AI Platforms: GOOGL, MSFT, AAPL – Research Note (3 Oct 2026)

Key takeaways

Three ways to own AI. Alphabet is the full-stack builder (TPUs, Gemini, Cloud, Search distribution); Microsoft monetises AI through enterprise software plus Azure; Apple is a low-capex device distributor whose AI story (Siri AI) is still unproven commercially.
Valuation leaves little margin of safety. Probability-weighted values: GOOGL US$350 (+1.9%), MSFT US$534 (+3.1%), AAPL US$267 (-19.9%). GOOGL and MSFT are fairly priced on our base case; AAPL needs earnings above our bull case to earn 8% a year. Estimate
Capex is the swing factor. The largest hyperscalers are on track for roughly US$800bn of 2026 capex [S18]. Alphabet guides US$195–205bn for 2026 [S7]; Microsoft spent US$115.9bn of cash capex in FY26 (plus finance leases) and expects FY27 to be higher [S9][S11]. Free cash flow has been squeezed (GOOGL H1-26 FCF US$4.3bn).
The downside is driven by the multiple. Bear cases are −46% (GOOGL), −34% (MSFT) and −48% (AAPL). These come mainly from lower P/E multiples, not from collapsing earnings. MSFT has the most even risk/reward because recurring software revenue supports its earnings.
The late-October prints are the next test. Watch Search growth against agentic assistants (Meta's Muse) and model price cuts; Azure growth vs the 44–45% cc guide; and Apple's December-quarter guide and gross margin as tariff-refund help fades and memory costs rise.

Investment thesis

The AI investment cycle has moved from who has the best model to who can convert compute spending into per-share cash flow. All three companies have kept growing revenue strongly through 2026 (latest quarter: GOOGL +24.2%, MSFT +18%, AAPL +16.4% y/y) [S6][S8][S12]. Their capital needs differ sharply, though, and so does the quality of their reported earnings.

GOOGL – cheapest multiple, heaviest reinvestment. Trades at ~23x our 2027E core EPS. Reported 2026 EPS is inflated by ~US$136bn of H1 securities gains, so we value core operating EPS only. Upside depends on Cloud (+82% in Q2, ~US$514bn backlog [S16]) and Search holding up against agents.
MSFT – highest quality, fairly priced. US$678bn commercial RPO (+84%), Azure +43%, Copilot >30m paid seats [S8]. At ~26.5x FY27E core EPS the market already pays for strong growth. Finance-lease-funded capacity (lease liabilities US$66.6bn, up from US$46.2bn a year earlier [S9]) is the hidden cost.
AAPL – best cash conversion, most demanding price. 9M FY26 FCF of US$110bn and continuing buybacks, but at ~33x FY27E core EPS the price implies FY28 EPS of ~US$13.59, above our bull case of US$13.28. Estimate

Interpretation, not fact: stances are analytical labels on a fixed 8% required return. They are not recommendations to buy or sell.

Industry drivers

1. Hyperscaler capex and the depreciation wave

Goldman Sachs estimates the five largest hyperscalers will spend about US$800bn on 2026 capex, up ~94% y/y [S18]. Reported plans include Microsoft ~US$175bn for calendar 2026 and Meta US$130–145bn [S18]. S&P has warned that capex is rising faster than expected and financing is getting more complex [S18]. That spending becomes depreciation in 2027–28, which is why our models carry D&A rising sharply and margins flat to down.

2. Token price war

On 22 Sep OpenAI and Anthropic cut prices on new models; OpenAI's GPT-6 Sol/Luna cost about half as much per token as their predecessors [S16]. Google Cloud's CEO claims a payback period of under 2 years on AI servers, and about half that on its own TPUs (management claim, unaudited) [S16]. If prices fall faster than unit costs, cloud AI margins compress.

3. Agents vs. search and app distribution

Meta's Muse assistant reached No. 1 on the US App Store on 21 Sep. Alphabet fell nearly 4% on 23 Sep as investors priced in disintermediation risk [S16]. Against that, an Evercore survey found 78% of respondents named Google their primary search engine in Aug 2026, up from a 70% low [S16]. Alphabet launched Gemini 4 "Argon" on 30 Sep [S17].

