Broadcom designs the custom AI chips Google, Meta, Anthropic and OpenAI use instead of buying Nvidia’s, plus the networking chips that connect them, and owns VMware. AI revenue is growing more than 200% a year. The shares have fallen about a third from their high on worries about how the customers will pay for it all.
Today’s close against the report’s numbers, updated every weekday. The analysis is as of October 22, 2026; the rating as of October 22, 2026. Both change only when the report is revised.
The cash-flow model’s bear-to-bull range, the probability-weighted value, the price, its 52-week range and the analysts’ average target. The copper line marks the base case; the fine dotted line is today’s close.
Written on October 22, 2026, ahead of its fourth-quarter results on December 9. Figures are from the company’s filings and releases unless noted. Not investment advice. The author holds no AVGO shares directly; Broadcom is held inside the VanEck Semiconductor ETF (SMH) in the author’s portfolio. Anthropic, which makes Claude (the AI model that helped draft this report), is expected to be Broadcom’s largest custom-chip customer in 2027.
Broadcom has two segments. Semiconductor Solutions (about 70% of revenue) designs custom AI accelerators, which it calls XPUs, for companies that want an alternative to Nvidia: Google’s TPUs, Meta’s MTIA, and chips for Anthropic and OpenAI. It also makes the Ethernet switches (Tomahawk) and optical parts that connect tens of thousands of chips in a data center, plus older chips for phones, broadband and storage. Infrastructure Software is mainly VMware, bought in 2023, plus mainframe and security software.
Broadcom designs the chips; TSMC manufactures them. The business needs little capital of its own, which is why about 46% of revenue becomes free cash flow.
In the quarter to August 2, revenue rose 86% to $29.6 billion and AI semiconductor revenue 221% to $16.7 billion; software grew 29% to $8.8 billion. Broadcom guided fourth-quarter revenue to about $34.8 billion, with AI at $21.7 billion.
Management said it has secured supply for about $115 billion of AI revenue in fiscal 2027 and about $230 billion in 2028, across six custom-chip customers, and expects earnings above $30 a share in 2028. It named Anthropic as likely to be its largest custom-chip customer in 2027 and OpenAI the second. The gross margin is falling, to about 73% from 78%, because complete systems carry more memory bought from others.
| FY2021 | FY2022 | FY2023 | FY2024 | FY2025 | TTM Aug 2026 | |
|---|---|---|---|---|---|---|
| Revenue | $27.4B | $33.2B | $35.8B | $51.6B | $63.9B | $89.1B |
| Operating margin | 31.0% | 42.8% | 45.2% | 26.1% | 39.9% | 48.0% |
| Net income | $6.7B | $11.5B | $14.1B | $5.9B | $23.1B | $38.3B |
| Operating cash flow | $13.8B | $16.7B | $18.1B | $20.0B | $27.5B | $40.7B |
| Capital spending | $0.4B | $0.4B | $0.5B | $0.5B | $0.6B | $1.2B |
| Free cash flow | $13.3B | $16.3B | $17.6B | $19.4B | $26.9B | $39.4B |
The operating margin fell in fiscal 2024 because of the amortization of VMware’s intangible assets, an accounting cost that is not a cash cost. Fiscal years end in early November; the last column covers the four quarters to August 2026. The quarterly view shows each quarter.
At August 2 Broadcom had $24.0 billion of cash and $59.4 billion of debt, $35.4 billion of net debt left from the VMware purchase; it repaid $5.6 billion in the quarter and $1.5 billion more after it. It pays a $0.65 quarterly dividend.
A newer item: with Apollo and Blackstone, Broadcom set up a financing vehicle to help Anthropic and OpenAI pay for their chips. The first $35 billion closed in June for Anthropic’s first gigawatt. Broadcom says it may give “modest” guarantees on the hardware’s resale value but has not said how large they could become. In October chief executive Hock Tan said financing access and costs might become key limits on AI growth.
| Assumption | Base case | Basis |
|---|---|---|
| Revenue, fiscal 2026 | $106B | First three quarters plus Q4 guidance (estimate) |
| Growth, 2027 and 2028 | 55%, then 25% | Below management’s outlook for 2028; fades to 3% by year 10 |
| Operating margin | 54% easing to 50% | After stock pay, before VMware amortization; systems carry more memory |
| Reinvestment | $1 for every $4 of new revenue | Working capital and customer financing |
| Discount rate | 10.7% | Beta 1.30 for large chip companies (Broadcom’s own is 1.86), little debt |
| Terminal growth | 3.0% | With a 25% return on new capital |
| Bear | Base | Bull | |
|---|---|---|---|
| Growth, 2027 / 2028 | 30% / 0% | 55% / 25% | 60% / 45% |
| Long-run operating margin | 45% | 50% | 53% |
| Discount rate | 11.7% | 10.7% | 9.7% |
| Value per share | $102 | $358 | $867 |
| Probability | 25% | 50% | 25% |
Weighted 25/50/25: $421. The range is unusually wide. The bear case is AI customers unable to fund their build-outs after 2027, leaving Broadcom with a year of growth and then none; the bull case is management’s 2028 outlook delivered in full. The base case takes 2027 roughly as guided and haircuts 2028.
| Discount rate | 9.0% | 9.5% | 10.0% | 10.5% | 11.0% |
|---|---|---|---|---|---|
| Value per share | $473 | $434 | $400 | $371 | $345 |
Every figure in this valuation comes from one model script, rerun and checked against this page each time the site is built. The XLSX model rebuilds it in live formulas: change any blue input and every year’s cash flow, the bridge to the value, the scenarios and the sensitivity table recalculate.
