Broadcom (AVGO)
Semiconductores (silicio de IA a medida) / Software de infraestructura
An exceptional-quality business —co-designed custom AI silicon + infrastructure software at a 77% margin— with a $164.6bn contracted AI backlog. But at $358 (~51× operating income) it is priced for perfection: Overvalued. Not even a generous base case (semis doubling over five years) justifies today's multiple.
- Price
- $358.27
- Intrinsic value (5y, base)
- $270
- Total annual return (5y)
- -4.4%
- Status (nominal)
- Overvalued
- Margin of safety
- No margin
The essentials
- Two quality engines: co-designed custom AI silicon with the hyperscalers (semis, ~63% of revenue) + subscription infrastructure software with ~77% segment margin (VMware, ~37%). Capital-light, ~42% free cash flow margin.
- The contracted artificial intelligence backlog exploded: the RPO jumped from $33.3bn to $164.6bn in one quarter on a custom accelerator contract — signed revenue that anchors growth, not extrapolated.
- But at $358 it is priced for perfection: ~51× operating income (about 45× economic operating income) → Overvalued. The central risk is concentration: a single customer = 42% of revenue in the latest quarter.
Intrinsic value — two valuation methods
Total return at 5 years: -4.4%/year = -5.5% appreciation + 1.1% dividend. The target price ($270) is ex-dividend; the $18 in dividends collected over 5 years are added separately.
By both methods, the value today (DCF $286 · Multiples $232) is below the market price ($358).
Pillars of the analysis
The verdict — today vs 5 years
Today — expensive, no margin of safety: at $358 trades ~54.2% above its value discounted to today (~$232); the expected return does not even reach the risk-free rate (4.5%).
At 5 years — Sobrevalorado: the expected total return is negative — the price already discounts a demanding scenario that, if not met, results in a loss.
The bridge: the return at 5 years falls below the risk-free rate (4.5%) — which is why there is not even a discount to today's value. To require a 15% annual return, it would need to be bought at ~$146.
Thesis
The business
Broadcom is an exceptional-quality business: two complementary engines —co-designed custom AI silicon with the hyperscalers and an infrastructure software annuity with ~77% segment margin— capital-light (capex <1% of revenue), with a free cash flow margin of ~42% and a return on capital well above the 10% bar. The contracted artificial intelligence backlog ($164.6bn of RPO) anchors growth with revenue already signed. The flip side: an extreme customer concentration (one customer = 42% of revenue in the latest quarter) and net debt of ~$45bn inherited from the VMware acquisition, which is being paid down with the giant cash flow.
The valuation
It is valued by sum-of-the-parts: each engine with its metric, on operating income after taxes (EV/NOPAT), adding back the amortization of acquisition intangibles (~$8bn/year, a non-cash charge from the VMware acquisition) but expensing stock-based compensation (~28% of free cash flow, a real cost). Semiconductors at ~20× (quality and AI growth); infrastructure software at ~18× (mature subscription annuity). The base scenario projects ~$270/share over five years.
At $358, the problem is the starting point: Broadcom trades at ~51× operating income (about 45× economic operating income), a multiple that compresses toward ~24× over five years as NOPAT compounds — but not even a generous base case (semiconductors doubling over five years) makes the five-year value exceed today's price. The base return at market price is -4%.
The margin of safety
The verdict is Overvalued: No margin of safety: the price already discounts a demanding scenario.. At $358 the five-year value falls below the current price, with a negative expected return — Broadcom is priced for perfection. Today's multiple (~45× economic operating income) discounts that the artificial intelligence ramp and the software margins hold for years; even the favorable scenario, with the backlog converting in full and semiconductors nearly tripling, barely reaches today's price. There is no margin of safety: it is an exceptional business at a price that already charged for the excellence in advance.
What to watch
The main disconfirmer is customer concentration: a single customer represented 42% of revenue in the second quarter of FY2026 (it was 28% two years earlier), from custom artificial intelligence silicon. If that program is internalized, cancelled or the customer fails to execute, the $164.6bn backlog —today the main justification of the price— suffers. In the background: the sustainability of the AI ramp (it depends on a few programs from one hyperscaler), the cyclicality of the non-AI portion of semiconductors (wireless/broadband, ~$16.7bn) and the ~$45bn debt (amply covered by cash flow, but a drag if rates or the rating move). The bear thesis is not that the business is bad —it is exceptional— but that the price already reflects it in full.
Educational / informational. Does not constitute investment advice.