4. Custom silicon vs. merchant GPUs (NVDA reference)

NVIDIA's Q2 FY27 (to 26 Jul 2026) revenue was US$96.2bn (+106%), with Data Center at US$89.0bn (+117%) and Q3 guidance of ~US$108bn [S19]. It trades at 19.4x forward earnings, below all three platforms, which suggests the market discounts how long capex-driven demand will last. Alphabet's TPUs, along with in-house silicon programmes at the other large clouds (AWS, Azure), aim to bend the cost curve.

5. Regulation

Google: a federal judge declined to force a sale of the ad exchange and ordered behavioural remedies (2 Sep 2026) [S16]; Search distribution remedies are under cross-appeal [S6]. Apple: EU DMA penalties and Epic anti-steering limits on App Store commissions [S12]. Microsoft: AI-partnership and cloud-licensing scrutiny, which we treat as a standing risk; no new formal action was verified this month.

Chart 3 – Fact. Latest reported period cash capex vs FCF (US$bn). Periods differ by company (labelled). Microsoft figures exclude finance-lease additions. Sources: [S6][S9][S12].

Coverage table

Ticker	Price (US$)	Mkt cap (US$ tn)	Fwd P/E cons.	P/E on next-yr core EPS (model)	YTD	Bear / Base / Bull (US$)	Prob-weighted (US$)	vs price	Entry zone (US$)	Stance
GOOGL Alphabet (Class A)	343.50	4.20	25.6x	23.3x	+9.7%	184 / 372 / 544	350	+1.9%	293–323	Neutral
MSFT Microsoft	517.53	3.84	26.2x	26.5x	+7.0%	344 / 565 / 741	534	+3.1%	446–490	Neutral
AAPL Apple	333.69	4.87	36.3x	32.8x	+22.7%	175 / 276 / 384	267	-19.9%	218–239	Cautious
NVDA supplier reference	233.95	5.64	19.4x	n/a	+25.4%	Not modelled – shown as AI-capex supplier reference				—

Facts: price = Nasdaq last sale, close 2 Oct 2026, delayed [S1]; market cap per Nasdaq [S2]; YTD vs 31 Dec 2025 close (Yahoo chart API) [S3]; forward P/E = consensus per stockanalysis.com [S4]. Estimates: scenario values, PW value, model P/E (on 2027E core EPS for GOOGL, FY27E for MSFT/AAPL) and entry zones, from the author's model (Methodology). Entry zone = price range that would give a 15–20% annualised base-case return to end-2028. It is an analytical level, not a price forecast.

Chart 1 – Estimate. Scenario present values (US$/share) vs 2 Oct close.

Revenue and profit mix

Chart 2 – Fact. Latest reported periods (US$bn); periods differ by company. GOOGL "other" = subscriptions, platforms & devices, Other Bets and hedging (balancing item). Sources: [S6][S8][S12].

GOOGL – Full-stack AI at the lowest multiple

US$343.50 (+1.56% on day) · mkt cap US$4.20tn · 52-wk 235.84 – 408.61 · div US$0.88/yr · Neutral

Business mix Fact

H1 2026 revenue US$229.7bn (+23.1%): Search & other 123.7 (+17.9%), YouTube ads 20.9 (+11.8%), Google Cloud 44.8 (+73.1%), subscriptions/platforms/devices & other 40.3 [S6].
Cloud H1 operating income US$15.4bn (34.4% margin), vs 19.3% a year earlier [S6].

Latest quarter Fact

Quarter	Q2 2026 (to 30 Jun), reported 22 Jul 2026 [S6]
Revenue	US$119.8bn, +24.2% y/y
Operating income / margin	US$40.8bn / 34.0% (+1.6pt)
Search / YouTube / Cloud	63.3 (+16.8%) / 11.1 (+12.9%) / 24.8 (+81.8%); Cloud margin 35.6%
Diluted EPS	US$9.11 reported, incl. ~US$99bn securities gains; core well below. Fact
Cash flow	Q2 OCF 39.1 – capex 44.9 = FCF −5.9; H1 FCF 4.3
Balance sheet 30 Jun	Cash+securities 242.5, debt 100.2, net cash 142.3; H1 raised 30.5 equity, 19.1 mandatory convertible preferred, 56.2 net debt; no buybacks; shares +1.17% YTD

