Skip to content
Some content is members-only. Sign in to access.

Company Fundamentals Analysis

By KAPUALabs

The fundamental record of Broadcom Inc. presents a question that the material can answer only in part. Revenue and margins have strengthened together, and the artificial intelligence (AI) business is the principal engine of that strength. The balance sheet, capital deployment and the earnings basis for valuation are less settled. Throughout this chapter, reported figures are distinguished from derived calculations and from interpretation, and every figure is traced to the reference markers carried in the source material.

1. Introduction and Data Basis

The material draws on a mixture of sources: reported results, quarterly commentary, management statements, aggregator projections and social-media posts. It does not identify a primary SEC filing for each figure, and it does not state whether each figure is reported, adjusted or trailing twelve months (TTM). Where the basis is unstated, this analysis says so. The Form 10-Q and earnings transcript citations that the brief requests cannot be confirmed from the supplied material, and no such filing is cited here by date.

The VMware acquisition is the most important consideration for historical comparability, and the material does not quantify it. The sources mention acquisitions that have expanded the infrastructure software business 53 and integration risk that accompanies them 55, and they note European Union antitrust scrutiny related to VMware 37,55,57. They do not report VMware's contribution to revenue, its purchase price, its pro-forma effect on margins or its effect on the prior-year base. The reader should therefore treat year-on-year growth rates as containing an unquantified acquisition component.

Data unavailable: Segment revenue for Semiconductor Solutions and Infrastructure Software; segment margins; VMware revenue contribution; VMware pro-forma adjustments; VMware synergy targets and realised synergies; EBITDA; operating cash flow; cash and net debt; interest expense; return on equity; return on invested capital.

The valuation figures also carry a date problem. Market data is cited for October 2026, while the most recent balance-sheet figure is dated November 2025 55,83. The balance sheet is therefore the oldest financial input relative to the market data and should be read as a dated snapshot.

2. Financial Performance

Revenue

Fiscal 2025 revenue was nearly $63.9 billion 28,30,54,55,79,80,82,98, growing roughly 24% on fiscal 2024 53,54,57,79,82,98. A second source gives the growth as approximately 23.9% 52,80. These figures agree closely and are the most robust annual result in the material.

Within that total, the AI-chip business is the clearest identified driver. AI-chip sales rose 65% to $20 billion in fiscal 2025, equal to 31% of total revenue 24,26,27,87, with growth attributed to custom accelerators and AI networking 31,42,99. The material offers no equivalent breakdown for infrastructure software, so the contribution of VMware software revenue to the total cannot be isolated. The semiconductor cycle is likewise unquantified: the material says nothing about inventory levels or cyclical pricing, and the cyclical risk discussed in Section 7 therefore rests on assessment rather than on reported data.

The most recent quarter shows acceleration. Third-quarter fiscal 2026 net revenue reached $29.6 billion, up 86% from $16.0 billion a year earlier 45,68,75,78,83, a figure corroborated across six sources. Growth exceeded the 48% recorded in the second quarter 68, and AI sales were the principal source of that acceleration 68. The AI-chip segment generated $16.7 billion in the quarter, more than tripling year over year 73, and a 221% growth figure appears in more than one source 66,99.

These quarterly AI figures carry a caveat that the annual figures do not. One item states that the $16.7 billion, the 221% growth and the $21.7 billion guidance figure appear in a Reddit comment and are neither verified nor detailed in the post itself 95. A Reddit commenter was also the first to assert the $16.7 billion figure 95. The direction of the AI growth is well supported. The precise quarterly AI numbers rest on thinner evidence than the multi-source headline revenue. A separate measure of 106% year-over-year AI-chip revenue growth 1,2,3,4,5,6,8,9,11,13,14,16,21,26,27,36,38,56,68 is not reconciled with the 221% figure, and the material does not say whether differing periods or definitions explain the gap.

