AVGO/Non-AI Semiconductor — The Legacy Base

Non-AI Semiconductor — The Legacy Base

$17/share(6% of AVGO)anchored

The key question

Do legacy semiconductor businesses maintain margin as R&D resources shift to AI programs?

$16.9/shNon-AI Semi Contribution5.6% of equity value, 3-scenario DCF model

The non-AI semiconductor business is Broadcom's legacy foundation -- broadband, wireless, server/storage, and industrial chips that generate stable cash flows with limited growth. Revenue has been flat at roughly $16.7B, and the segment's share of semiconductor solutions is declining rapidly as AI explodes (from 45% in FY2024 to 33% in Q1 FY2026).

~$16.7B
FY2025 Non-AI Semi Revenue
Flat YoY, Q1 FY2026: $4.1B flat
~55%
Estimated OPM
Blended semi segment 57.6%, non-AI slightly lower
33%
Share of Semi Segment
Down from 45% in FY2024 as AI grows

The modest bull case rests on DOCSIS 4.0 broadband upgrades and Wi-Fi 7 adoption driving periodic recovery. The bear case is secular decline as some end markets (set-top boxes, DSL) shrink permanently and enterprise storage shifts to cloud-native architectures. Either way, this segment contributes less than 6% of equity value and is not a material driver of the investment thesis.

Mature cash cow, not a growth driver

Non-AI semi is a high-margin, low-growth cash generator. At 55% operating margins and a fabless model, it produces substantial free cash flow on flat revenue. The investment debate for Broadcom is entirely about AI semiconductor and VMware -- this segment is background noise.

Non-AI Semi Business Lines
SegmentRevenueGrowth DriverRisk
Broadband (DSL/DOCSIS/fiber)Largest non-AI segmentDOCSIS 3.1 / fiber build-outTelco capex cycles
WiFi/BluetoothSignificantWiFi 7 (802.11be) upgrade cycleMediaTek competitive pressure
Storage NetworkingModerateSAS/PCIe Gen 5 enterprise HDDNVMe directly threatening SAS
Set-top Box / VideoDecliningIPTV transitionStreaming shifts to software

Non-AI Semi — Scenario Comparison

bull (20%)base (55%)bear (25%)
GrowthWiFi 7 + broadband refresh drives 8-12% CAGRSteady harvest; 3-5% CAGRDeclining — WiFi/broadband refresh delayed
MarginsExpand on product mix shift to higher-ASP siliconStable; BRCM pricing discipline maintainedCompressed; pricing power lost to Qualcomm/MediaTek
R&D AllocationSpin-out creates standalone value unlockMaintained at current levelsStarved by AI program demands
DCF Value$105B$84B$51B
Per Share$22.2$17.8$10.7

Sensitivity: Probability Shifts

Bull Prob.Bear Prob.Implied ValueΔ from Current
30%20%$18.7/sh+$2
30%20%$18.7/sh+$2
10%40%$14.3/sh-$3
15%35%$15.1/sh-$2

So What?

Non-AI semi contributes .9/sh at the base case — a minor contributor to Broadcom's overall value. These businesses are valued as steady cash generators, not growth engines. The main scenario risk is internal: if AI XPU success causes Broadcom to under-invest in legacy product lines, accelerating customer attrition, the bear case ($11/sh) is plausible. The bull ($22/sh) requires a WiFi7 super-cycle or storage networking refresh that exceeds current forecasts. Neither outcome moves the stock materially — this segment's real value is as a cash flow floor while the AI thesis plays out.

Sources

Company Filings
Broadcom 10-K FY2024 · Broadcom Q4 FY2024 Earnings Transcript
Industry Data
Wi-Fi Alliance WiFi 7 certification data · IDC broadband silicon market share
Analyst Research
Barclays AVGO segment decomposition · Morgan Stanley non-AI semi note
The key question

What is the DOCSIS 4.0 upgrade timeline and Broadcom's share of new modem chip designs?

Scenario Model$17/share

Broadband: DOCSIS 4.0 & Cable Infrastructure

5 evidence
25 GbpsDOCSIS 4.0 SpeedUnified chipsets developed with Comcast and Charter

Broadcom's broadband chip business is entering an upgrade cycle driven by DOCSIS 4.0, the next-generation cable standard that enables speeds up to 25 Gbps over existing cable infrastructure. Broadcom holds a near-monopoly on DOCSIS silicon and is jointly developing unified chipsets with Comcast and Charter. The AI-powered NPU embedded in these chips adds intelligence to the network. While this is a cyclical business tied to cable operator capex, the DOCSIS 4.0 cycle provides a multi-year revenue tailwind.

2026
DOCSIS 4.0 Retail Modems
Expected on shelves this year
Near-monopoly
Market Position
Minimal competition from Qualcomm/MaxLinear
U-shaped
Broadband Recovery
Improvement in FY2025 after inventory correction

Cyclical opportunity, not structural growth

DOCSIS 4.0 provides a genuine multi-year upgrade cycle, but broadband remains cyclical. Cable operators invest in waves, and the long-term threat is fiber-to-the-home deployments that may reduce the addressable market for cable modem silicon over time.

Wireless: Wi-Fi 7, Bluetooth & Apple Relationship

5 evidence
N1 chip replaces BroadcomApple ImpactiPhone 17, Apple TV, HomePod (2025); iPad, Mac (2026)

Broadcom's wireless business faces a structural revenue headwind as Apple transitions to its own N1 networking chip. Apple has been Broadcom's largest wireless customer, and the shift to in-house Wi-Fi 7 and Bluetooth in iPhone 17 removes a significant revenue stream. However, this is partially offset by continued RF filter sales and the Baltra cloud server chip collaboration. Broadcom's Wi-Fi 7/8 portfolio targets enterprise and non-Apple mobile markets.

2026
Wi-Fi 7 Adoption
Pivotal year for North American adoption
RF filter only
Apple Retention
Plus Baltra cloud server chip collaboration
Late 2027+
Wi-Fi 8
Products not expected before then

Apple in-sourcing is the structural headwind

Apple's decision to develop its own Wi-Fi and Bluetooth chip is part of a broader in-sourcing strategy that Broadcom cannot reverse. The revenue loss is partially offset by enterprise Wi-Fi growth and the emerging cloud server chip partnership, but the wireless business will be structurally smaller without Apple as a component customer.

Legacy Semi: Storage, Industrial & Secular Trends

5 evidence
33%→decliningNon-AI Semi ShareOf semiconductor solutions, down from 45% in FY2024 as AI grows

Broadcom's legacy semiconductor portfolio -- server/storage connectivity, enterprise networking, and industrial chips -- is a mature, cash-generative business in secular decline relative to the AI semiconductor explosion. Server storage connectivity revenue declined 25% YoY in Q4 FY2025 as enterprise storage shifts to cloud-native architectures. The overall non-AI semi business is flat with management guiding for stability in FY2026, not growth.

~$16.7B
FY2025 Non-AI Semi
Flat YoY, estimated 55% operating margins
$861M
Storage Connectivity
Q4 FY2025, -25% YoY decline
+2%
Q4 FY2025 Growth
First positive quarter, cyclical trough may be behind

Cash cow with limited upside

Non-AI semiconductor contributes less than 6% of equity value and is not a material investment thesis driver. It generates stable cash flows at high margins on a fabless model. The primary risk is secular decline in storage and industrial end markets. The primary opportunity is DOCSIS 4.0 and Wi-Fi 7 upgrade cycles, which are covered in separate child nodes.

Open questions

?How large is the Wi-Fi 7 upgrade cycle addressable market?
?Is enterprise storage connectivity in secular decline or cyclical trough?