Guidance Fact – management

2026 capex US$195–205bn (raised from 180–190); 2027 capex to "significantly increase" (Q2 call, 22 Jul) [S7].
Implied H2-26 capex US$114–124bn. Estimate

Since the last report (news flow to 2 Oct)

2 Sep – ad-tech case: court declined to force divestiture of the ad exchange; behavioural remedies [S16].
8 Sep – Cloud CEO: AI server payback <2 yrs, own silicon about half that (unaudited claim) [S16].
21–23 Sep – Meta Muse tops US App Store; OpenAI/Anthropic price cuts; GOOGL closed US$337.83 on 23 Sep (down nearly 4%) [S16].
30 Sep – Gemini 4 "Argon" launched; a same-day Bloomberg report described internal doubts about it [S17]. Quarterly dividend US$0.22 (ex-date 4 Sep) [S2].

Bull case

Search keeps growing double digits as AI Overviews/AI Mode monetise; survey share rising [S16].
Cloud backlog (~US$514bn) converts; TPU cost advantage keeps Cloud margins above 35%.
Bull 2028E core EPS US$21.22 at 30x → US$544 PV.

Bear case / strongest counter-argument

Strongest counter-argument: agents (Muse, ChatGPT, Gemini rivals) take over commercial queries, so Search volume, not just price, erodes.
Growth is being financed with equity, convertibles and debt rather than per-share FCF; D&A rises from ~US$48bn (2026E) to ~US$100bn (2028E) in our base model.
Bear 2028E EPS US$10.69 at 20x → US$184.

Catalysts

Q3 2026 results – late Oct (est. 28 Oct; not yet confirmed by company) [S4][S5]. Consensus revenue ~US$127bn (~+24%) [S16].
Gemini 4 adoption and pricing over Oct–Dec.
Search-case appeals; ad-tech remedy implementation.
2027 capex guide with Q4 results (typically early Feb 2027).

Falsification signals Assumption – research discipline

Search growth <8% for two consecutive quarters with falling operating margin.
Cloud growth <25% or Cloud margin <25%.
2027 capex >US$260bn without matching backlog/revenue explanation.
2028E owner FCF (after SBC) forecast <US$40bn.

Valuation scenarios Estimate

Scenario (prob.)	Core EPS 2026E / 2027E / 2028E (US$)	Terminal P/E · EV/EBIT	P/E value · EV value	Blended PV (US$)	vs price	Annualised to end-2028
Bear (30%)	10.97 / 10.91 / 10.69	20x · 16x	182 · 187	184	-46.3%	-18.1%
Base (50%)	12.57 / 14.77 / 16.82	26x · 21x	370 · 374	372	+8.2%	+11.8%
Bull (20%)	13.51 / 17.31 / 21.22	30x · 25x	537 · 550	544	+58.3%	+32.5%
Probability-weighted	30% bear / 50% base / 20% bull			350	+1.9%	—

Estimate. Reverse-engineered: at today's price and the base multiple, an 8% return needs terminal EPS of US$15.62 vs 2028E base EPS 16.82. Base-multiple ±4x moves the P/E value to US$313–426.

MSFT – Enterprise AI monetisation, priced for it

US$517.53 (+0.92% on day) · mkt cap US$3.84tn · 52-wk 349.20 – 553.72 · div US$3.92/yr · Neutral

Business mix Fact

Q4 FY26 revenue US$90.0bn: Productivity & Business Processes 37.8 (+14%), Intelligent Cloud 39.3 (+32%), More Personal Computing 12.9 (−4%) [S8].
FY26 (to 30 Jun 2026): revenue US$331.8bn (+18%), operating income US$155.2bn (+21%, 46.8% margin), Microsoft Cloud Q4 US$59.3bn (+27%) [S8][S9].