Profit Margins

Net income for fiscal 2025 was $23.1 billion 30,52,54,55,57,79,80,82, and the net margin of about 36.2% 28,80 was reported by several outlets. The operating margin of 48.6% is described as an all-time high 79, which indicates that profitability has strengthened alongside revenue.

Gross margin is the less settled measure. One figure of about 64.96% rests on a 24-source base 6,7,8,9,10,15,18,21,30,32,33,35,52,76,81, while 66.54% appears in quote cards and snapshots 66,75,77. The figure with the larger source base is the better supported, but the material does not explain the discrepancy.

Assessment on the software-semiconductor margin differential. The brief anticipates that semiconductor gross margins sit near 60 to 70 percent and software gross margins near 80 to 90 percent, and that the blended figure would move with the mix shift toward software. The supplied material contains neither segment gross margins nor any mix data, so this differential cannot be evidenced. Its contribution to the consolidated margin, and the degree to which VMware has lifted or diluted it, remain unquantified. The consolidated margins above are therefore the only margin evidence available, and they cannot be attributed to either business.

Data unavailable: Segment gross and operating margins; VMware integration cost savings; EBITDA and its margin.

Cash Flow and Earnings Quality

Free cash flow of approximately $26.9 billion, defined as operating cash flow minus capital expenditures 28,80,82, appears in the material both as a projection and as a realised figure. The reader should treat it with care until the basis is clarified. Stock-based compensation equals roughly 27.5% of operating cash flow 52,55,79,82, which means that more than a quarter of operating cash generation is absorbed by equity pay before any shareholder return.

Using the figures as reported, free cash flow equalled about 1.16 times net income and about 42% of revenue (see the Appendix). Such conversion would be strong, but its reliability depends on the unresolved basis of the free cash flow figure, and the operating cash flow and capital expenditure inputs are not separately reported. The sustainability of the cash generation given the dividend commitment is considered in Section 6.

Debt and Liquidity

Total debt was nearly $67.12 billion as of the November 2025 balance sheet 55,83. The debt-to-equity ratio of approximately 0.8x 28,52,54,55,57,79,82 and the current ratio of approximately 1.7x 28,55,79 are reported for the same date. Cash balances are not reported, so net debt cannot be computed, and the leverage trajectory after the VMware acquisition cannot be measured against the material's own evidence. Commentators warn that high debt could constrain financial flexibility in a downturn 52.

Data unavailable: Cash and equivalents; net debt; debt maturity schedule; interest coverage; credit ratings; borrowing costs 65; pro-forma leverage post-VMware.

Summary Table: Reported Financial Metrics

Metric Period Value Source markers
Total revenue FY2025 ~$63.9 billion (up ~24%) 28,30,54,55,79,80,82,98, 53,54,57,79,82,98, 52,80
AI-chip revenue FY2025 $20 billion (up 65%; 31% of total) 24,26,27,87, 31,42,99
Net income FY2025 $23.1 billion 30,52,54,55,57,79,80,82
Net margin FY2025 ~36.2% 28,80
Operating margin FY2025 48.6% 79
Gross margin FY2025 ~64.96% (alternative 66.54%, unreconciled) 6,7,8,9,10,15,18,21,30,32,33,35,52,76,81; 66,75,77
Total revenue Q3 FY2026 $29.6 billion (up 86% from $16.0 billion) 45,68,75,78,83
AI-chip revenue Q3 FY2026 $16.7 billion (single-source caveat) 73; 95
Free cash flow FY2025 (basis unclear) ~$26.9 billion 28,80,82
Stock-based compensation Relative to OCF ~27.5% of operating cash flow 52,55,79,82
Total debt Nov 2025 ~$67.12 billion 55,83
Debt-to-equity Nov 2025 ~0.8x 28,52,54,55,57,79,82
Current ratio Nov 2025 ~1.7x 28,55,79
Operating cash flow FY2025 / Q3 FY2026 / TTM Data unavailable n/a
EBITDA FY2025 / Q3 FY2026 / TTM Data unavailable n/a
Net debt Nov 2025 / TTM Data unavailable n/a

The material contains no TTM figures for revenue, net income or cash flow, so the table reports the fiscal-year and quarterly values as stated.

