Analyst targets (selected, recent changes):

• Evercore ISI: Buy – $582
• Citigroup: Strong Buy – $500
• JPMorgan: Overweight – $500
• Morgan Stanley: Buy – $470

Broadcom reports Wednesday, June 3, 2026 after the close (5:00pm ET).

That’s the whole week, in one calendar line.

Here’s the thing: Broadcom has been one of the rare large-cap AI winners in 2026 that hasn’t needed a new story every month to stay interesting. It’s mostly been execution, plus the market slowly realizing the company sits in two places that matter at the same time: custom compute and the networking that makes AI clusters useful.

What matters is whether management can keep making the same argument feel more real: AI chip revenue above $100 billion in 2027. That’s not a casual comment. If you underwrite it, you can justify a very different long-term earnings picture. If you don’t, then you start asking uncomfortable questions about timing, customer budgets, and whether the “custom” opportunity is lumpy by nature.

Slight tangent, but it matters: everyone knows GPUs are scarce and expensive. The quieter shift is that hyperscalers want control – cost, power, workload fit, supply planning. Custom silicon is one way to get that control. It doesn’t replace GPUs overnight, but it does change who has leverage over the next three years.

Broadcom’s last quarter set a high bar. The hard part isn’t posting a strong quarter. The hard part is doing it again when expectations are already leaning in.

Anchors going into the call:

• Q1 FY2026 revenue: $19.3B (about +29% year over year)
• Q1 AI semiconductor revenue: $8.4B (about +106% year over year)
• Q2 FY2026 revenue guidance: about $22.0B (about +47% year over year)

If those are the anchors, the real tells are in the color around them: do we hear stronger conviction on the custom accelerator ramp, or more careful language? Does AI networking keep riding alongside the compute wins, or does it flatten in a way that hints customers are pausing builds?

One more point that gets skipped: concentration. Custom programs are sticky, yes. They’re also, by definition, tied to a small set of very large customers. If two of those customers push spend to the right by even a quarter, the market won’t be patient about it – not with AVGO already seen as a leader this year.

So what am I watching Wednesday night?

• Any reaffirmation (or softening) of the 2027 AI revenue line of sight
• Signals on custom accelerator program cadence: timing, volume, and how many ramps overlap
• AI networking demand: not just growth, but whether it’s still broad-based across customers
• Software: steady is fine, but any surprise volatility changes how investors think about the “ballast” in the model

The part people skip is that this earnings call is less about whether Broadcom is “good at AI” and more about whether the company is becoming a dependable supplier of infrastructure for AI. Dependable gets paid. Flashy gets questioned.

If Broadcom sounds calm, specific, and a little repetitive about demand and capacity, that’s probably a positive. If the answers start getting abstract, the market will do what it always does: it will assume the hard part is now the schedule, not the technology.

Worth a look Wednesday night. Not because the quarter matters by itself, but because this is where the 2027 claim either gets sharper… or it gets harder to defend.