Latest quarter Fact

Quarter	Q4 FY26 (to 30 Jun), reported 29 Jul 2026 [S8]
Revenue / operating income	US$90.0bn (+18%) / US$40.6bn (+18%)
Diluted EPS	GAAP US$4.81 (+32%); non-GAAP (ex OpenAI) US$4.74 (+23%). Includes a US$3.2bn Anthropic gain; company cites a US$0.27 net benefit from discrete items vs guidance
AI KPIs	Azure & other cloud +43%; Azure >US$100bn annual revenue; M365 Copilot >30m paid seats; commercial RPO US$678bn (+84%)
Cash flow FY26	OCF 182.9 – cash capex 115.9 = FCF 67.0; Q4 FCF 19.6 (OCF 55.4, capex 35.8) [S9]
Balance sheet 30 Jun	Cash+ST investments 76.8, debt 40.3, finance-lease liabilities 66.6 (vs 46.2 a year ago); FY26 buybacks 22.3, dividends 26.4; diluted shares 7.453bn [S9]

Guidance Fact – management

FQ1 FY27 (Sep qtr) revenue US$89.9–91.0bn; Azure growth 44–45% in constant currency; demand exceeds supply (secondary report of the call) [S11].
FY27 capex to rise y/y; Q4 FY26 capex incl. finance leases ~US$41bn [S11].
Quarterly dividend raised 8% to US$0.98 (15 Sep), payable 10 Dec [S10].

Since the last report (news flow to 2 Oct)

14 Sep – Microsoft set limits on future frontier-model development (industry-wide throttling debate) [S18].
25 Sep – redesigned Copilot (Home, Code, Autopilot) and consumption-based "Copilot Credits"; stock +3.7% that day [S11].
YTD +7.0% but up ~48% from its 52-week low of US$349 [S2][S11].

Bull case

RPO of US$678bn gives multi-year revenue visibility; Azure growth above 40% sustained into FY28.
Copilot moves from per-seat to seat + consumption pricing, lifting revenue per user at software margins.
Bull FY29E core EPS US$28.87; terminal 32x → US$741 PV.

Bear case / strongest counter-argument

Strongest counter-argument: capacity bought via finance leases and capex turns into depreciation faster than Copilot revenue arrives. The operating margin falls toward 40% while the stock is still priced as a 46%-margin compounder.
Backlog concentration among a few AI-lab customers is not fully disclosed. Competition from Gemini on Windows and Workspace pressures Office seat pricing.
Bear FY29E EPS US$19.31 at 22x → US$344.

Catalysts

FQ1 FY27 results – late Oct (est. 28 Oct by stockanalysis; Nasdaq/Zacks algorithm shows 4 Nov; unconfirmed) [S4][S5].
Copilot Code / Autopilot broader rollout after the Frontier programme (end-Sep) [S11].
Ignite conference (typically November) for AI agent pricing.
FY27 capex trajectory at the January (FQ2) print.

Falsification signals Assumption – research discipline

Azure cc growth <30% for two consecutive quarters while capex keeps rising.
Full-year operating margin <42%.
Commercial RPO growth <25% y/y.
FY27 FCF after finance-lease principal turns negative.

Valuation scenarios Estimate

Scenario (prob.)	Core EPS FY27E / FY28E / FY29E (US$)	Terminal P/E · EV/EBIT	P/E value · EV value	Blended PV (US$)	vs price	Annualised to end-2028
Bear (30%)	17.86 / 18.23 / 19.31	22x · 17x	355 · 332	344	-33.6%	-10.0%
Base (50%)	19.51 / 22.06 / 25.05	28x · 23x	563 · 567	565	+9.1%	+12.3%
Bull (20%)	20.44 / 24.50 / 28.87	32x · 27x	726 · 756	741	+43.2%	+26.7%
Probability-weighted	30% bear / 50% base / 20% bull			534	+3.1%	—

Estimate. Reverse-engineered: at today's price and the base multiple, an 8% return needs terminal EPS of US$21.64 vs calendar-2028 EPS proxy (avg FY28/FY29) 23.55. Base-multiple ±4x moves the P/E value to US$483–642.

Model notes (Assumptions). FY26 base: revenue US$331.8bn, EBIT US$155.2bn. Base growth FY27/28/29 17%/15%/13%, EBIT margin 46%/45%/45%; bear 12%/9%/8% and 44%/41%/40%; bull 20%/18%/16% and 47%/47.5%/48%. Tax 19%, net share count −0.5%/yr. FY26 core EPS proxy US$16.87, vs non-GAAP US$17.28 which includes the Anthropic gain. EV method uses a net position of −US$30bn at end-2028 in the base case (cash minus debt and finance leases, −US$30.0bn at 30 Jun 2026); bear −60, bull 0.