3. Earnings, Guidance and Momentum

Management Outlook

Management's guidance points to further expansion, although the figures are of differing reliability. For the fourth quarter of fiscal 2026, Broadcom projected revenue of approximately $34.8 billion, up 93% 75, and an AI-related revenue outlook of $21.7 billion, up 236% 56. The AI figure carries the same verification caveat as the third-quarter AI data 95. For the full fiscal year, management's AI revenue guidance is cited at $58 billion, an expected increase of 186% 67, and the same reporting describes the AI outlook as anticipated to double for three consecutive years 67.

An aggregator projects fiscal 2026 revenue of $106 billion, up 66%, with profit almost doubling to more than $44 billion 57. This is a third-party projection and not company guidance, and it is included only to show the range of expectations in circulation.

Longer-dated targets are the largest figures in the material. Broadcom expects AI chip revenue of about $115 billion for the fiscal year ending October 2027 43,44,53,67,69, a claim asserted by twelve sources and therefore the most robust forward-looking statement in the material. Hock Tan said AI revenue could reach $115 billion in fiscal 2027 68, and the same target is repeated across several independent sources 72,93. Reporting frames the 2027 and 2028 goals as two successive doublings 93. Tan has also said Broadcom's AI revenue in 2028 could reach four times the company's current annual revenue 39,66,68. One post cites a fiscal 2028 target of about $230 billion 63, a single-source figure to be treated with caution. Supply commitments are reported to be expected to double AI revenue again the following year, though timing and terms are not specified 52. Cumulative AI-chip revenue since late 2023 is put at roughly $68 billion 73, which indicates how quickly the base has been built.

Assessment of Momentum

Taken together, the evidence shows a business whose growth has been accelerating rather than decelerating. Third-quarter growth of 86% 45,68,75,78,83 against 48% in the prior quarter 68 is a marked change in direction, and AI sales account for it 68. In the quarter reported, the $16.7 billion AI-chip figure 73 equals roughly 56% of the $29.6 billion total 45,68,75,78,83, a derived share that is informative only if the AI figure is confirmed. The forward targets are management's and are therefore claims about future performance, not realised results. Their credibility depends on the supply commitments and customer financing discussed in Sections 6 and 7, which are the mechanisms by which such targets would be delivered.

Two items in the guidance are not addressed in the material. The first is the semiconductor inventory cycle, on which the material offers no outlook. The second is software subscription transition progress and VMware renewal or cross-sell performance, on which the material offers no metrics. These are data gaps rather than neutral findings, and they bear directly on whether revenue growth is durable across both business lines.

Data unavailable: Consensus estimates for the last two to four quarters; actual-versus-consensus comparisons; one-time items; VMware renewal rates and expansion metrics; semiconductor inventory data.

4. Ratios and Peer Benchmarking

Valuation Multiples

Valuation is the most contested element of the material, and the disagreement is largely about the earnings basis rather than about the company's condition. Forward earnings multiples range from roughly 14 to 20 times on the more bullish reports, including 20.38x 49,54, under 19 times fiscal 2027 estimates 76, about 20 times 72, and 17 times 58. Other sources cite a forward P/E of 29.7x with a price-to-sales ratio of 18.5x 54,80, a forward P/E of 29.9 55, and a trailing P/E of 41.97 52,55, with some places reporting the trailing figure as 48 81. These measures are not on the same basis and should not be averaged. The gap between about 20 times and about 30 times forward earnings determines whether the shares appear cheap or expensive, and it is the first question an investor must resolve.

Market capitalisation is also inconsistently stated: $1.57 trillion 75,92, $1.6 trillion 55,82, $1.71 trillion 68 and $1.8 trillion 6,7,8,9,10,18,21,23,29,32,33,48,49,52,76,81. The last figure is asserted by twenty-one sources, but its latest report date is October 14, which makes it the least current.

