In plain English
What the company actually sells
Two halves. The semiconductor half designs custom AI accelerators for a small number of enormous customers who want an alternative to buying merchant chips, plus the networking silicon that connects data centres together. The software half — largely built by acquiring VMware — sells the software that runs corporate data centres, converted from one-off licences into subscriptions.
Why the share price moves
How much custom AI silicon it expects to ship, and for whom. Because a handful of programmes represent so much of the growth, news that a specific customer's programme is expanding or slipping moves the stock immediately. The software half is valued more like an annuity and moves it far less.
If you already own it
Unusual among AI-levered names: there is a real, growing dividend here. That changes the income arithmetic and it introduces ex-dividend assignment risk that a non-payer like Amazon does not have.
The technical read
What actually drives results
Four variables explain most of what this company reports. Everything else is commentary.
Designing bespoke silicon for hyperscalers is a business with few customers and enormous contracts. Growth is programme-driven and therefore lumpy: it arrives in steps as designs move to volume, not as a smooth curve.
As AI clusters grow, the interconnect becomes the bottleneck, and switching silicon becomes strategically important. This is a genuine second AI franchise rather than a rounding error attached to the first.
Moving an acquired installed base from perpetual licences to subscriptions raises revenue per customer sharply and antagonises those customers. It is highly profitable and it creates a durable incentive to migrate away, on a multi-year timetable.
Large acquisitions were debt-funded. Cash generation is directed at paying that down while still raising the dividend, which constrains how much can go to buybacks.
The moat, stated as a mechanism
Co-design lock-in. Once a customer's accelerator is jointly engineered for their workload, switching means redesigning the chip — a multi-year commitment, not a supplier change. On the software side, the moat is simply that the infrastructure layer is load-bearing and migration is dangerous.
A moat you cannot describe as a mechanism is not a moat — it is a track record. The test is whether you can name the specific reason a well-funded competitor has not simply taken the business.
What would break it
The parts an owner should be able to state without prompting. None of these is a prediction; each is a live risk with an identifiable mechanism.
- Extreme customer concentration, arguably the highest in the top ten. A small number of buyers drive the growth narrative, and each has the engineering capability to change direction.
- Programme timing risk. Because revenue arrives in steps, a single delayed ramp can miss a quarter badly without anything being structurally wrong.
- Aggressive software repricing invites long-run substitution; the revenue is booked now and the customer resentment compounds.
- The non-AI semiconductor segments serve cyclical end markets and can be weak while the AI story is strong, muddying the reported numbers.
The line to read first in the next earnings release
Guided AI revenue and any disclosure about the number of custom-silicon customers moving to volume production. That figure, more than earnings per share, is what the stock trades on.
How it behaves in the options market
This is the section no ordinary stock write-up contains, and it is the one that decides whether an income overlay on these shares is sensible. Two companies with identical fundamentals can be completely different option underlyings.
High — comparable to Meta and well above the software mega caps, reflecting both programme lumpiness and semiconductor cyclicality.
Large moves are routine, and they are frequently driven by forward AI revenue commentary rather than the quarter reported. The guide is the event.
Good and much improved since the split, with active weeklies, though strike granularity is not at Apple's level.
A 10-for-1 split in July 2024 reset the share price into a range where covered calls are practical for ordinary accounts; divide any pre-split strike or premium by ten. The dividend is substantial by semiconductor standards, so ex-dividend early assignment on in-the-money calls is a live consideration rather than a theoretical one.
Covered-call character
This grades the option trade, not the company and not the stock. A “Poor fit” means the chain pays too little to justify capping upside — nothing more.
Dividends and buybacks
A large and consistently raised dividend, unusual among AI-exposed semiconductor names, funded by strong cash generation while debt from acquisitions is repaid. The yield is high enough here that a covered-call overlay genuinely stacks on top of a meaningful income base.
The tax wrinkle specific to this name
The one name in the top ten where the qualified-dividend holding period is worth real money, because the dividend is large enough to matter. Writing in-the-money calls against these shares can suspend that holding period and convert dividends taxed at long-term capital-gains rates into dividends taxed as ordinary income. The premium may not cover what that costs — check it in the Calculator with your own rates.
Now check today’s numbers
Everything above is durable. Everything that is not — price, implied volatility, the actual credit on an actual strike — lives in the tools, with AVGO already filled in.
The Calculator is free and needs no account. The Screener reads a live options chain, so it asks you to sign in first — the link will bring you back here.
Read next
What a covered call really costsThe premium is visible; the upside you cap and the tax bill if you’re assigned are not. The lesson behind every verdict on this page.Educational information only — not investment, tax, or legal advice, and not a recommendation to buy, sell, or hold AVGO or any other security. Nest Money is not a registered investment adviser or broker-dealer, and receives no compensation from any company covered here. This analysis is our own summary of publicly available information as at the date shown; it may be incomplete or out of date, and you should verify anything you intend to act on against the company’s own filings. Options involve substantial risk of loss, including loss greater than the premium collected.