AAPL – The cash machine whose AI must still pay

US$333.69 (+1.02% on day) · mkt cap US$4.87tn · 52-wk 243.42 – 345.34 · div US$1.08/yr · Cautious

Business mix Fact

9M FY26 revenue US$364.4bn (+16.2%): iPhone 196.5 (+22.4%), Services 91.7 (+14.1%), Mac/iPad/Wearables 76.1 [S12].
9M gross profit: Products 108.8 (39.9% margin), Services 70.0 (76.3%). Services is 25% of revenue but 39% of gross profit [S12].

Latest quarter Fact

Quarter	Q3 FY26 (to 27 Jun), reported 30 Jul 2026 [S12]
Revenue	US$109.4bn, +16.4%; iPhone 54.3 (+21.7%); Services 30.7 (+12.1%); Greater China 18.8 (+22.4%)
Gross margin	50.1%, incl. ~2pt from tariff refunds (non-recurring)
Diluted EPS	US$2.02 (+28.7%); ~US$1.91 ex tariff refund Estimate
Cash / capital return	9M OCF 117.0 – capex 6.8 = FCF 110.2; buybacks 62.1; dividends 11.8; shares −1.11%
Balance sheet 27 Jun	Cash+securities 146.5, debt 84.3, net cash 62.2; R&D +32.5% (infrastructure incl. AI)

Guidance Fact – management

FQ4 (Sep qtr) revenue growth 9–11%; gross margin 47–48%, including ~1pt tariff-refund benefit; component-cost relief expected to shrink after September (secondary reports of the call) [S14].

Since the last report (news flow to 2 Oct)

9 Sep – iPhone 18 Pro/Pro Max and foldable iPhone Duo launched (A20 Pro, iOS 27, Siri AI); on sale 18 Sep [S13][S15].
15 Sep – GF Securities called pre-orders "lukewarm"; 23 Sep – UBS saw Pro lead times up 2–3 days, kept Neutral (US$296 PT) [S15].
30 Sep – Counterpoint: iPhone 18 Pro family China launch-window units +12% vs 17 Pro (a three-day vs one-week comparison) [S15].
Siri AI not available in China pending regulatory approval; initially unavailable in the EU on iPhone [S13].

Bull case

Siri AI drives a multi-year upgrade cycle; foldable Duo lifts ASPs.
Services compound at 12–14% with stable take rates; buybacks shrink share count ~1–2%/yr.
Bull FY28E EPS US$13.28 at 34x → US$384.

Bear case / strongest counter-argument

Strongest counter-argument: AI does not raise willingness to pay for iPhones. The FY26 China rebound and tariff refunds flatter a cyclical peak, while Epic/DMA rulings erode App Store economics.
Memory cost inflation pressures product gross margin after September.
Bear FY28E EPS US$8.48 at 24x → US$175.

Catalysts

FQ4 FY26 results – late Oct (est. 29 Oct; unconfirmed) [S4][S5]; December-quarter guide is the key read on iPhone 18.
Siri AI user-beta rollout timing; China approval.
Epic anti-steering case at the US Supreme Court; EU DMA appeals.

Falsification signals Assumption – research discipline

Normalised product gross margin <37% or Services growth <10% for two consecutive quarters.
FY27E core EPS falls below US$9.50.
Greater China revenue declines y/y for two quarters.
Buybacks funded by net new debt.

Valuation scenarios Estimate

Scenario (prob.)	Core EPS FY26E / FY27E / FY28E (US$)	Terminal P/E · EV/EBIT	P/E value · EV value	Blended PV (US$)	vs price	Annualised to end-2028
Bear (30%)	8.62 / 8.44 / 8.48	24x · 19x	173 · 177	175	-47.5%	-18.9%
Base (50%)	9.05 / 10.18 / 11.01	29x · 24x	271 · 280	276	-17.4%	-0.8%
Bull (20%)	9.35 / 11.58 / 13.28	34x · 28x	382 · 386	384	+15.0%	+14.9%
Probability-weighted	30% bear / 50% base / 20% bull			267	-19.9%	—

Estimate. Reverse-engineered: at today's price and the base multiple, an 8% return needs terminal EPS of US$13.59 vs FY28E base EPS 11.01. Base-multiple ±4x moves the P/E value to US$234–308.