| To justify the price, you need | Against |
|---|---|
| Revenue growth of about 28% a year in 2027 and 2028, then the same fade | 55% and 25% |
| Or a long-run operating margin of about 45% | 50% |
The price assumes about half the growth management says it has already secured supply for. Analysts average about $529. The shares fell 15% in one day in June when guidance was not raised, and again in October on worries that the AI labs cannot finance their plans; the model’s wide bear case is the same worry, quantified.
| Risk | Why it matters |
|---|---|
| Customer funding | Anthropic and OpenAI need outside money to buy their chips; Broadcom is now helping finance them. |
| Concentration | A handful of customers make up most AI revenue; losing one design generation would show quickly. |
| Competition | Nvidia, and design houses in Taiwan, compete for the same custom-chip work. |
| Supply | Leading-edge wafers, substrates, HBM memory and lasers are all tight; Broadcom depends on TSMC. |
| Margins | Complete systems carry more bought-in memory, pulling the gross margin down. |
Upside: the 2028 outlook delivered in full, more custom-chip customers, Ethernet winning inside AI racks, and VMware growing faster than expected.
Broadcom has secured supply for about $115 billion of AI revenue next year and says demand is higher. The price assumes about half the growth in its outlook, because two of its largest customers, Anthropic and OpenAI, need outside financing that Broadcom itself is now helping arrange.
What would change the value: Fourth-quarter results on December 9, the size of the guarantees in the financing vehicle and any change to the 2027 AI outlook are the next inputs.
Ranges, not three cases. The expected value is the average of 4,000 runs of this report’s own model, with every input drawn from a range: the base case is the middle, and the bear and bull cases are the 10th and 90th percentiles. Business inputs move together; the discount rate moves on its own. The 25/50/25 weighting in the valuation section gives $421. How the ranges work
Two different questions. The rating is relative: it asks whether the stock is cheaper or dearer than the S&P 500 on the same assumptions. The absolute view asks whether it is worth its price at all: on its own the stock trades 23% below our estimated value.
On the same assumptions every report uses (a 5.15% Treasury yield and a 4.5% equity premium), the S&P 500 itself is worth about 68% of its price. The rating compares each stock with that: Buy if it is at least 15% cheaper than the market on this basis, Sell if at least 15% dearer, Hold in between. It is the same rule for every company, whoever owns it. The rating is the same whether the index is assumed to grow 6% or 10% a year. How ratings work
Against the Street: Analysts average about $529, closer to management’s outlook. This model takes 2027 roughly as guided but haircuts 2028 by about a quarter and uses a discount rate of 10.7%.
| Measure | Broadcom | Industry median | Rank | |
|---|---|---|---|---|
| Revenue growth, last fiscal year | 23.9% | 15.3% | 7 of 15 | Better than the median |
| Operating margin | 48.0% | 26.0% | 3 of 15 | Better than the median |
| Price to earnings | 43.0× | 46.4× | 7 of 14 | Better than the median |
| Free cash flow yield | 2.4% | 2.5% | 10 of 15 | Worse than the median |
Peers: the 14 other S&P 500 companies in Semiconductors with complete data (ADI, AMD, FSLR, INTC, MCHP, MPWR, MRVL, MU, NVDA, NXPI, ON, QCOM, SWKS, TXN). Last twelve months where reported, from SEC filings; prices from the latest close. A lower price-to-earnings ratio ranks higher. These figures update with each data run; the rating does not.
The orders are there; the question is whether the customers can pay.
| What | Figures | Source |
|---|---|---|
| Latest results | $29.6B (Revenue, Q3 FY2026, +86%), $16.7B (AI semiconductor revenue, +221%), $13.7B (Free cash flow), $34.8B (Q4 revenue guidance) and others | Q3 FY2026 results; Earnings call |
| The numbers (five-year table) | Five years of annual figures and the last twelve months | SEC XBRL company facts; latest: 10-K for the period to 2025-11-02, 10-Q for the period to 2026-08-02 |
| Price at writing | $323.24 (Close, Oct 22, 2026) | Yahoo Finance; the time of the quote is stated beside the price |
| Discount rate inputs | 10-year Treasury 5.15%, equity risk premium 4.5% | Fixed for every report so they can be compared; How ratings work |
| Valuation | Every value, case and sensitivity | This site’s model: XLSX, CSV |
| Checked by hand | 12 figures on October 23, 2026 | All match. Audit log |
| Input | Bear (10th percentile) | Base (median) | Bull (90th percentile) |
|---|---|---|---|
| Growth, first forecast year | 30.0% | 55.0% | 60.0% |
| Growth, second forecast year | 0.0% | 25.0% | 45.0% |
| Operating margin, near term | 48.0% | 55.0% | 57.0% |
| Operating margin, long run | 45.0% | 50.0% | 53.0% |
| Discount rate | 11.7% | 10.7% | 9.7% |
| Simulation result | |
|---|---|
| Expected value (average of 4,000 runs) | $397.76 |
| Likely range, 10th to 90th percentile | $129.54 to $726.10 |
| Median | $348.18 |
| Chance the value is above the price | 54% |
| Chance it is cheaper than the S&P 500 | 73% |
Each input is drawn from a split normal: its bear and bull values are the 10th and 90th percentiles, so the two sides can differ. Business inputs share a common factor (correlation 0.6); the discount rate moves on its own. Inputs not listed are held at their base value. How the simulation works.