Broadcom · AVGO ·
Broadcom: when AI ambitions become cash for shareholders
What if Broadcom delivers on management’s growth ambitions and custom AI chips become enduring, profitable infrastructure? A conditional model through October 17, 2029 examines not just revenue, but customer financing, dilution and the price the market might pay for cash flow.
English translation of the original Czech edition.
Documented management ambitions
Sources
- [S01] Broadcom Q3 FY2026 financial results and dividend — company releaseBroadcom Inc. / PR Newswire · 2 September 2026 · Checked 17/10/2026, 08:05
- [S02] Broadcom Form 10-Q, quarter ended August 2, 2026Broadcom Inc. / SEC EDGAR · Checked 17/10/2026, 08:05
- [S03] Broadcom Q3 FY2026 earnings call — management remarks, transcript hosted by Stock AnalysisBroadcom management / Stock Analysis (transcript host) · 2 September 2026 · Checked 17/10/2026, 08:05
- [S04] AVGO historical regular-session pricesStock Analysis / S&P Global Market Intelligence · 16 October 2026 · Checked 17/10/2026, 08:05
Management ambitions
Broadcom is an interesting choice for the series’ first edition because management has set out a concrete growth ambition. On its September earnings call, management outlined AI semiconductor revenue of approximately USD 115 billion in fiscal 2027 and USD 230 billion in 2028. This is management’s outlook, not reported orders or assured receipts. Management’s earnings-call transcript, September 2, 2026.
The company’s original filing also describes the AI XPV platform’s ambition to enable more than 20 GW of compute capacity through 2028. Gigawatts are not Broadcom revenue, however. Form 10-Q.
The reported starting point is considerably smaller: Q3 FY2026 revenue was USD 29.591 billion and free cash flow USD 13.665 billion; revenue guidance for the following quarter was USD 34.8 billion. Earnings release.
ChatGPT wrote this article and built the model. It examines successful delivery of those ambitions and its own conditional extension of growth into 2029. It is neither the base case nor an expected outcome, has no assigned probability and is not investment advice. This article does not change the standard analysis or its scores.
What must go right
The commercial mechanism for success is straightforward: major customers would keep using custom accelerators and networking because the complete solution offers attractive compute economics. Broadcom would need to retain an important role across subsequent chip generations. Expanding AI budgets alone are insufficient if a competitor or the customer itself offers better pricing or technology.
Deliveries, data-center commissioning, energy availability and, above all, customer funding must work together. It is a chain: an order can be genuine, yet delayed construction or a capital shortage can postpone both delivery and collection. The optimistic model therefore does not equate management ambition with money that already belongs to shareholders.
A contingent guarantee is not automatically a loss. The SEC filing reports maximum potential exposure of approximately USD 29 billion under the existing backstop after full deployment of the relevant racks; nothing had been paid under it at that point. This is a defined risk of a particular arrangement, not cash debt equal to that entire amount. Future agreements could nevertheless change the economics of growth. Backstop terms in the 10-Q.
Economics in three years
The horizon is exactly October 17, 2029. To value the shares on that date, I use target annual economics corresponding approximately to FY2029; that fiscal year will not yet have closed. All the following figures are ChatGPT assumptions, not additional company guidance.
The model assumes AI revenue of USD 280 billion and other revenue of USD 60 billion, for a total of USD 340 billion a year. The assumed continuation of AI growth beyond 2028 is approximately 22%, rather than another doubling. The remainder would grow from a more mature base. That distinction prevents optimism from simply extending the fastest growth rate forever.
I assume a cash-flow margin of 40% after tax, interest on existing financing, normal working capital and capital expenditure, but before additional cash absorption from customer financing. From the resulting USD 136 billion, I deduct a separate modeled annual burden of USD 4 billion for that support and associated risks. This leaves USD 132 billion of adjusted FCF. The burden equals approximately 1.2% of revenue: it allows an ongoing cost of support, but assumes successful collection and no widespread customer defaults. This allowance is neither an accounting estimate disclosed by the company nor a guaranteed ceiling on possible losses.
I assume 5.1 billion diluted shares. For comparison, management expected approximately 4.94 billion for Q4. Earnings call. The model therefore allows further net dilution and assumes no repurchase that eliminates it. Stock-based compensation is not deducted again as cash: its economic effect is captured by the larger denominator.
The bridge is 340 × 40% − 4 = USD 132 billion; divided by 5.1 billion shares, this produces USD 25.88 of adjusted annual FCF per share. This is not GAAP earnings. Refinancing debt on manageable terms is a condition; I add no further benefit from increased borrowing.