Key risks (sector-wide)

Overbuild / depreciation: 2026–27 capex becomes 2027–29 depreciation. If utilisation or pricing disappoints, margins and multiples fall together, which is the main path to our bear cases.
AI price deflation: token prices halving per model generation [S16] could outrun unit-cost declines.
Disintermediation: agentic assistants could bypass search, app stores and office suites, the toll booths these three companies own.
Regulation and litigation: Search/ad-tech remedies (GOOGL), App Store rulings (AAPL), cloud/AI partnership scrutiny (MSFT).
Supply chain and geopolitics: Asia-concentrated hardware assembly, tariffs, export controls, China AI-feature approval.
Power, memory and chips: capacity constraints delay revenue recognition; memory cost inflation hits devices.
Rates and multiples: at a 6% vs 10% discount rate, base values move about ±4%. The terminal multiple matters far more: ±4x moves values by roughly ±14–16%.

What to watch in the next 1–2 quarters

Item	When	Supportive	Negative
GOOGL Search growth	Q3 print, late Oct	≥15% y/y with stable TAC	<10%, or paid clicks falling
Google Cloud growth / margin	Q3 print	>60% growth, margin ≥33%	Growth <40% or margin slipping after Gemini price cuts
Azure cc growth	FQ1 print, late Oct	≥45% (at or above guide)	<40%, capacity-constrained misses
MSFT capex incl. leases	FQ1/FQ2	Growth slowing relative to RPO conversion	>US$50bn/qtr with margin dip
Apple Dec-qtr guide	FQ4 print, late Oct	Double-digit growth, GM ≥47% ex refunds	Single-digit growth, GM <46%
NVDA FQ3 (supplier read-through)	est. 18 Nov	In line with ~US$108bn guide, orderly	Big beat (more capex pressure) or miss (demand doubt)
2027 capex guides	Jan–Feb 2027	Growth decelerating	Another step-up without revenue to match

Methodology & assumptions

Normalised (core) EPS: EBIT × (1 − tax) ± after-tax net interest, divided by diluted shares. Excludes mark-to-market investment gains (Alphabet's securities gains; Microsoft's OpenAI/Anthropic effects; Apple's one-off tariff refund). Tax: GOOGL 18%, MSFT 19%, AAPL 17% (base). Assumption
Two valuation methods, equally weighted: (1) terminal price = terminal EPS × scenario P/E; (2) (terminal EBIT × EV/EBIT + terminal net cash) ÷ terminal shares. Terminal date is 31 Dec 2028; the terminal year is 2028 (GOOGL), FY28 to Sep-2028 (AAPL), and the average of FY28/FY29 to June (MSFT).
Discounting: present value = terminal price ÷ 1.08^2.245 + PV of the current annualised dividend. 8% is a required equity return, not a WACC.
Probabilities: 30% bear / 50% base / 20% bull. These are subjective and skewed conservative; they have not been statistically validated.
Refresh scope: the GOOGL and AAPL operating forecasts are carried over from the author's 12 Sep 2026 model, because no newer quarter has been reported since. Price, horizon and dividends were refreshed on 3 Oct. The MSFT model is new, built from the FY26 10-K. Main GOOGL drivers: Cloud growth 2027/28 45%/32%, Cloud margin 36%, capex US$200/240/240bn (2026–28). AAPL: products/services growth 7%/12% (FY27), 5%/11% (FY28).
Entry zone: prices at which the base-case terminal value plus dividends gives a 15–20% annualised return to end-2028.
Limitations: no paid consensus data. Forward P/E figures are third-party consensus with opaque definitions. Some guidance items come from secondary reports of earnings calls (flagged). Prices are delayed closes, not real-time. Model risk dominates decimal precision.