The share price has fallen while the reported fundamentals have strengthened. The most frequently repeated quote is $328.75, down $15.58 or 4.53%, at the close on October 16, 2026 68,75,76,77. Commentary places the stock 26% below a June peak 68,71. The 52-week range is itself disputed: $293.41 to $481.57 41,84,88,89,90,91, $289.96 to $495.00 53,72, and an older snapshot of $153.09 to $414.61 6,7,8,9,18,21,24,32,33,34,40,72,76,85,86. The last is clearly stale, so the more recent ranges carry more weight. The material does not establish a cause for the October price weakness.

A cross-check on the trailing multiple is instructive. Dividing the $1.8 trillion market capitalisation by fiscal 2025 net income of $23.1 billion gives roughly 78 times. The trailing P/E figures of 41.97 and 48 imply trailing earnings of roughly $43 billion and $38 billion respectively. The difference is plausibly explained by the fiscal 2026 earnings growth that a TTM basis would include, but the material does not provide TTM net income, so this explanation is an inference that requires verification.

Peer Comparison

Peer evidence is partial, and the brief's benchmark set can be only partly addressed. For the semiconductor group, the material reports that Broadcom trades at a premium to Nvidia on future earnings 56, and it describes Broadcom as having higher profitability than peers, which is presented as justifying that premium 57. Nvidia's forward multiple is reported inconsistently: 14.5 times fiscal 2028 analyst estimates 76, 23.8x 80, and a next-twelve-month price-to-forward-earnings of 18.7x against a five-year mean of 36.3x 99. A Bluesky post attributing to a Bank of America analyst a view that Nvidia, Broadcom and Micron trade at about 13 times forward earnings 59 is secondhand and sector-level, and it is not a Broadcom-specific benchmark.

Marvell comparisons depend on the measure chosen. Broadcom's AI revenue is described as growing faster than Marvell's 52, another item expects Marvell to grow slightly faster than Broadcom 77, and Marvell is reported to have faster overall fiscal-year revenue growth 52. Intel and AMD are not addressed in the material at all.

For the software and infrastructure group, the material contains no VMware standalone history, no Red Hat data and no enterprise-software comparables. Over a longer window, Broadcom's 10-year return of 2,580% against 327% for the S&P 500 53 is striking, but it is a historical outcome and not a valuation benchmark.

Data unavailable: Sector-median ratios; Intel and AMD multiples; VMware standalone historical financials; Red Hat and enterprise-software comparables; Broadcom and peer EV/EBITDA; peer ROE and ROIC.

Ratio Summary

Ratio Broadcom Peer reference Status
Forward P/E ~17x to ~30x (basis disputed) 54,58 Nvidia 14.5x to 23.8x, with NTM 18.7x 76,80,99 Partly comparable; bases differ
Trailing P/E 41.97 to 48 52,55,81 Data unavailable Unreconciled
EV/EBITDA Data unavailable Data unavailable Not computable
ROE Data unavailable Data unavailable Not computable
ROIC Data unavailable Data unavailable Not computable
Debt-to-equity ~0.8x 28,52,54,55,57,79,82 Data unavailable Peer basis missing
Net debt / EBITDA Data unavailable Data unavailable Not computable
Current ratio ~1.7x 28,55,79 Data unavailable Peer basis missing
Interest coverage Data unavailable Data unavailable Not computable

Assessment on valuation premium. The premium to Nvidia on forward earnings 56 is asserted rather than demonstrated, and its justification rests on higher profitability 57 for which the material supplies only the 36.2% net margin 28,80 and the 48.6% operating margin 79, without the peer figures needed to compare them. Against the software group, no premium or discount can be established at all. The valuation conclusion must therefore remain conditional on the earnings basis.

