AVGO · Forward model · Bear case

The Bear case, 20 quarters out

Model as of 2026-09-03

Each segment is projected from its own operating driver, rolled up into consolidated cash flow, and discounted back to a fair value per share. The assumptions below are editable — change them and every number on this page moves with them.

Quarters are labelled by the calendar year the fiscal quarter ENDS in, matching the rest of r40: 2026 Q3 is Broadcom's third fiscal quarter, ended 2 August 2026. Broadcom reports two segments and this model carries three verticals, because the segment that matters splits into two businesses with almost nothing in common. Infrastructure Software ($8,752M in the basis quarter) is reported on the face of the release. AI semiconductor revenue ($16.7B) is management prose - stated in the release headline and again on the call - and is NOT a reportable segment: there is no reconciliation table and no prior-period series in the exhibit. Non-AI semiconductors is the arithmetic remainder of Semiconductor Solutions ($20,839M) less AI, so the three verticals sum to reported consolidated revenue ($29,591M) to the dollar. Nothing is apportioned. On the call Tan gives the non-AI line as approximately $4.2B; the $61M against the $4,139M remainder here is rounding inside the $16.7B AI figure. What the Q3 call changed: management stopped quoting a single fiscal-year AI target and started describing demand in gigawatts, customer by customer, and for the first time gave its own content per gigawatt - '$20 billion-$30 billion per gigawatt'. The capacity driver is built on that disclosure. The 759 MW basis figure is $16.7B at the $25M per MW midpoint and 88% utilisation, and is DERIVED: Broadcom has never disclosed gigawatts shipped. What is assumed rather than reported: the segment margin and capex splits. Broadcom discloses consolidated non-GAAP operating margin (67.9%) and consolidated capex ($532M) but neither by segment, so the three margin inputs are the author's and are calibrated to reproduce the consolidated figure. The arithmetic worth stating plainly. Management gave three AI numbers: $58B for fiscal 2026, approximately $115B for fiscal 2027 'secured', and line of sight to $230B in fiscal 2028. The first two are reproduced here exactly - Q4 AI of $21.7B and fiscal 2027 of $115.0B. The third is not reachable from the same build rate: $115B in fiscal 2027 requires the sequential ramp to DECELERATE hard through that year, and $230B in fiscal 2028 then requires it to RE-ACCELERATE. This model's base decelerates and AI plateaus near $132B a year, about 43% below the fiscal 2028 line of sight. That gap is the finding, not a bug, and it is what the bull case is for. Note that management's own customer roadmap describes exactly such a re-acceleration - Anthropic from 5 GW in 2027 to an incremental 10 GW in 2028, OpenAI from 1.3 GW to over 5 GW - so the disagreement is about whether that step happens, not about whether management said it would.

Where each case comes from

From cash flow to fair value

The projected path

The published model's 20 quarters. The horizon and sliders above do not change this table.

Every row is projected. A year-over-year change is shown only where the quarter it compares against exists — an em dash means there is no comparable quarter, not a flat year.

Model revisions

Assumptions are marked to reality as each quarter prints. Every change is appended here, with the fair value the model produced at the time, so the model's own history stays visible.