Sources (with timestamps)

[S1] Nasdaq quote-info API, GOOGL/MSFT/AAPL/NVDA – lastSalePrice, lastTradeTimestamp "Oct 2, 2026", isRealTime=false, market Closed. Retrieved 2026-10-03 03:02–03:10 UTC. api.nasdaq.com/api/quote/<sym>/info
[S2] Nasdaq summary API – market cap, 52-week range, annualised dividend, ex-dates. Retrieved 2026-10-03 03:03–03:10 UTC.
[S3] Yahoo Finance chart API (range=ytd) – 31 Dec 2025 close: GOOGL 313.00, MSFT 483.62, AAPL 271.86, NVDA 186.50. Retrieved 2026-10-03 03:04–03:10 UTC.
[S4] stockanalysis.com overview pages – forward P/E, estimated earnings dates. Retrieved 2026-10-03 03:05–03:10 UTC. stockanalysis.com/stocks/<sym>/
[S5] Nasdaq earnings-date API (Zacks algorithmic estimate). Retrieved 2026-10-03 03:15 UTC. Company confirmations not yet found on IR sites at retrieval time.
[S6] Alphabet Q2 2026 Form 10-Q and earnings release, 22 Jul 2026. SEC 10-Q · release
[S7] Alphabet Q2 2026 earnings call, 22 Jul 2026 (capex guide). abc.xyz
[S8] Microsoft FY26 Q4 earnings press release, 29 Jul 2026. Retrieved 2026-10-03 03:10 UTC. microsoft.com/investor
[S9] SEC XBRL company facts, Microsoft (CIK 789019), FY26 Form 10-K filed 29 Jul 2026 – OCF, PP&E purchases, buybacks, dividends, cash, debt, finance-lease liabilities, diluted shares. Retrieved 2026-10-03 03:11 UTC. data.sec.gov
[S10] Microsoft dividend announcement, PR Newswire, 15 Sep 2026. prnewswire.com
[S11] TIKR, "Microsoft Stock Is Up 48% From Its Low as Azure Growth Heads Toward 45%", 26 Sep 2026 (secondary report of FQ1 guide and Copilot launch). tikr.com
[S12] Apple Q3 FY26 Form 10-Q (31 Jul 2026) and results release (30 Jul 2026). SEC 10-Q · release
[S13] Apple Newsroom: iPhone 18 Pro and iPhone Duo (9 Sep 2026); Siri AI (8 Jun 2026). apple.com/newsroom
[S14] TIKR, "Apple Stock Is Up 34% Over the Past Year", 26 Sep 2026; Trefis, "What Will Decide Apple Stock Over The Next Six Months?", 29 Sep 2026 (secondary report of FQ4 guide). tikr.com
[S15] TECHi / Counterpoint China launch data, 30 Sep 2026; Yahoo Finance / UBS lead times, 23 Sep 2026; Investor's Business Daily / GF Securities, 15 Sep 2026. techi.com
[S16] TIKR, "Alphabet Stock Fell Nearly 4% in a Day on Muse and AI Price Cuts", 25 Sep 2026 (Muse, price cuts, Kurian payback claims, ad-tech remedy, ~US$514bn Cloud backlog, Q3 consensus). Retrieved 2026-10-03 03:05 UTC. tikr.com
[S17] Barron's, "Google Unveils New AI Model Gemini 4 Argon", 30 Sep 2026; Investing.com, report on internal doubts over Gemini 4, 30 Sep 2026. barrons.com
[S18] Yahoo Finance (Goldman Sachs ~US$800bn 2026 hyperscaler capex), 26 Sep 2026; 24/7 Wall St (Microsoft ~US$175bn, Meta US$130–145bn capex plans; AI-model limits), 14 Sep 2026; Axios (S&P on AI debt), 11 Sep 2026; CNBC, "Microsoft sets limits for future AI models", 14 Sep 2026. axios.com
[S19] NVIDIA Q2 FY27 results as summarised by TIKR (26 Sep 2026) and roic.ai (2 Oct 2026): revenue US$96.2bn, Data Center US$89.0bn, Q3 guide ~US$108bn. tikr.com
[S20] Author's working scenario model and fact packs, 12 Sep 2026 (filings-based; GOOGL/AAPL operating drivers reused here).

Disclaimer

Personal research notes for information only. Not investment advice. Figures marked Estimate or Assumption are the author's model outputs and may be wrong. Market data are delayed closing prices from public sources and may contain errors. Past performance does not predict future returns. Do your own research or consult a licensed adviser.