5. Management, Governance and Leadership

The material is dominated by the commentary of chief executive Hock Tan, and it reports no changes to the management team or the board. The brief asks for an assessment of board independence, compensation alignment and dual-class structure, and none of these is addressed in the supplied material. They are recorded as data gaps.

Tan's public positions are numerous. He has argued that the imbalance between demand and supply could persist through 2029 65, and that capital rather than chips is the largest constraint on AI expansion 62,73. He has characterised investment flowing into AI startups as uninformed speculation driven by extreme expectations rather than unit economics 73. His September statements that the fiscal 2027 and 2028 AI-chip projections remain intact 74 are management claims, and the material contains no independent verification of them. Some commenters have accused him of criticising customers or investors despite Broadcom's own exposure to the AI buildout 95, an allegation that the material does not substantiate.

The strategic franchise rests on custom silicon. Broadcom helped develop Tensor Processing Units for Alphabet 12,25,35,72,76, a point corroborated by five sources, and its XPUs are designed with hyperscalers including Alphabet, Meta Platforms, Anthropic and OpenAI 56. It works with Meta on the Meta Training and Inference Accelerator 22,67 and has six custom-AI-chip customers 72. The OpenAI relationship centres on the Jalapeño inference processor 17,19,20,66,67, and Alphabet is identified as Broadcom's largest customer 66. Broadcom's own statement that Alphabet's latest TPUs match or beat Nvidia's new systems 68 should be read as company positioning rather than independent assessment.

Assessment of leadership effectiveness. The strongest evidence for management's effectiveness is outcome-based: the revenue acceleration 45,68,75,78,83, the operating margin 79 and the custom-silicon customer base 72 are consistent with disciplined execution. The sources offer no independent evaluation of management's execution of the VMware integration, and the brief's named prior integrations, CA Technologies and Symantec, are not discussed in the material. The integration record therefore cannot be judged from the supplied evidence.

Governance concerns. Governance coverage is thin, and the concerns that are reported are unresolved. European Union antitrust scrutiny related to VMware remains a potential issue 37,55,57. Broadcom recently settled a dispute with AT&T over software-support options 52,82, although one source still lists an AT&T legal dispute as a live risk 55, so the status of that matter is itself unsettled. An insider-selling claim of $260.0 million in the prior 90 days with no purchases 61 comes from a single post and is uncorroborated, but it merits monitoring. One post reports $72.91 million in puts against $13.60 million in calls 70, though it does not state whether these are premiums or notional values. The material describes no governance improvements.

6. Capital Allocation

Dividends

Dividends are small in relation to the company's scale. Several sources cite a yield of 0.79% 66,72,76,81 and an annual dividend of $2.60 82, while another reports a declared-dividend yield of 0.7% 46,52. One article describes Broadcom as having evolved into a dividend-paying income source 88, which overstates what a yield below one percent can deliver on its own. The payout ratio cannot be computed from the material, because it reports neither dividends paid in aggregate nor a reliable free cash flow basis. Using the reported annual dividend and the $26.9 billion free cash flow figure would be inappropriate until that basis is confirmed. Sustainability cannot be assessed in full, but the dividend appears to absorb a very small fraction of the cash generation reported, subject to the same caveat.

Buybacks

The material reports no share buybacks. It does not state whether an authorisation exists, and no historical repurchase pattern is described. The absence is a finding of the material, not evidence that repurchases have not occurred.

Investments, Customer Financing and M&A

Most of the capital deployment described goes toward customers and infrastructure rather than toward shareholders. Broadcom agreed to lend Anthropic up to $42 billion to support infrastructure that will use its chips 47,50,51,66, a commitment asserted by six sources. One post states that the loan takes the form of convertible notes that can convert into Anthropic equity 94. Coverage also reports a $60 billion financing package 69, while a separate Bluesky post says the company is arranging more than $50 billion to finance chips for OpenAI 64. Those OpenAI figures are described as early-stage talks 69. Other commitments include a $30 billion Apple commitment 52, large Samsung commercial commitments 79, and a memorandum of understanding with Samsung reported at $200 billion 52, which is single-sourced and recent. One post claims that Broadcom backs Anthropic and OpenAI with $92 billion 60, and the material does not reconcile this with the other amounts. A Reddit-thread claim that Broadcom has guaranteed repurchase of leased chips if the two AI customers cannot meet their obligations 97 is uncorroborated.