How the market might value the business
Valuation uses free cash flow to equity after interest, not EBITDA. I apply three independently chosen valuation multiples: 20×, 25× and 30× target adjusted FCF per share. These correspond to cash-flow yields of 5%, 4% and approximately 3.3%. They are neither a sourced peer-group median nor a claim that the market must assign any of them.
The lower multiple allows for persistent customer concentration and investment cyclicality. The middle multiple requires confidence that cash generation is repeatable. The upper multiple would require persuasive evidence of longer growth durability and controlled financing exposure. Business success therefore does not automatically imply a higher valuation multiple.
The reference is Broadcom common stock, NASDAQ: AVGO, at the regular-session close on October 16, 2026, at 4:00 p.m. EDT: USD 328.75. After-hours trading is excluded. Dated stock quote. The returns below use that price, in USD, excluding dividend reinvestment, taxes, fees and changes in the Czech koruna exchange rate. The range measures the same favorable business scenario’s sensitivity to a multiple, not a statistical interval.
Calculations use unrounded FCF per share. They add USD 7.80 in total dividends, an independent assumption of an unchanged USD 0.65 each quarter for twelve quarters, anchored to the most recently declared dividend. Earnings release. The additional one-time adjustment is zero; net debt is not deducted again from an equity valuation.
One optimistic business case, three market multiples
The operating assumptions stay the same. The market can value the resulting performance differently. This is a sensitivity analysis, not a probability interval.
Price = per-share metric × multiple + one-time net adjustment. Total return includes the stated dividends; annualized return assumes a three-year period. No dividend reinvestment, taxes, fees or currency conversion.
ChatGPT’s own conditional model for October 17, 2029, with annual economics approximating FY2029: USD 280bn AI revenue + USD 60bn other revenue = USD 340bn; 40% FCF margin after tax, existing interest, normal working capital and capex = USD 136bn; less USD 4bn of additional annual customer-financing and risk burden = USD 132bn adjusted FCF; divided by 5.1bn diluted shares = USD 25.88235294117647 per share. The additional burden is an independent modeled allowance, not a disclosed accounting estimate or ceiling on losses. SBC is reflected through net dilution, not another cash deduction. No benefit from repurchases, increased borrowing or adding accumulated cash; manageable refinancing is a condition. The 20/25/30× multiples are independently chosen sensitivities (cash-flow yields 5/4/3.33%), not peer consensus or probabilities. Dividends are 0.65 × 12 = USD 7.80 without growth; no additional one-time adjustment and no repeated net-debt deduction from equity value. Returns against the USD 328.75 regular-session close are nominal USD, excluding reinvestment, taxes, fees and FX. The USD 4bn burden is approximately 1.2% of revenue and assumes successful collection without widespread defaults; a margin lower by 5 percentage points means −USD 83.33 per share at 25×, and an additional USD 4bn annual burden means −USD 19.61.
What may already be priced in
A known growth ambition is not, by itself, evidence that the market underestimates it. Without a consistent, current consensus dataset, I do not claim this model exceeds analysts’ expectations. It is more useful to reverse the calculation: what cash flow per share does today’s price require at a chosen multiple?
At 25×, today’s price would be matched, including modeled dividends and with no increase in value, by annual revenue of approximately USD 174 billion at the same margin, additional burden and share count. This is an algebraic break-even point, not an estimate of market expectations. Even if all USD 340 billion of modeled revenue materializes, a multiple of approximately 12.4× would merely return today’s value including dividends.
Under successful delivery of the optimistic model, growth in cash flow per share would be the main source of returns. But the multiple is a separate variable: investors might start demanding a higher cash-flow yield precisely when the business plan succeeds. Multiple contraction could consequently consume a substantial portion of the increase in value.
It would be equally dangerous to add all accumulated cash flows to the valuation even though they fund debt, operations or other capital requirements. This model adds neither a pile of future cash nor gains from assumed repurchases. It seeks to value a sustainable annual ability to generate cash rather than the sum of several overlapping narratives.
Sensitivity is not confined to the multiple. At unchanged revenue, a margin lower by five percentage points reduces annual FCF by USD 17 billion and the price at 25× by approximately USD 83 per share. Each additional USD 4 billion of annual customer-financing burden reduces that price by approximately USD 20. These independent mechanical calculations show why capital requirements cannot be ignored.
Fragile assumptions and checkpoints
The most fragile assumption is that infrastructure growth becomes cash flow per share rather than remaining tied up in capital or receivables. For subsequent results, I would monitor five specific checkpoints:
These are neither predictions of failure nor trading instructions. They are evidence for abandoning or recalculating this conditional narrative in time. The article does not claim AVGO is cheap today; it shows how demanding the path from ambition to shareholder value must be.