The Latent
EN▾
EnglishEspañol中文PortuguêsFrançaisالعربية日本語한국어
Sign Up
NEWSBENCHMARKSDATARESEARCHPODCASTNEWSLETTERPRESSPARTNER WITH US →
Data/Demand

Demand

Broadcom AI Semiconductor Revenue

Broadcom AI semiconductor revenue in USD billions: custom XPUs plus AI networking. A hyperscaler infrastructure-demand proxy that complements NVIDIA Data Center revenue. Quarters are Broadcom fiscal (Q1 is Nov–Jan), aligned by quarter-end with NVIDIA, not by matching Q1 to Q1.

Key takeaway

Broadcom AI revenue rose from an estimated $0.6B in Q1 FY23 to a reported $10.8B in Q2 FY26, tracking custom silicon and AI cluster networking rather than merchant GPUs alone.

Broadcom AI Semiconductor Revenue

  • Broadcom
Broadcom AI Semiconductor RevenueAs of 2026-01-01, Broadcom AI semiconductor revenue was $10.8 in the latest reported quarter.$15B$10B$5B$0BBroadcom, 2025-01-01T00:00:00.000Z: $4.4BBroadcom, 2025-04-01T00:00:00.000Z: $5.2BBroadcom, 2025-07-01T00:00:00.000Z: $6.5BBroadcom, 2025-10-01T00:00:00.000Z: $8.4BBroadcom, 2026-01-01T00:00:00.000Z: $10.8BQ2 FY25Q3 FY25Q4 FY25Q1 FY26Q2 FY26Broadcom: $10.8B on Q2 FY26
SOURCE: The Latent
ZOOMALLYTD12M3M1M

API access · Pro coming soon

As of 2026-01-01, Broadcom AI semiconductor revenue was $10.8 in the latest reported quarter.

Pro API coming soon

Methodology

Values are Broadcom AI semiconductor revenue compiled from company disclosures and management commentary. Quality labels in the downloadable data mark reported figures, approximations, lower bounds, and a few FY23 estimates implied from later YoY growth comments.

Broadcom's fiscal year ends in late October. Fiscal Q1 is November–January, Q2 February–April, Q3 May–July, and Q4 August–October. Axis labels keep Broadcom's own fiscal quarters (Q1 FY23 through Q2 FY26).

Because NVIDIA's fiscal year ends in late January, Broadcom Q1 is not the same window as NVIDIA Q1. Points are placed by the calendar period they occupy so Broadcom Q1 FY23 lines up with NVIDIA Q4 FY23 (both end in January). Compare the two series by date, not by matching Q1 to Q1.

The series includes custom AI accelerators (XPUs) and AI networking—Ethernet switches, fabrics, NICs, optics, and interconnect. In Q2 FY26 networking was almost 40% of Broadcom AI revenue. It is a proxy for hyperscaler AI infrastructure deployment, not the size of the entire AI semiconductor market.

Frequently asked questions

How is this different from NVIDIA Data Center revenue?

NVIDIA Data Center revenue is the best public proxy for merchant accelerated computing (mostly GPUs plus NVIDIA networking). Broadcom AI semiconductor revenue captures custom XPUs and a large share of AI cluster networking, including networking sold into GPU as well as XPU deployments. Together they are closer to total hyperscaler AI infrastructure demand. If spending shifts from GPUs toward custom ASICs, NVIDIA can slow while Broadcom accelerates—or the reverse.

Why does Broadcom Q1 not line up with NVIDIA Q1?

The two companies use different fiscal calendars. NVIDIA's year ends in late January (Q1 is February–April). Broadcom's year ends in late October (Q1 is November–January). We keep each company's own labels, but plot both by the calendar period of the quarter, so Broadcom Q1 FY23 sits next to NVIDIA Q4 FY23. Align by quarter-end date, not by matching Q1 to Q1.

Is $10.8B the size of the AI semiconductor market?

No. It is Broadcom AI segment revenue: a concentrated hyperscaler-customer proxy that can move with product cycles, mix between XPUs and networking, pricing, and share. The trajectory from about $0.6B in Q1 FY23 to $10.8B in Q2 FY26 is the signal, not the absolute level as a market-size estimate.

Related charts

  • AI chat traffic by site
  • AI Chat Search Interest
  • NVIDIA Data Center Revenue

The Latent

AI industry news. A sister publication to The Block.

Editorial

  • Standards
  • Corrections
  • Commercial policy
  • Contact

Company

  • Data
  • Benchmarks
  • Score
  • Methodology
  • About
  • Team
  • Privacy Policy
  • Terms of Service
  • Security
  • The Block
  • Add The Latent as a preferred source