Acquisitions have expanded the infrastructure software business 53, but integration risk accompanies them 55. The material gives no VMware purchase price, integration spending or pipeline of further deals, so the discipline of the acquisition programme cannot be measured.

Assessment. The central gap is quantification. The material supplies no confirmable exposure amounts for the customer-financing arrangements 96, no repayment terms 96 and no loss estimates 96, and it provides no borrowing-cost data for Broadcom 65. The customer-financing commitments are, by the material's own account, the largest capital claims on the balance sheet, and they are the least quantified. The capital allocation case therefore cannot be weighed against the total debt figure of $67.12 billion 55,83 until those terms are established. The sceptical reading is that the company is extending credit to its largest customers in order to secure demand for its chips, which would convert a future revenue claim into a present balance-sheet exposure. The material does not confirm this reading, but it does not exclude it.

Data unavailable: Capital expenditure and R&D intensity by segment; software maintenance spending; share repurchase authorisation and history; dividends paid in aggregate; borrowing costs; VMware integration spending.

7. Risks and Catalysts

Principal Risks

The first risk is customer and financing concentration. The AI growth figures depend on a small number of large customers, with Alphabet identified as the largest 66 and six custom-AI-chip customers 72. The customer-financing commitments discussed in Section 6 would transfer part of the risk of those customers' performance to Broadcom's balance sheet 47,50,51,52,66. Probability is moderate and magnitude is high, because the financing terms are unquantified and the debt load is substantial 55,83.

The second risk is semiconductor cyclicality. The material contains no inventory data, so the cycle cannot be modelled. Its magnitude is potentially large, given how heavily growth is concentrated in one product category, with AI chips providing 31% of fiscal 2025 revenue 24,26,27,87 and a larger share of the latest quarter. Probability cannot be fixed from the evidence, but the concentration makes the magnitude substantial.

The third risk is integration and leverage. The VMware acquisition introduces integration risk 55 and regulatory scrutiny 37,55,57, and the leverage effect is unquantified. Commentators' warning that high debt could constrain flexibility in a downturn 52 is the most direct statement of this risk. Its probability is uncertain. Its magnitude would rise sharply if the forward growth targets were missed, because the debt would then be serviced from a smaller cash base, and the material gives no interest coverage figure to test that.

Competitive pressure in networking and custom silicon is a further risk that the material touches but does not measure. The comparison with Marvell's growth rate is contested 52,77, and no share data is given.

Near-Term Catalysts

The first catalyst is AI networking and custom-chip design wins. The custom-silicon franchise 12,25,35,56,72,76 and the six-customer base 72 support the forward AI targets of $115 billion in fiscal 2027 43,44,53,67,69. Confirmation of new design wins would strengthen the most robust forward-looking claim in the material.

The second catalyst is VMware cross-sell and renewal success. The material contains no renewal rates or expansion metrics, so this catalyst is identifiable but unmeasured. Its evaluation depends on the data gaps recorded in Sections 2 and 3.

The third catalyst is the semiconductor inventory recovery, together with margin expansion from any software mix shift. The material offers no inventory outlook and no segment margin data, so neither can be confirmed. Both would become testable once segment disclosure is available.

8. Investment Implications

Does Valuation Reflect Fundamentals?

The fundamentals support a company with strong and accelerating revenue, solid reported profitability and a large, corroborated AI growth outlook. The valuation question is whether the share price, around $328.75 68,75,76,77, reflects that record. The answer depends on the earnings basis. On forward earnings in the range of about 17 to 20 times, the shares look inexpensive relative to a company growing at the reported pace. On forward earnings near 30 times, the shares look priced for a considerable portion of the forward AI targets. The material supports neither conclusion alone.

