-of: fundamentals — annual statements to 2025-09-28 (fiscal 2025) from the vendor payload; quarterly income statement to 2026-06-28; most recent filing-verified financials 2026-06-28, from the 10-Q filed 2026-07-29 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873600 = 2026-08-04T20:00:00Z ($162.67, +7.32%; 50-DMA $196.48; 200-DMA $168.96; RSI 38.1; MACD −10.49) · knowledge-base claims 2026-08-04. Fiscal note: Qualcomm's fiscal year ends the last Sunday of September. "FY2025" means the year ended 2025-09-28 and "Q3 FY2026" means the quarter ended 2026-06-28. All figures come from the Synthos vendor data file for QCOM or from the SEC filings in the QCOM archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2025-11-05 (fiscal 2025); 10-Q filed 2026-04-29 (March 2026 quarter); 10-Q filed 2026-07-29 (June 2026 quarter); 8-K filed 2026-06-24; 8-K filed 2026-07-29 (third-quarter results); 8-K filed 2026-07-31. All six carry preserved financial tables (tables: true), which is why the balance sheet, customer-concentration table, income-tax note and investment-income breakdown in this dive are filing-sourced.
• Where the filings contradicted or corrected the vendor (detailed in Section 7): the entire trailing GAAP earnings series rejected as a valuation basis, after the 10-Q identified a $5.7 billion valuation allowance on federal deferred tax assets taken in Q4 fiscal 2025 under the One Big Beautiful Bill Act and RELEASED as a $5.7 billion benefit in Q2 fiscal 2026 under Treasury Notice 2026-07 — producing a minus $3,117M quarter, a $7,370M quarter on $2,309M of operating income, a fiscal-2025 effective tax rate of 56.2% against fiscal-2024's 2.2%, and a nine-month fiscal-2026 tax BENEFIT of $4,136M; $726M of unrealised marks on marketable equity securities identified inside the June quarter's $2,002M of net income (36% of the total), with $605M in the nine months; net debt rebuilt from $8,530M to $6,966M (the vendor omits $4,635M of short-term investments and is nine months stale; the 10-Q shows total debt of $15,270M against cash of $4,533M and marketable securities of $3,771M), with enterprise value rebuilt from $181.541B to $177.770B and net-debt-to-EBITDA from 0.797x to 0.54x; the vendor's cash figure shown to exceed the filing's by exactly $2,323M — the restricted-cash line the filing discloses separately; the vendor's own enterprise value and net-debt fields shown to disagree by $2.2 billion; epsdiluted found NULL in all fifteen income-statement rows, with every GAAP EPS figure computed and cross-checked against priceToEarningsDilutedRatioTTM; customer concentration added entirely from the filing — 23%, 20% and below-10% against 18%, 21% and 13% a year earlier, with no vendor field of any kind; a $2,916M unexplained increase in goodwill identified between the two balance-sheet dates; est.ebitAvg rejected as arithmetically impossible (below netIncomeAvg in fiscal 2029 and 2030) and on a fixed-ratio signature (34.04% and 29.69% of revenue in all eight years); seg_geo reported with an explicit three-year staleness warning (most recent entry fiscal 2023, showing China at 62.5% of revenue); and lastDividend of $3.59 identified as the TTM total against $3.49 of fiscal-2025 dividends per share. Where vendor and filing AGREED — worth recording: the share count (10-Q cover's "1,050 million at July 27, 2026" against an implied market-cap count of 1,050,000,000 — an exact match to the million, the cleanest verification in this batch alongside ADI); the quarterly revenue series; and seg_prod summing to within 0.8% of reported revenue.
• Basis note: analyst estimates in est.epsAvg are non-GAAP EPS — verified because the fiscal-2025 row of $11.908 matches the sum of the earn_cal quarterly actuals and does not match computed GAAP diluted EPS. The GAAP series is additionally unusable for the reasons above, so the GAAP-to-non-GAAP gap is not a meaningful measure here and no trailing GAAP multiple is used in any conclusion. All forward and trailing valuation in this dive is on the non-GAAP basis; the trailing GAAP multiple of 18.6x is shown once, labelled tax-distorted, and not used.
• Estimate coverage: 18 analysts on FY2026 EPS and 24 on revenue; 21 and 26 on FY2027 — the anchor year and the deepest coverage in the file; only 9 and 24 on FY2028; 4 and 12 on FY2029; 5 and 12 on FY2030 — FY2029 and FY2030 are excluded entirely. The FY2028 EPS row, which carries the bull case, rests on 9 analysts with an $11.17 to $15.45 range, and that is disclosed at every point of use.
• Peer note: the vendor peer set mixes semiconductors with unrelated names — Uber and Intuit are not comparables, and Applied Materials, Lam Research and KLA are wafer-fab equipment rather than fabless design. The defensible names are Texas Instruments ($259.1B), Intel ($509.1B), Arm and loosely Sony. MediaTek, which the knowledge base names as Qualcomm's direct competitor with an "unspoken truce dividing the market by price tier," is NOT in the payload. ADI, also covered in this batch, appears superficially comparable and is not. No peer multiples are supplied, so no peer-multiple comparison is drawn.
• Fair-value caveat: the $122 / $180 / $260 anchors are multiples of the fiscal-2027 and fiscal-2028 consensus non-GAAP EPS distributions — 13x the FY2027 low of $9.42, 17.4x the FY2027 mean of $10.346 (21 analysts), and 16.8x the FY2028 high of $15.45 (9 analysts, disclosed). Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 15x FY2027E gives $155, 4.7% BELOW spot; 20x gives $207, essentially the street's target. The whole answer lives in a 15-20x band. Note that the base case assumes modest multiple EXPANSION — unusual for this house — on the reasoning that 15.3x on a trough year with a 7.4% shareholder yield and a 0.54x levered balance sheet is a distressed multiple; that reasoning is stated in Section 5b so it can be graded. The bear case sits 1.6% above the street's own $120 low target; the street's $400 high is treated as an outlier and no anchor approaches it.
• Timing: fiscal third-quarter 2026 results were released 2026-07-29, six days before this dive, MISSING non-GAAP EPS by two cents ($2.21 against $2.23) while beating on revenue ($9,947M against $9,695M). The next print is 2026-11-04, 92 days away, and consensus for it is already minus 25% year on year on earnings and minus 9.2% on revenue. 2026-08-04 was a sector-wide semiconductor up-day and QCOM's +7.32% carried no company-specific news in this file.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.