The disinterested assessment is that the price weakness of October 2026, which the material does not explain, sits against a fundamental record that has strengthened. The divergence is the principal reason for caution. It may reflect concern about financing exposure, the balance sheet or the reliability of the forward targets, but the material does not establish which. Until the customer-financing terms and the earnings basis are resolved, the valuation cannot be declared fair or unfair from this evidence. The probability of the tendency toward a re-rating therefore rests on questions that the material leaves open.

Critical Follow-Up Questions

  1. What are the terms, exposure and loss provisions of the Anthropic and OpenAI financing? The commitments range from $42 billion 47,50,51,66 to $92 billion 60 in the material, and the repayment and loss terms are absent 96.
  2. What is the segment performance of Semiconductor Solutions and Infrastructure Software, including VMware revenue and renewal rates? Without this, neither the software margin differential nor the VMware synergy case can be tested.
  3. Is the AI-chip quarterly figure confirmed by company filings? The $16.7 billion and 221% figures rest on unverified commentary 95, and the 106% alternative 1,2,3,4,5,6,8,9,11,13,14,16,21,26,27,36,38,56,68 remains unreconciled.
  4. What earnings basis underlies the forward and trailing multiples, and what are the net debt and cash balances after the VMware acquisition? These determine whether the valuation and leverage conclusions above hold.

Appendix: Calculations, Inputs and Source Details

Derived Calculations

Net margin check: net income of $23.1 billion divided by revenue of $63.9 billion gives 36.2% 28,30,52,54,55,57,79,80,82,98, consistent with the reported 36.2% net margin 28,80.

Free cash flow to net income: $26.9 billion divided by $23.1 billion gives approximately 1.16x 28,30,52,54,55,57,79,80,82. This figure is provisional because the free cash flow basis is unresolved.

Free cash flow margin: $26.9 billion divided by $63.9 billion gives approximately 42.1% 28,30,54,55,79,80,82,98. Provisional, for the same reason.

Quarterly revenue growth check: $29.6 billion divided by $16.0 billion gives 1.85x, or 85%, against the reported 86% 45,68,75,78,83. The difference reflects rounding in the reported inputs.

AI share of quarterly revenue: $16.7 billion divided by $29.6 billion gives approximately 56% 45,68,73,75,78,83. Provisional, pending confirmation of the AI figure 95.

Market capitalisation to fiscal 2025 net income: $1.8 trillion 6,7,8,9,10,18,21,23,29,32,33,48,49,52,76,81 divided by $23.1 billion 30,52,54,55,57,79,80,82 gives approximately 78 times. Not comparable with trailing P/E figures without TTM earnings.

Implied trailing earnings from trailing P/E: $1.8 trillion divided by 41.97 gives approximately $42.9 billion 6,7,8,9,10,18,21,23,29,32,33,48,49,52,55,76,81; divided by 48 gives approximately $37.5 billion 6,7,8,9,10,18,21,23,29,32,33,48,49,52,76,81. These are inferences requiring verification.

Cumulative AI-chip base: roughly $68 billion since late 2023 73, against FY2025 AI revenue of $20 billion 24,26,27,87 and Q3 FY2026 AI revenue of $16.7 billion 73.

Ratios not computed for lack of inputs: EV/EBITDA, ROE, ROIC, net debt to EBITDA, interest coverage, payout ratio, and segment margins.

Source Details by Area

Data Gaps Register

More from KAPUALabs

See all
| Free

Macroeconomic and Global Factors

By KAPUALabs
/
| Free

Market Sentiment and Analyst Coverage

By KAPUALabs
/
| Free

Industry and Sector Analysis

By KAPUALabs
/
| Free

Business Operations and Strategy

By KAPUALabs
/